TRIUMF: Canada's national laboratory for particle and nuclear physics

KAON Factory study : accelerator design report TRIUMF-KAON Project 1990

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


51833-TRI_KAON_1990_02.pdf [ 159.7MB ]
JSON: 51833-1.0355485.json
JSON-LD: 51833-1.0355485-ld.json
RDF/XML (Pretty): 51833-1.0355485-rdf.xml
RDF/JSON: 51833-1.0355485-rdf.json
Turtle: 51833-1.0355485-turtle.txt
N-Triples: 51833-1.0355485-rdf-ntriples.txt
Original Record: 51833-1.0355485-source.json
Full Text

Full Text

KADN FACTORY STUDY ACCELERATOR DESIGN REPORT TRIUMF 4004 WESBROOK MALL VANCOUVER, B.C., CANADA V6T 2A3 KADN FACTDRY STUDY ACCELERATOR DESIGN REPORT TRIUMF 4004 WESBROOK MALL VANCOUVER, B.C., CANADA V6T 2A3 TABLE OF CONTENTS EXECUTIVE SUMMARY 1. INTRODUCTION 2. BEAM OPTICS 2.1 Accelerator and Storage Ring Optics 2.2 Beam Transfer 3. MAGNET AND KICKER SYSTEMS 3.1 Magnets 3.2 Magnet Power Supplies 3.3 Kicker Magnets 4. ACCELERATING SYSTEM 4.1 RF Voltage Program 4.2 Beam Loading 4.3 RF Systems 5. BEAM STABILITY AND INSTRUMENTATION 5.1 Collective Instabilities 5.2 Stabilizing Systems 5.3 Beam Instrumentation 6. BEAM PIPE AND VACUUM SYSTEM 7. ACTIVATION AND SHIELDING 8. H- EXTRACTION FROM THE CYCLOTRON 9. CONTROL SYSTEM 10. ALIGNMENT 11. PARAMETER TABLES 12. INDEX OF DESIGN NOTES EXECUTIVE SUMMARY To produce the intense beams of kaons and other exotic sub-atomic particles desired at the KAON Factory the atoms in a solid target must be bombarded with high intensity beams of high energy protons (nuclei of the lightest atom, hydrogen). Experience at Brookhaven National Laboratory in the USA and at CERN (the European Centre for Nuclear Research) in Switzerland has shown that the protons should have a kinetic energy of at least 30 GeV (giga- or billion electron-volts). At this energy the protons are travelling at 99.95% of the speed of light and are about 30 times more massive than at rest. Their energy is raised to this level step by step in a series of particle accelerators (commonly known as atom-smashers). To produce 100 times more kaons than have been available at Brookhaven and CERN the KAON Factory must be capable of accelerating beam intensities of 100 microamperes (6 X 1014 protons per second). The aim of this Study was to review the design of the accelerators, to build prototypes of critical components and to update the cost and manpower estimates. The Accelerator Design Report describes the results of the Study on the design of the overall complex and on the individual subsystems. Costs and manpower are dealt with in a separate volume. In this sUImnary we review first the overall design and then the progress on each subsystem. ACCELERATORS FOR THE KAON FACTORY Accelerators come in many varieties from the humble TV set to the free electron laser, from the X- and gamma-ray generators in hospitals to the synchrotron light sources used for processing modern micro-chips, from electron welding machines to cyclotrons for medical isotope production, from laboratory electron microscopes to the 53-mile-long Supercon-ducting Super Collider. In spite of their variety of shapes and sizes all particle accelerators have the same basic components: • an evacuated chamber to carry the beam of electrically charged particles, normally protons or electrons; • high-voltage generators, usually working at radio frequencies, to provide the electric fields to accelerate the particles; • magnets to steer the particles in a closed path for many passages through the same accelerating device, and to focus or contain them in a narrow-diameter beam. • electric power supplies to drive the various systems • cooling systems to limit electrical heating • diagnostic and control systems for both beam and hardware. Considering the wide variety of uses for accelerators in both basic and applied science, especially in high technology and medicine, Canada, like other advanced industrial coun-tries, has found it expedient to keep at the forefront of accelerator technology. Major centres have developed at AECL Chalk River, the Saskatchewan Accelerator Laboratory and TRIUMF. The first stage of acceleration for the KAON Factory will be provided by the existing TRIUMF 0.5 GeV cyclotron. This is one of only 4 accelerators in the world capable of providing proton beams of the required 100 microamperes intensity in this energy range. To raise such an intense beam to higher energies it is necessary to use synchrotron accel-erators in conjunction with storage rings. Because the protons become harder to bend at higher energy these machines have large circumferences, over 1 kilometre at 30 Ge V, and are therefore installed in tunnels underground. Each machine consists essentially of a string of electromagnets and radio-frequency accelerating cavities, through which is threaded a rectangular vacuum pipe for the beam, typically 15 cm x 10 cm in diameter. Each ac-celerator is followed by a storage ring which prepares the beam for the next stage. This separation of acceleration from storage allows more pulses of beam to be passed through the system per second, maximizing the intensity. Five new rings will be required: • the Accumulator ring and Booster synchrotron, located in the Booster tunnel, 216 me-tres in circumference, and cycling 50 times per second. • the Collector ring, the Driver synchrotron and the Extender ring, located in the 5-times-Ionger main tunnel, and cycling 10 times per second. Accumulator Ring. This ring gradually accumulates the regular stream of beam bunches produced by the cyclotron at intervals of 43 nanoseconds (billionths of a second) into 40 notional "buckets" around the circumference. The process is similar to filling a moving circular train of boxcars from regularly spaced piles on a conveyer belt. It continues over 1/50 second, time enough for about 20,000 orbits of the ring. Finally the 40 bunches are transferred from the Accumulator to the Booster, which has the same circumference. Booster Synchrotron. In the next 1/100 second this synchrotron accelerates the beam bunches from 74% to 97% the speed of light, increasing their energy six-fold to 3 GeV. The use of a booster stage reduces the cost of the whole system. Because the beam shrinks in diameter during acceleration it enables the magnets in the much longer 30 Ge V synchrotron to be significantly reduced in size and cost. It also reduces the complexity and cost of the radio-frequency accelerating system by restricting the problem of providing a large frequency rise, matching that in velocity, to the Booster, a small ring where only a low radio-frequency voltage is required (750 kilovolts). 11 MAIN TUNNEL BOOSTER C'\ ~EXPERIMENT AL HALL Figure 1: Layout of the site, with cross-sections through the tunnels showing the five rings: A - Accumulator, B - Booster, C - Collector, D - Driver, E - Extender Collector Ring. This fixed-energy ring, located in the main tunnel, is 5 times longer than the Booster and is used to collect five successive bunch trains from it end to end Qver a period of 1/10 second. Special radi<rfrequency cavities are used to lengthen the bunches and avoid the microwave instabilities which would otherwise be expected at full intensity. Driver Synchrotron. This is located immediately beneath the Collector and is used to accelerate the 200-bunch train assembled there from 3 Ge V to 30 Ge V. The speed of the protons increases from 97% to 99.95% of the speed of light and they become 8 times more massive. The frequency of the accelerating voltage (2550 kilovolts) must also increase by 3%. The bunch train extracted from the Driver is about 3 millionths of a second long and can be used directly for neutrino experiments. The neutrino is so weakly interacting that it is beneficial to use them in sharp pulses in order to better distinguish real signals from background. Extender Ring. For experiments with strongly interacting particles, like kaons, pions and antiprotons, a sharp pulse would produce too many simultaneous interactions, indis-tinguishable from one another. For these the beam is transferred to the Extender. This is a deliberately imperfect storage ring from which the protons are gradually extracted into the main experimental hall over the full 1/10 second cycle time. III PROGRESS DURING THE STUDY The following paragraphs list the highlights of progress on the various projects. Accelerator Design • To accommodate a more efficient slow extraction system, the shape of the Extender and other large rings has been changed from circular to racetrack. Computer tracking studies have shown that the use of an additional electrostatic deflector in the long straights can reduce the beam spill ten-fold to below 0.2%. • The Extender will be located 4 metres horizontally outside the Driver, rather than 1 metre below it. This is more convenient for installation and servicing and allows more space for local shielding in the extraction regions. • The main tunnel is to be located to the west of the present buildings, rather than around them. This avoids congestion and provides more space for the experimental halls. This relocation together with changes in the magnet lattices have required the complete redesign of all beam transfer lines. • Provision has been made for complete debunching of the beam in the Extender, as required for certain rare-decay experiments. • The Booster quadrupole focusing magnets have been reversed to make injection and extraction easier. • The injection section of the Accumulator has been redesigned and the scheme for painting the narrow beam from the cyclotron over the much larger aperture has been revised to take advantage of the development of stripping foils with two free sides. • Detailed schemes have been developed for accelerating spin-polarized beams. • To avoid beam spill due to coupling resonances the horizontal and vertical emittances are now taken to be equal. This increases the magnet apertures somewhat. • Beam dumps in the transfer lines are used in preference to fast abort systems. Magnets • Detailed magnetic field computations have been carried out to optimize the design of the Booster and Driver dipole (bending) and quadrupole (focusing) magnets. • A prototype Booster dipole magnet has been constructed. • A prototype Booster quadrupole magnet is under construction. lV • A magnet measurement system with several testing stations has been designed ca-pable of dealing with up to 25 magnets per week, the maximum production rate. Magnet Power Supplies • The test stand using old magnets from the NINA synchrotron operated successfully at dual frequency, with the field rise times three times longer than the fall. (This feature will allow the radio-frequency voltage required for acceleration in the Driver to be reduced by 1100 kilovolts.) • The frequency-changing capacitor-disconnect switch has been redesigned, eliminating a GTO thyristor which did not meet specifications. • A power supply suitable for testing the Booster prototype dipole magnet at full power has been designed and delivered. • The basic building block for regulated dc supplies was selected to be a 450 volt, 1000 ampere unit and one of these was constructed based on the chosen technique of high-frequency link power conversion. • Multi-cell resonant circuits have been designed capable of powering all the dipole magnets in each synchrotron. • The dc by-pass chokes have been designed in conjunction with Canadian industry. A combined choke for each cell is ruled out by size and weight, and so separate chokes will be used. Kickers These are pulsed ferrite magnets with rise times better than 100 billionths of a second, used to kick the beam into or out of a ring. • Two kicker magnets were obtained on loan from CERN and were carefully studied with regard to both their magnetic and mechanical design. • A pulser was also obtained from CERN and its cycling rate successfully increased from 1 to 50 cycles per second. Sufficiently flat 40 kilovolt pulses were obtained, 600 billionths of a second long, with better than 30 billionths of a second rise and fall times. • A kicker magnet suitable for operation with this pulser has been designed and built. v • A novel pulsed chopper has been designed and a prototype is being built. This device will be installed in the injection line from the cyclotron in order to deflect away 5 out of every 45 beam bunches. This will provide a gap in the beam 100 billionths of a second long in each ring for turning the injection and extraction kickers on and off. To do this a 30 kilovolt pulse of this length is sent to the strip-line deflector every millionth of a second. The power requirement is drastically reduced by energy storage in a low-loss transmission line. Radio-frequency Accelerating System • A prototype Booster accelerating cavity was completed, based on that in operation at the Fermilab booster synchrotron. It was successfully tested at signal level over the entire frequency range (46-61 million cycles per second) using air-cored tuners. • An alternative prototype, developed at Los Alamos and shown there to give improved performance at fixed frequency, has been obtained on loan. This cavity has now been completely reconstructed at TRIUMF to allow the frequency to be swept over the full range 50 times per second; initial tests look promising. • A structure to damp higher-order modes in the Los Alamos cavity has been success-fully tested at signal level. These modes could induce instabilities in both the beam and radio-frequency systems. • Modified forms of the HERA proton cavities developed at AECL, Chalk River, are proposed for the other rings. • A detailed theoretical study of beam loading has been carried out in order to gain a thorough understanding of the feedback loops needed to avoid instabilities induced in the radio-frequency system by the high beam power. • A 2.4 kilowatt solid-state amplifier with time delays less than 50 billionths of a second has been developed and tested. This meets the requirements for the fast-feedback control loops. • Collaboration has been continued on the design of radio-frequency cavities and am-plifiers with the Los Alamos and the Superconducting Super Collider groups. Beam Pipe and Vacuum • More stringent vacuum tolerances have been imposed to avoid beam instabilities. • Pumping systems have been specified in detail for each of the rings and transfer lines. VI • Ceramic vacuum chambers will be used within the fast-cycling magnets of the Booster and Driver in order to avoid eddy current heating and magnetic fields. These must, however, incorporate metallic "radio-frequency shields" to provide low-impedance paths for the image currents induced by the beam. Three contracts were placed for prototypes. The first was with the Rutherford Appleton Laboratory in the U.K. based on the design used successfully in their ISIS accelerator. This design, which uses an internal wire cage as shield, has been completed and delivered. The second design, developed by SAIC in San Diego, provides the shield in the form of metal strips laid in grooves on the internal surface of the ceramic; a partial prototype is under construction. The third contract with Omega Slate of Calgary involves construction of the ceramic pipe only. • A vacuum test stand has been constructed, enabling the outgassing rates of various materials to be evaluated. Extraction of the Cyclotron Beam The existing cyclotron accelerates H- (negative hydrogen) ions - a proton surrounded by two electrons. To extract the beam the circulating H- ions are intercepted by a thin carbon foil in which the two electrons are stripped off while the much heavier proton passes through. Having a positive electric charge, the proton is bent out of, rather than back into, the cyclotron magnetic field. This stripping process, however, is essential for efficient injection into the Accumulator over many thousands of turns. The H- ions must therefore be extracted whole from the cyclotron and transported to the Accumulator. An extraction system for H- ions has been under development at the cyclotron for some years. The first two elements, a radio-frequency deflector and electrostatic septum, had already been built and tested, showing that 90% of the H- beam could be cleanly separated from the circulating beam. • The layout and design of all the extraction elements has been finalized. • Extensive magnetic field computations have been carried out for the 5 new elements, all magnetic channels. • A prototype of the third channel, which uses some iron elements, has been built and measured to check the calculations. • To protect the electrostatic septum from irradiation and heating, a fine stripping foil can be placed in front of it, in principle at multiples of 1200 upstream of the septum. Experiments have now confirmed that the 1200 shadow is indeed stable enough to protect the septum. • A test of the electrostatic septum at full beam intensity which had been planned for fall 1989 was postponed because of the cancellation of the regular cyclotron shutdown. It will now take place in spring 1990. Vll Control System • The logical structure of the control system has been analysed in considerable detail using Structured Analysis Structured Design methods. • The hardware architecture is based on a local area network supporting control buses and data buses. • A thorough study of the hardware and interface requirements has been completed. • The synchronization of the rings and the timing of transfer between them have been studied in detail. Shielding • The transverse shielding requirement for the maximum allowed spill in the accelerator tunnels was determined to be 8 metres of earth backfill or an equivalent density of concrete. • The necessity of providing low-sodium concrete for tunnel construction and shielding was identified. • The use of beam collimator systems to localize beam losses in the accelerator rings to specially shielded regions was examined. Systems Integration This activity involved studies of how the various accelerator components would be installed, powered, serviced and aligned in the accelerator tunnels, and how this would impact the tunnel size and the location of equipment buildings. • A design for the magnet transporter was selected based on a towed fork-lift vehicle similar to that in use at Fermilab. • A cabling study identified the preferred conductor configurations for providing the high current to the magnets. • An in-tunnel network of surveying monuments tied to a surface geodetic network through a number of surface-to-tunnel penetrations was specified to provide align-ment of magnets to the necessary tenth of a millimeter tolerance. Vlll INTRODUCTION Chapter 1 1 INTRODUCTION 1-1 1.1 Main Accelerator and Injector . ..... 1-1 1.2 Booster Synchrotron and Storage Rings 1-3 1.3 Time Structure of the Beam ... 1-6 1.4 Control of High-Intensity Beams 1-7 1.5 Polarized Beam .. . ....... 1-8 1.6 Design Changes since the 1985 Proposal 1-9 1 1 INTRODUCTION To accelerate the 100 pA proton beam from the TRIUMF cyclotroh to 30 Ge V a chain of 2 fast-cycling synchrotrons and 3 dc storage rings is proposed. 452 Me V H- ions from the cyclotron are injected by stripping into the Accumulator ring, 216 m in circumference. A 50 Hz Booster synchrotron in the same tunnel then accelerates the proton pulse to 3 Ge V, where the rf frequency swing is almost complete (46 to 61 MHz). In the main tunnel, lO78 m in circumference, are the Collector ring, which collects 5 Booster pulse trains, the lO Hz Driver synchrotron, which accelerates them to 30 GeV, and the dc Extender ring, from which the beam is gradually dispensed by slow resonant extraction. The accelerator designs have various features, such as H- stripping injection, high transition energy, and bucket-to-bucket beam transfer which will avoid or reduce beam loss. Dual-frequency magnet power supplies provide a 3:1 rise:fall time ratio in the Driver, reducing the peak rf voltage requirement from 3700 kV to 2550 kV at 61- 63 MHz. 1.1 Main Accelerator and Injector The specifications for the KAON Factory call for the accelerator to provide 100 pA proton beams at 30 GeV. This choice of energy satisfies the desire for intense fluxes of high-energy kaons as well as stopping kaons, antiprotons and neutrinos. The 100 pA current (6 X 1014 protons/s) is chosen to provide a significant (80-fold) improvement over beams which have been available in this energy region and to make possible experiments which have hitherto been impractical. In light of these specifications the KAON Factory accelerator has been based on a rapid-cycling 10 Hz 30 GeV proton synchrotron. At lower energies other types of accelerator could be considered, but above about 15 Ge V a synchrotron is the only practical choice. The fast cycling rate keeps the charge per pulse down to N = 10 pC (6 X 1013 protons) and restricts the time available for instabilities to develop. The circulating current, a measure of the likelihood of beam instability, is 2.8 A, not quite double the 1.5 A at which the CERN PS operates and a little less than the 4 A so far achieved at the Rutherford ISIS synchrotron. Intensity-dependent effects, such as tune shift, instabilities and beam loading, therefore lie in a well-explored region. For reference, Table 1.1.1 lists the major beam properties of various medium-energy proton synchrotrons. The existing 30 Ge V synchrotrons are limited in beam intensity both by their low cycling rates « 1 Hz) and by their low injection energies (200 MeV into their first synchrotron stages). The injection energy is crucial because space charge reduces the transverse focusing strength by an amount strongly dependent on energy, and largest near injection; it also varies within every beam bunch. There results a spread in tune (number of betatron oscillations per orbit) which, at low energies, for a machine of fixed radius and magnet aperture, is given by D..v ex: -N/ (32,,/; here (3 and 'Yare the usual relativistic speed and 1-1 Table 1.1.1: Medium-Energy Proton Synchrotrons Average Rep. Protons/ Circulating Energy Current Rate Pulse N Current i (GeV) (pA) (Hz) (x 1013) (A) Fast Cycling Argonne IPNS 0.5 14 30 0.3 4.0 Rutherford ISIS 0.75 130(200) 50 1.6(2.5) 4.0(6.1) AGS Booster (1.5) (20-40) (7.5) (1.8-3.5) (4-8) Fermilab Booster 8 7 15 0.3 0.3 SSC LEB (11) (8) (10) (0.5) (0.5) Slow Cycling KEK PS 12 0.32 0.6 0.4 0.6 CERN PS 26 1.2 0.38 2 1.5 Brookha.ven AGS 28.5 0.9 0.38 1.6 0.9 - with Booster (4) (0.38) (6) (4) Kaon Factories TRIUMF KAON 30 100 10 6 2.8 KA 0 N Booster 3 100 50 1.2 2.7 Moscow KF 45 125 6.25 12.4 3.2 Moscow Booster 7.5 250 50 3.1 3.2 total energy parameters. Assuming an upper limit on /).11 (-0.2 if serious resonances are to be avoided) the accelerable charge per pulse N increases as j32'l, evaluated at injection. It was to take advantage of this effect that the injection energies for the Brookhaven AGS and CERN PS were raised from the original 50 MeV to 200 MeV and 800 MeV, respectively, providing improvement by a factor of 5 or more. To achieve the 50 to 100-fold higher intensity specified for the KAON Factory, without demanding a very high (and expensive) cycling rate, requires an injection energy of at least 400 MeV (a factor of 2.6 better than 200 MeV). The TRIUMF cyclotron provides the basic performance required for the KAON Factory injector: H- beams at energies up to 520 MeV, with 150 J-lA currents delivered with good reliability (over 85% in recent years). Moreover, local low~cost electricity will make it possible to continue operating on the present 9 month/year schedule. The crucial question is how to match this cw machine, producing a continuous stream of beam bunches at 23 MHz, with a 10 Hz synchrotron accelerating 3-ps-Iong pulses every 100 ms. Fortunately, the key to an answer is already availa.ble in the use of H- ions in the TRIUMF cyclotron. If these are injected into the next stage by stripping them to protons as they pass through' a thin foil, Liouville's Theorem on phase space conservation may be circumvented and beam injected over the many thousands of turns required. At present, the stripping process is 1-2 used for extracting protons from the cyclotron; a new extraction system will be required to extract the H- ions intact. Extensive testing has shown that this will be straightforward, especially if beam is extracted at 452 MeV, taking advantage of radial precession induced by the Vr = 3/2 resonance nearby. Extracting below 500 MeV will also avoid most of the beam spill caused by Lorentz stripping of the H- ions in the cyclotron's magnetic field; this amounts to 8% between 452 and 500 MeV but is only 1% below 452 MeV. The /)2,3 factor is 3 times larger at 452 MeV than at 200 MeV. The choice of a high cycling rate for the synchrotron implies a need for high energy gain per turn and therefore high aggregate rf voltage (about 4 MV). To ease this requirement, an asymmetric magnet cycle will be used in the main synchrotron with a rise time 3 times longer than the fall; this reduces the voltage required by one third, and the number of cavities in proportion. This not only simplifies the rf system but, equally importantly, reduces the cavity gap impedance capable of driving coupled-bunch instabilities. 1.2 Booster Synchrotron and Storage Rings The rf system presents two additional challenges. One is associated with the high beam current, implying an rf power capability well above the 3 MW in the beam itself. The other is the frequency swing, which amounts to a factor 1.37 as the energy rises from 450 MeV to 30 GeV. In order to separate the problems of providing high rf voltage and power from that of providing frequency swing, an intermediate Booster synchrotron is proposed to accelerate the proton beam from 450 MeV to 3 GeV. This covers almost the entire frequency range (a factor 1.33), but involves only 300 kW beam power and requires only 750 k V rf voltage; (an asymmetric magnet cycle is not economically worthwhile in this case, since half the voltage is needed to form the rf bucket and the potential savings amount to only 120 kV). The Booster is one-fifth the circumference of the main "Driver" synchrotron (215.66 m rather than 1078.30 m), so that 5 pulses from it completely fill the Driver circumference, and it cycles 5 times faster at 50 Hz. The Driver itself is left with a 2550 kV voltage requirement but a frequency swing of only 3%. The chief parameters of the two synchrotrons are listed in Table 1.2.1. The use of a Booster also allows the aperture and cost of the main ring magnets to be significantly reduced - the beam width being reduced a factor 2 by adiabatic damping during passage through the Booster. The lower magnetic fields used, together with the large number of rf cavities required, combine to make the circumferences of these two synchrotrons somewhat larger than those of slow-cycling (low-intensity) machines of the same energy. 1-3 Table 1.2.1: Synchrotron Design Parameters BOOSTER DRIVER Energy 3 GeV 30 GeV Circumference 4.5 x2rr RJ = 215.66 m 22.5 X 2rr RJ = 1078.30 m Current 100 J.LA = 6 x 1014/S 100J.LA = 6x1014/s Repetition Rate 50 Hz 10 Hz Charge/Pulse 2 J.LC = 1.2 X 1013 ppp 10 J.LC = 6 X 1013 ppp Long Straights 2 X 156 m No. Arc Superperiods 6 12 Lattice }{ Focusing FODO FODO Structure Bending OBOBBOBO BBBB No. Focusing Cells 24 68 Maximum f3x X f3y 17.5 m X 15.4 m 38.0 m x 31.8 m Dispersion 7]max 7.2 m 7.4 m Transition It 15 30 i Tunes Vx X Vy 5.23 X 7.22 13.23 X 14.18 Space Charge ~Vy -0.17 -0.10 Emittances }{ f nx X f ny 60rr X 60rr (J.Lm)2 100rr X 100rr (J.Lm? at Injection fl ong 0.048 eV-s 0.192 eV-s Harmonic 45 225 Radiofrequency 46.1-60.8 MHz 60.8-62.5 MHz Energy gain/turn 316 keY 2040 keY Max RF Voltage 750 kV 2550 kV RF cavities 12 X 75 kV 18 X 150 kV To allow time for injection or for slow beam spill for counter experiments, it is conventional to "flat-bottom" or "flat-top" the magnet cycle of a synchrotron. In the present case, however, starting with 100 J.LA beams from the TRIUMF cyclotron, such a procedure would result in average beam currents at 30 GeV of only 50 J.LA for neutrino production (fast extraction) or 33 J.LA for counter experiments (slow extraction). Instead, it is proposed to follow each of the three accelerators by a relatively inexpensive dc storage ring, so that the TRIUMF cyclotron would be followed by a chain of 5 rings, as follows: • A Accumulator: accumulates cw 452 Me V beam from the cyclotron over 20 ms periods • B Booster: 50 Hz synchrotron; accelerates beam to 3 Ge V • C Collector: collects 5 Booster pulses and expands longitudinal emittance • D Driver: main 10 Hz synchrotron; accelerates beam to 30 GeV • E Extender: 30 Ge V storage ring for slow extraction. As can be seen from the energy-time plot (Fig. 1.2.1) this arrangement allows the cy-1-4 clot ron output to be accepted without a break, and the B and D rings to run continuous acceleration cycles; as a result the full 100 p,A from the cyclotron can be accelerated to 30 GeV for either fast or slow extraction. The practicality of multi-ring designs has been thoroughly demonstrated at the high-energy accelerator laboratories, and new projects such as HERA, LEP and SSC also use a large number of stages. } 30 GeV 10 Hz ~3 GeV 50 Hz } 452 MeV 23 MHz rl "",,-0, T,,,-oT"I,,,,"-r~~~, ~I~,-r~~~>_ Time (ms) o 100 200 Figure 1.2.1: Energy-time plot showing the progress of the beam through the five rings The Accumulator is mounted directly above the Booster in the small tunnel, and the Col-lector above the Driver on the inner side of the main tunnel (Fig. 1.2.2). The Extender is installed on the outer side of this tunnel, separated 4 m horizontally from the Driver, allow-ing it and the C ring to be shielded from the beam spill associated with slow extraction. A major change from the 1985 proposal is the switch from circular- to racetrack-shaped lat-tices for the large rings. The primary motivation was to provide a longer straight section for slow extraction from the Extender to accommodate collimators and an electrostatic pre-septum; simulations have shown that this latter device could reduce the beam spill by almost an order of magnitude, to less than 0.2%. Dispersionless long straights are also advantageous for the C and D rings, simplifying the beam transfer systems, permitting special insertions (such as for a Siberian snake) and allowing power supplies to be grouped more economically. On the other hand, the racetrack shape makes the main tunnel fit very tightly around the buildings on the existing site; it is therefore proposed to locate it to the west, an arrangement which also provides more space for the experimental areas. 1-5 MAIN TUNNEL BOOSTER EXPERIMENTAL HALL Figure 1.2.2: Proposed layout of the accelerators and cross-sections through the tunnels Figure 1.2.3 shows schematically the magnet lattices, the beam transfer lines and the location of the rf stations. Similar lattices and tunes are used for the rings in each tunnel. This is a natural choice providing structural simplicity, similar magnet apertures and straightforward matching for beam transfer. The need for the Accumulator ring would disappear if, instead of the TRIUMF cyclotron, a high-intensity pulsed H- linac were used as injector. The cost of such a machine, ri-valling LAMPF in performance, would, however, be formidable, about $100 million even for 452 MeV, based on recent estimates for other projects. By comparison, the cost of the Accumulator is estimated to be about $10 million. The Collector ring, besides marshalling the 5 bunch trains from the Booster, is used to dilate the longitudinal emittance of each bunch to reduce the danger of microwave insta-bility in the Driver. It might also be dispensed with, as in the LAMPF AHF proposal, although this option is not tied to the choice of a linac as injector. Whatever the injector, the lack of a C ring necessitates flatbottoming the main synchrotron (D) magnet cycle for collection of the Booster pulses. Maintaining the same final average current (100 /-LA) then requires increasing either the repetition rate or the number of protons/pulse (and hence the magnet apertures) for both Band D rings. The costs involved in such changes would considerably exceed any savings achieved by eliminating the C-ring, the cost of which is $35 million. 1-6 C Beam Stop -c::r Snake/Spin Rotator -c::r Dipole Magnet -{}- Quadrupole Magnet • RF Station I Figure 1.2.3: Arrangement of magnets, rf cavities and beam transfer lines (schematic) 1-7 1.3 Time Structure of the Beam To minimize beam loss, bucket-to-bucket transfer is planned between the rings. To achieve this, the rf frequency at injection into the Accumulator ring must be some simple multiple of that of the TRIUMF cyclotron, 23.05 MHz. Double this frequency, 46.1 MHz, has been chosen, making the frequency at top energy 62.9 MHz. For this frequency range the cavities are of reasonable size and tubes and power supplies are readily available. Moreover, operational experience is available on the fast cycling 30-53 MHz system at the Fermilab Booster and on the 52 MHz proton system at HERA. In order to populate all the rf buckets, which are spaced only half as far apart as those at the extraction radius RJ in the injector cyclotron, the radius of the A and B rings is chosen to be an odd half-integer multiple of the cyclotron's, 4.5 RJ = 34.4 m. The circumference (215.8 m) will then contain 45 rf buckets compared to the 5 bunches circulating in the cyclotron at double the spacing. 4.5 turns from the cyclotron will encircle the A ring once, populating every other bucket; the next 4.5 turns will populate the remainder, the bunches being automatically interleaved (Fig. 1.3.1). To provide a 110 ns gap for kicker rise or fall, ••••••• • • • • • • • • • • • • • • • • • 0 • • • <> • • 0 • • • • <> • • • 0 0 • • 0 • • 0 • • • • • •• • • • •••••• • e. • • • t t • • • • • • • • <> <> • Figure 1.3.1: Populating the 45 Accumulator buckets over two turns of injection. Five buckets are kept empty by kicking out selected bunches using a 1 MHz chopper in the injection line. 1-8 5 adjacent buckets are kept empty by removing the appropriate bunches en route from the cyclotron, using a stripline chopper in the injection line, pulsing at the orbit frequency 1.024 MHz. This allows the 40 full buckets to be extracted cleanly from the Accumulator and injected cleanly into the Booster. This 40 full + 5 empty bucket pattern is retained through the B, C, and D rings, serving the same purpose for injection and extraction kickers, although the gap shrinks to 80 ns at 30 Ge V. The C and D rings, being 5 times larger, will contain 225 rf buckets. Because of the higher frequency, the time separation of beam bunches will be 15.9 ns at 30 GeV, compared to 21.7 ns at 450 MeV. If larger separations should be required, they can be achieved, with some loss of intensity, by pulse programming the ion source. For instance, the 1:5 pulser already in operation would increase the bunch separation to 80 ns. Each bunch in the fast extracted beam from the Driver is 2.6 ns long; each train of 40 bunches is 0.62 J-lS long, while the five trains extend over 3.48 J-lS. In the Extender the extra circumference provides 230 buckets, and one of the five gaps is lengthened from 5 to 10 buckets. To some extent bunch length can be adjusted to users' requirements by changing the transition energy, the rf voltage or the slow extraction mode. A completely debunched beam can also be achieved within rv 1 ms by allowing the beam to coast with rf off; bunch rotation by a quarter turn prior to this will reduce the momentum spread by a factor 6 to 0.05% (FWHM). 1.4 Control of High-Intensity Beams Successful operation of a high-intensity accelerator depends crucially on minimizing beam losses and the activity they produce. Sources of loss must be identified and either con-trolled or eliminated, the beam and any spill must be carefully monitored, losses must be localized through the use of collimators and beam dumps, suitable materials must be used for absorption and shielding of spilled beam and, where activation cannot be avoided, equipment must be capable of being handled remotely. Careful attention to these features in the case of the TRIUMF cyclotron has enabled twice the originally specified beam cur-rent to be accelerated with reduced exposure to personnel. The same principles would be followed for the five rings of the KAON Factory accelerator. Several processes which give rise to losses in many existing machines have been avoided entirely in this design. The use of H- ions for injection will almost entirely eliminate injection spill in the Accumulator ring. The use of bucket-to-bucket transfer between the rings will avoid the losses inherent in recapturing coasting beams. The buckets will not be filled to more than 80% of their height; this should avoid beam losses while providing a high bunching factor to minimize the space-charge tune spread at low energies, and sufficient spread in synchrotron tune to give effective Landau damping. The rf voltage is programmed 1-9 to maintain a constant bucket area over the early part of the cycle. The magnet lattices are designed to place transition, where the phase focusing passes through zero, well above top energy in all the rings, thus avoiding the instabilities and losses associated with that passage. Moreover, with the beam always below transition, it is no longer advantageous to correct the natural chromaticity, so that sextupole magnets are needed only for correcting field imperfections. Beam instabilities will be suppressed or carefully controlled. Although all 5 rings have large circulating currents, the rapid cycling times give the instabilities little time to grow to dangerous levels. Longitudinal and transverse coupled-bunch modes, driven by parasitic resonances in the rf cavities and by the resistive wall effect, will be damped using active damping by electronic feedback. The longitudinal microwave instability is a separate case, which because of its rapid growth will be avoided by making the longitudinal emittance suf-ficiently large at every point of the cycle and by minimizing the high frequency impedance of the vacuum chamber as seen by the beam. At this stage of the design it is not possible to make accurate estimates of beam blow-up due to instabilities or non-linear resonances, but to be safe, the magnet apertures have been designed to accommodate 20-30% growth in emittance in each ring, and the horizontal and vertical betatron acceptances are taken equal to twice the nominal 20' injection, emittances. In the event of a malfunction in any ring the ion source will be shut off and any surviving beam will be directed to a dump located off the subsequent transfer line. A fast abort system is not deemed necessary, the charge per pulse being not much larger than in existing medium-energy machines. 1.5 Polarized Beam The new optically-pumped source on the TRIUMF H- cyclotron has already provided 5 pA proton beams with 50% polarization, and further improvements are expected as commissioning proceeds. To accelerate polarized protons to higher energies is difficult, because of the number of depolarizing resonances to be crossed, but not impossible, as Brookhaven, Saclay and KEK have shown. Imperfection resonances occur every 0.524 GeV but are relatively weak; intrinsic resonances occur less frequently, depending on the lattice and the tune, but are stronger. The KAON Factory Booster can be tuned to eliminate all but one of the latter, and this should be crossable successfully using the conventional pulsed-quadrupole tune-jumping technique. This would be too complicated for the Driver; instead a Siberian snake is being considered for one long straight to keep the spin tune away from resonances. A snake will also be required in the Extender to provide all directions of polarization in the slow extracted beam. A beam with 60% polarization should be attainable at 30 GeV. 1-10 1.6 Design Changes since the 1985 Proposal For convenience, major design changes since the 1985 Proposal are listed here, starting with some general considerations and then working alphabetically through the rings: • Horizontal and vertical beam emittances are now assumed equal to avoid beam growth due to coupling resonances. • Detailed schemes are presented for accelerating polarized beams. • More stringent vacuum tolerances have been imposed to avoid beam instabilities. • Transfer line dumps are used in preference to fast abort systems. • The gap in the beam for kicker rise or fall will be created by a 1 MHz chopper in the injection line. • The H- injection insertion has been redesigned, the Accumulator superperiodicity changed to 3, and the painting scheme revised to take advantage of recently-developed stripping foils with two unsupported edges. • The Booster quadrupoles have been reversed to make extraction easier and now agree in polarity with the Accumulator. • The Booster magnets will be powered from a single-frequency 50 Hz circuit. • The shape of the C, D and E rings has been changed from circular to racetrack. • The main tunnel is located to the west of the present buildings, rather than around them. • The lattice changes and ring relocations have required the complete redesign of all beam transfer lines. • The E ring is separated 4 m horizontally from the C and D rings, rather than 1 m vertically. • A pre-septum in the slow-extraction system should reduce beam spill below 0.2%. • Provision is made for complete debunching of the beam in the Extender. 1-11 BEAM OPTICS Chapter 2 2 BEAM OPTICS 2-1 2.1 Accelerator and Storage Ring Optics 2-1 2.1.1 Lattice Designs . . . .... 2-1 2.1.2 Beam Emittance and Beam Pipe Acceptance 2-9 2.1.3 Single-Particle Instabilities 2-11 2.1.4 Tracking Studies 2-14 2.1.5 Polarized Beam . 2-19 2.2 Beam Transfer ... 2-24 2.2.1 H- Injection 2-24 2.2.2 Transfer Line Design . 2-37 2.2.3 Slow Extraction 2-48 2 BEAM OPTICS 2.1 Accelerator and Storage Ring Optics 2.1.1 Lattice Designs The magnet lattices of the two synchrotrons and three storage rings are all of the separated function type with FODO focusing structure. As mentioned in the introduction, very similar lattices are used for the A and B rings in the Booster tunnel and the C,D, and E rings in the Driver tunnel. The similar structures, with dipoles and quadrupoles mounted above or beside each other, provide engineering convenience and permit easy beam transfer between the rings with straightforward matching of beta functions and dispersion. All five lattices are designed to have a high transition energy, far above the operating ranges of the machines. This allows instabilities and beam loss due to space charge in the zero phase-focusing regime at transition to be completely avoided. Several authors[l] have suggested ways of obtaining this feature: A superperiodicity is introduced in the lattices of the arcs and the relevant horizontal tune set just below the number of superperiods. For the Booster a regular quadrupole structure has been used, with some full and some empty cells. This has been chosen for its good optical properties and its excellent stability under large space-charge detuning. The superperiodicity is created by arranging the dipoles in clusters. This has the advantage that the peak beta function is not increased, and the missing magnet cells automatically provide space (and superperiodicity) for the rf cavities and for the injection and extraction systems. The Accumulator is similar but has three of the long straight sections modified as insertions, with one accommodating the H- injection system. For the large rings a racetrack lattice has been developed, primarily to provide straight sections longer than 100 m as required for slow extraction with low losses. In the Driver missing-magnet cells would take too much space in the arcs. Therefore the superperiodicity is created by modulating the quadrupole strength and thus, the beta functions. As each arc has a phase advance of 5 x 271", a superperiodicity per arc of 6 is chosen. In the dispersion-free straight sections dR / dp = a and therefore only a small adjustment to It in the arcs is needed to compensate for the increased length. The alternative approach of eschewing superperiodicity and using a regular lattice, where It ~ vx, with a high tune is impractical. To produce tunes Vx > 2,max with phase advances per cell in the favourable region around 75° would require more than 10,max cells. This would require at least 40 cells in the Booster and 300 in the Driver, with impractically 2-1 small quadrupole spacings of only 2.7 m and 1.8 m, respectively. One consequence of the relatively high rf voltages associated with rapid cycling is relatively high synchrotron tunes. Suppression of synchrobetatron resonances driven by the rf-cavity fields is therefore of concern. In the large rings this is achieved straightforwardly by placing the cavities in the dispersion-free long straight sections. In the small rings, where dispersion-free straight sections do not exist, the suppression is achieved by arranging the rf cavities with a certain superperiodicity around the ring. The five synchrotron lattices are described in more detail below. Table 2.1.1 summarizes the lattice parameters for all the rings. Compared to slow-cycling machines of the same energy our rapid-cycling synchrotrons have relatively large circumferences. The Booster circumference is 215.66 m, compared to 105.6 m for Saturne II, and the Driver circumference 1078.3 m, compared to 807 m for the Brookhaven AGS. The extra size arises from the relatively low maximum magnetic fields which are economic to achieve. For the Booster dipole magnets cycling at 50 Hz a maximum field of 1.12 T has been assumed; for the Driver dipoles, at 10 Hz, 1.38 T. The maximum pole-tip fields in the quadrupole magnets were chosen so that their saturation curves matched those of the dipoles, 0.72 T in the Booster and 0.98 T in the Driver. For the lattice studies extensive use has been made of the second order lattice code DIMAD from R.V. Servranckx.[2] Plots of the lattice and envelope functions and of the machine layout are made using data from the DIMAD output files by the graphics postprocessor DIPLOT developed at TRIUMF.[3] Small Rings • Booster The Booster synchrotron accelerates the proton beam injected from the Accu-mulator at 450 MeV to 3 Ge V, cycling at 50 Hz. The Booster lattice consists of 24 half-full regular FODO focusing cells, each about 9 m long, with about 4 m space between the quadrupoles for the dipoles, rf cavities, and injection and extraction systems. Figure 2.1.1,a shows the lattice functions for one superperiod and the arrange-ment of quadrupoles and dipoles. There are 4 dipole magnets in each super-period occupying 4 half-cells and leaving the other 4 empty. The 6-fold su-perperiodicity is obtained by reversing a regular BOBO arrangement in every other pair of cells. This brings two dipoles together to make a full cell and two straight sections together to make an empty cell. Choosing a horizontal tune somewhat smaller than the dipole periodicity of six, the dispersion function is modulated from a high of 7.2 m to a low of -3.1 m and the average dispersion 2-2 Table 2.1.1: Lattice Parameters for the Five Rings Accumulator Booster Collector Driver Extender Energy (Ge V) 0.45 0.45-3 3 3-30 30 Circumference (m) 215.66 215.66 1078.30 1078.30 1102.26 Repeti tion rate (Hz) de 50 de 10 de Intensity (pC/pulse) 0-2 2.0 0-10 10 10-0 N urn ber of cells/arcs 24 24 48 48 48 Number of cells/straight 10 10 8 Number of superperiods/arcs 6 6 12 12 12 Horizontal tune 3.73 5.24 13.23 13.23 13.23 Vertical tune 5.71 7.22 14.19 14.19 14.19 Max. hor. beta/arcs (m) 17.8 17.5 36.5 38.0 38.0 Max. vert . beta/arcs (m) 17.8 15.5 31.4 31.8 32.9 Max. hor. beta/str. (m) 36.5 38.0 100.8 Max. vert. beta/str. (m) 25.4 28.3 40.0 Max. dispersion (m) 5.2 7.2 5.2 7.3 6.9 Transition Energy 'Yt 4.0 15.0 00 36i 11-30i Dipoles Number 24 24 96 96 96 Length (m) 1.01 2.99 1.0 4.89 3.90 Min. field (T) 0.88 0.30 0.83 0.17 1.75 Max. field (T) 0.88 1.12 0.83 1.38 1.75 Quadrupoles (in arcs) Number 51 48 94 94 94 Length (m) 0.3 0.36/0.46 0.2/0.4 0.94-1.69 0.82-1.8 Strength dB/dx(T/m) 2.8-4.2 12.5/8.5 7.3-9.8 14.7 8.76-19.46 Quadrupoles (in straights) Number 42 42 34 Length (m) 0.2 0.75- 1.55 0.70- 1.60 Strength dB/dx(T /m) 7.5-10.4 14.20-15.10 11.42-24.06 Long straight sections Number 6 6 2 2 2 Length (m) 12.5 8.8 154 154 154 Correction elements: Sextupoles Number 24 24 48 48 48 Length 0.2 0.2 0.2 0.2 0.2 Horiz. orbit correctors 68 68 64 length 0.1 0.2 0.2 Vert . orbit correctors 24 24 68 68 64 length 0.1 0.1 0.1 0.1 0.1 2-3 is reduced. In this way, It is raised to about 15 in the Booster, well beyond the top value of I of 4.2, although the tune is only 5.2. The vertical tune, Vy = 7.2, has been chosen two units higher than the horizontal tune in order to avoid the zero-harmonic coupling resonance Vc = Vy and the structural integer resonance Vy = 6. Alternatively, the vertical tune can be set to Vy = 4.2. This second working point has advantages for polarized beams as outlined in Sec. 2.1.5, although for high intensity Vy = 7.2 is preferred because it gives a higher space-charge limit and a larger acceptance. The focusing action of the parallel entrance and exit faces of the dipoles introduces only negligible modulation in the vertical beta function. The short straight sections in the half-empty cells have relatively low dispersion. They are used for rf cavities - six pairs of which occupy six straight sections -and for transverse collimators and various diagnostic elements. Two of the six long straight sections are occupied by the injection and the extraction system. As injection and extraction occur vertically, the whole system can be located within one straight section. Another such section contains the longitudinal loss collimator, taking advantage of the high momentum resolution at this location. Small correction dipoles are placed near the defocusing quadrupoles for correc-tion of the vertical closed orbit distortion. The horizontal closed orbit distortion will be minimized with backleg windings on the dipoles. The beam envelopes at injection, for a momentum spread D.p/Po = 3.42 x 10-3 and an rms emittance of € = 6071" mm-mrad are shown in Fig. 2.1.1,b. At extraction the beam sizes have shrunk by roughly 50%. This lattice differs very little from that proposed in 1985. The quadrupoles have been reversed in order to ease the injection and extraction requirements and to avoid modulating the quadrupole strengths. There are now just two quadrupole families, F and D, promising better tracking with the dipoles. A number of racetrack designs were also investigated for the Booster and Accu-mulator, with a particular view to the advantages of dispersion-free straight sec-tions for placement of rf cavities and injection and extraction systems.[4] With the small circumference, however, suitable lattices proved difficult to design. Moreover, having a periodicity of 2 they tended to be unfriendly to polarized beams, while the use of Siberian snakes in the Booster is prohibited by the low energy. • Accumulator The Accumulator ring accumulates the 450 MeV H- beam from the TRIUMF cyclotron during the 20 ms cycle time of the Booster. Injection into this machine is via charge exchange employing a stripper foil. The magnet arrangement and lattice functions are illustrated in Figure 2.1.2,a. The machine lies 2 m above the Booster and has the same circumference and the same lattice structure as the Booster except for the injection section and two symmetrically located sections, giving it a superperiodicity of 3. The injection 2-4 (a) (b) 15 8' ~ 0+-------------------------------+ .<l (m) O+-~--------~--T---------~ ~ -5 ~ .., - 60,00" mm-mr «I 1l DISTANCE (m) DISTANCE (m) Figure 2.1.1: Lattice functions (a) and beam envelopes (b) for the Booster. 20 15 {3x (3y 10 'r/X (m) 5 0 0 10 20 30 40 50 60 8' ~10 (a) :5 ." '~ S " 5 " ~ M N ~ 0 8' ~ .c ~ -5 ~ S " " ~>--10 70 80 N ~ 0 £z - .45.00 1f mm-mr 6p/ p - 0,00350 ty = "5.00 7r mm-mr DISTANCE (m) I DISTANCE (m) Figure 2.1.2: Lattice functions (a) and beam envelopes (b) for the Accumulator. section was modified to create a double waist 'at the stripper foil and also to provide a 7 m long drift unobstructed by quadrupoles in order to accommodate the injection bump magnets. The tunes of the Accumulator, Vx = 3.73 and Vy = 5.71, are lower than for the Booster in order to obtain a momentum resolution 1J/R ~ 1.2 m1/ 2, as required for the phase-space painting process (see Sec. 2.2). Because of the lower horizontal tune the transition energy is It = 4, substantially lower than in the Booster. The quadrupoles and dipoles are much shorter than in the Booster, because of the lower energy. Dipoles with a magnetic induction of 0.9 T and a length of 1 m are placed in the middle of drifts, between quadrupoles. Here the beta functions are substantially lower than in the quadrupoles and hence the apertures are smaller. The dipoles, with parallel entrance and exit faces, perturb the regular vertical focusing structure of the quadrupoles causing some modulations in the 2-5 (b) betafunctions. The maximum dispersion is 5 m. Fig. 2.1.2,b shows the beam envelope at the end of the accumulation cycle. Large Rings The lattices for the large C, D, and E rings have been completely redesigned since the 1985 proposal. The primary motivation was to facilitate the design of an ultra-low-loss slow extraction system including collimators in order to allow hands-on mainte-nance of most parts of the E ring. This requires the availability of very long straight sections and also moderately high beta functions at the extraction septa. To facili-tate slow extraction of both bunched and debunched beams it is desirable to be able to set Ft in the Extender to either a relatively low value, about 10, or to an imaginary value as in the Driver synchrotron. In addition, some control of the dispersion in the straight sections is necessary to be able to use both an achromatic extraction scheme with zero dispersion in the extraction straight and a chromatic extraction scheme with finite dispersion at the septa. These requirements led to the design of a separated-function racetrack lattice with FODO focusing structure. [5] As the use of special dispersion-suppressor cells was not compatible with the requirement of high Ft, the arc cells are completely filled with bending magnets, maximizing the length of the straight sections (154 m). Dispersion in the straights is suppressed by choosing the tune of each arc to be an integer value, IIx = lIy = 5, making each 24-cell arc a second-order pseudo-achromat. If necessary the dispersion in the straights can readily be adjusted by changing the horizontal tune in the arcs. The average phase advance per cell in the arcs is 75°, very close to the optimum value for the lowest beta functions. The overall tune of the machine will be set using the straight sections. For all rings the tunes are chosen to be IIx = 13.2 and lIy = 14.2, again avoiding the zero-harmonic difference resonance IIx = lIy • Since the bending structure in the arcs is completely regular, transition has to be pushed up by modulating the beta function rather than the bending radius. To this end, a sixfold periodicity is introduced in the arcs by modulating the focusing-quadrupole strengths in three families, FDFIDFDF2D. Again the dispersion function 'fJ develops oscillations, bringing its average value below zero and giving imaginary Ft. In fact, Ft can be tuned over a wide range by adjusting the quadrupole modulation, providing the flexibility necessary in the Extender ring . • Driver The Driver accelerates the beam from 3 GeV to 30 GeV, cycling at 10 Hz. The overall layout is shown in Fig. 1.2.3. In this machine high values of the beta function in the straight sections are not necessary, in fact not desirable. Consequently the focusing structure of the straight sections has been chosen to be as similar as possible to that of the arcs. In order to be able to control the working point, each half-straight is made up of a two-cell transformer followed by a regular three-cell FODO section. The lattice functions in the regular section are chosen to be about the same as in the arcs. In order to be able to vary 2-6 the tunes and the lattice functions in the regular FODO section independently, seven quadrupole families are needed. The lattice functions of the machine are shown in Fig. 2.1.3,a, for a value of It of 36i. The modulation of the lattice functions is clearly visible, but the peak value of the beta functions does not exceed 38 m horizontally and 31 m vertically. Dispersion reaches a maximum of 7.3 m in the arcs. For a maximum field of 1.38 T the dipole magnets in each half-cell must be 4.88 m long. The size of a half-cell, 8 m, is long enough to allow space for placing correcting elements for closed orbit distortions, sextupoles, pumps, bel-lows and beam position monitors between quadrupoles and dipoles. Although the saturation curves of quadrupoles and dipoles have been matched to min-imize tracking errors to bet ter than 10-3 , trim windings on the quadrupoles are provided to allow correction of any remaining tracking errors in the arcs, if necessary, in order to avoid spurious dispersion in the straight sections. The injection and extraction systems and the rf cavities are placed in the straight sections, taking advantage of the space available and the absence of dispersion. The length of a half-cell varies from 7.3 m to 8 m. One straight section will also accommodate the Siberian snake needed for acceleration of polarized beams (see Sec. 2.1.5). The quadrupoles in the straight sections will be programmable in families, so that no extra trim quadrupoles will be needed there. The beam envelopes for an rms emittance of € = 247r mm-mrad are shown in Fig. 2.1.3,b. At extraction, the beam size has shrunk by almost 70%. 30 I I I I \ ~ ,I 1\ /I ~ 'I I ~ ,I II /I II 'I II \ I \ R 20 I ,I II I 'I' I ! II I • II I (a) N "-(b) I"x I I I , I I 1(' I Y ,I, '1,1 II, '\ ,II " \1 II \ {3 I~ I~ lfil' I I II I I \ ' I - 0+------------------------------+ 7Jx 10 I \ I \I \ I \I I' \ I \' I \,' 1\, \ V { I \ \I~ , V '\/ ~ J , I VI , \' (m) VI • v, ,,- \ , V \ ~ . \ / " 1'-· o " I ".J -~~~~~~~~~~~~~~~~ o 50 100 150 200 250 300 DISTANCE (m) S ~ ~ -5 .~ e CI ., &l ",-~ a N "-I l:y = 24 .00 7f mm-mr 50 100 150 200 DISTANCE (m) Figure 2.1.3: Lattice functions (a) and beam envelopes (b) for the Driver . • Collector 250 The Collector operating at 3 Ge V collects 5 pulses from the Booster cycling at 50 Hz before injection into the Driver cycling at 10 Hz. Also, the longitudinal emittance of the beam will be blown up by a factor of 4 in the Collector in order to prevent microwave instability in the Driver. 2-7 300 ]: 10 ""'""",'""""""""""""",'",',,,',',' , ' , ' , ' , (a) t. - 20 .00" mm-mr .c: <1p/p = 0.00370 (b) ~ 30 ~, ,\ A; ~ ,! , 1\ 1\ " /I I L I I 1\ I I I I I I , I \ 1\ 1\ A A {3x 20 I I 1\ I , \ I \ I .~ ~ 5 Z .... \ i \I \ I \I (3y 1\ I I, I' I 111~I/\I\l1 "> 0-+----------------+ TJx 10 I \1 II I I \1 I, \ I I, I' \1 \I \I \I I \ \ / \1 I VV \ \1 \ i\l / ~ Y V V V I (m) VI\~ \ .\~ I V, o / \. I ' - , / , -, '---10-r,...,..........,.,..,........,........"..-r-r"TT-r-r-r-r...,-r-r-r,..,,.....,..,.....,..+ o :10 100 1:10 200 2:10 DISTANCE (m) e $ tT = 20.00" mm-mr 50 100 150 200 DISTANCE (m) Figure 2.1.4: Lattice functions (a) and beam envelopes (b) for the Collector. 250 The lattice functions of the Collector are shown in Fig. 2.1.4,a. Except for the much shorter magnets - the dipoles are 1 m long at 0.83 T - the lattice is identical to the Driver lattice. Transition I has been set to 00 as in the Driver in order to maintain stability against microwave instability before the longitudinal emittance blowup. This ring will be located 1.6 m above the Driver synchrotron, allowing straightforward vertical beam transfer between the identical straight sections of the two rings. In Fig. 2.1.4,b the beam envelope in the Collector is shown for an rms emittance of E = 2071" mm-mrad . • Extender The 30 Ge V Extender accepts one pulse from the Driver and is emptied over the time period of 100 ms, while the Driver is ramping down and accelerating another pulse. The lattice is similar to that of the Driver with the exception of a slightly longer circumference, 1102.26 m rather than 1078.3 m, and a modified straight section with more widely spaced quadrupoles (12.7 m) and a maximum beta function of 100 m for the slow-extraction system. Fig. 2.1.5,a shows the lattice functions of the Extender, for a value of It of 00. The ring will be mounted side-by-side with the Driver in the same tunnel, with 4.1 m separation of the extraction straight section from the C and D rings to facilitate shielding of the Driver straight section from the potentially much more radioactive slow-extraction system.[6] The circumference of the Extender has been chosen to give a harmonic number of 230, rather than 225 as in the C and D rings. The beam envelope is shown in Fig. 2.1.5,b for a horizontal emittance of Ex = 2571" mm-mrad and a vertical emittance of Ey = 371" mm-mrad. The large value for the horizontal emittance reflects the increase in beam size due to phase-space distortion during slow extraction. This has to be taken into account in determining the apertures of the magnets. The bending magnets have a length of 3.6 m at 1.75 T field. 2-8 300 a 120 ~ 10 .<: .. 100 .., i s .. 5 80 .. J> {3x H 60 N {3y ~ 0 T/x 40 a ~ (m) :9 -5 .., 0 i s t. = 3.00 1r mm- mr .. .. -20 J> -10 0 50 100 150 200 250 300 ~ 0 50 100 150 200 250 N DISTANCE (m) '-"I DISTANCE (m) Figure 2.1.5: Lattice functions (a) and beam envelopes (b) for the Extender. 2.1.2 Beam Emittance and Beam Pipe Acceptance The transverse emittances were made equal to decrease susceptibility to coupling reso-nances. The initial transverse emittance was derived by imposing an upper limit of 0.18 on the magnitude of the space-charge tune shift, !:111. The following relation, along with slight corrections due to the effect of dispersion on the beam size and due to the effect of image charges at 450 MeV, was used to determine the minimum normalized emittance Here Np is the number of protons in the ring, Be is the ratio of average current to peak current, and rp is the classical proton radius. For our parameters, we find in rings A and B, En,min =30 7rmm-mrad. This emittance is 'minimum' because it corresponds to a transverse distribution (the Kapchinsky-Vladimirsky distribution) in which the space-charge force is linear, so that all particles at a given longitudinal position have the same tune shift . Another property of this distribution is that for any of the four 2-D projections, all of the beam is contained inside a uniformly populated ellipse whose area is 4 times the rms emittance Eq. Realistically, we expect the distribution to fall smoothly away from the origin so that the 4Eq emittance will be somewhat larger. Based upon painting studies (see Section 2.2.1), a reasonable value for the unnormalized 4Eq in the Accumulator is 45 7rmm-mrad. The beam emittance and momentum spread used to evaluate the beam size are given in Table 2.1.2. A transverse emittance blowup of 30% has been assumed in the Booster and 20% in each of the large rings, based on our tracking studies. The choice of longitudinal emittance and the reason for the factor of 4 blow-up in the Collector are explained in Section 4.1. 2-9 300 Table 2.1.2: Emittances and Momentum Spread A B C D E 4€q,x( rrmm-mrad) 45 60 20 24 25 4€q,y( rrmm-mrad) 45 60 20 24 3 D.p/p x 103 3.48 3.48 3.73 3.85 1.64 Bucket heightxloJ 4.66 4.68 4.88 4.90 6.14 The horizontal and vertical beam-stay-clear regions within the quadrupoles and dipoles are summarized in Table 2.1.3. They are calculated using the following formula: where Ah is the half width of the beam-stay-clear region, €i the 4€q beam emittance at injection energy, (6.p/p)mAX is the maximum bucket height and c.o.d. stands for closed-orbit distortions that are scaled with ViJ. The factor 2 was applied to the emittance as a safety factor in order to ensure that no more than 1 % of the beam is lost. By using the bucket height rather than the momentum spread of the beam in the dispersive term it is assured that all particles that can be accelerated are contained within the acceptance. Only in the Extender is the base momentum width of the beam used, since the bunching factor is very small (0.1) and no further acceleration takes place. Table 2.1.3: Beam Stay-Clear Apertures (half-widths in cm) A B C D E H V H V H V H V H V Arc quadrupoles .j2iJi 4.0 4.0 4.6 4.3 3.8 3.5 4.2 3.5 4.4 1.4* 7]( D.p/p ) max 2.4 3.4 3.0 2.5 1.1 c.o.d. 0.4 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 Total 6.9 4.5 8.5 4.8 7.3 4.0 7.2 4.3 6.0 1.9 Dipoles .j2iJi 3.2 3.6 4.0 3.8 3.1 2.7 3.9 3.6 3.9 1.2 7](6.p/p)mAX 1.4 1.3 1.7 2.2 0.9 c.o.d. 0.4 0.5 0.5 0.5 0.4 0.3 0.5 0.4 0.4 0.4 Total 5.0 4.1 5.8 4.3 5.2 3.0 6.6 4.0 5.2 1.6 *) based on slow extraction dynamics 2-10 2.1.3 Single-Particle Instabilities Instabilities can be classified either as 'single-particle' or 'collective' depending on whether they affect the motion of individual particles or of the whole beam. Single-particle in-stabilities arise through resonances between the natural oscillations of the particles and components - both intrinsic and accidental - of the magnetic guiding and rf accelerating fields. They are controlled through the rf and magnet lattice designs and the setting of appropriate tolerances on magnetic and rf field errors. Betatron instabilities may arise when there is resonance between the transverse oscillations of particles about the equilibrium orbit and harmonic components of the magnetic guiding field; i.e. when kllx ± Illy = n where lIx and lIy are the horizontal and vertical betatron tunes (numbers of oscillations per turn) and k, 1 and n are integers. The integer n refers to the driving harmonic of the magnetic field. Imperfections in the magnetic field will be present for all values of n and will drive relatively weak 'imperfection resonances'. If, however, n is an integer multiple of some structural feature of the lattice, such as the total number of cells N or the superperiodicity S (n = pN, n = qS), then the magnetic field component will be relatively strong and this 'systematic resonance' potentially more serious. The importance of a resonance also depends on its 'order' (k + 1), since the excitation term depends on the (k + 1- l)th derivative of the magnetic field. (Thus the third order resonance 311y = 16 is driven by skew sextupole components of the 16th harmonic magnet field component). The sign in the equation above distinguishes 'sum' and 'difference' resonances. Sum resonances are characterized by a stopband of width e around the resonance line, within which the motion becomes unstable. For difference resonances there is no unstable region, but a width e can still be calculated to provide a measure of the amplitude growth. In addition, 'synchro-betatron' resonances may occur involving the longitudinal (syn-chrotron) oscillations, where kllx ± Illy ± mlls = n. Here m is another integer and lis is the synchrotron tune. In proton machines lis is much smaller than lIx or lIy (at least two orders of magnitude for the KAON Factory) so that synchro-betatron resonances (m 2:: 1) are generally treated as satellites of the m = 0 betatron resonance. Also, only satellites of the integer resonances lIx = n, lIy = n' are sufficiently strong to be important. They are driven by harmonic components of the energy gains due to the accelerating fields. There are several causes for the resonance. The primary cause is dispersion at rf cavity locations [7,8,9]. Synchro-Betatron Resonance and RF Cavity Distribution Synchro-betatron resonance driven by dispersion in rf cavities is expected to play an im-portant role in the Booster and the Accumulator because the dispersion in the straight 2-11 sections is large and the synchrotron tunes are also rather large. In the larger rings, this resonance is not important because the straights are designed to be dispersionless. More-over, the synchrotron tune and the space charge tune shifts are smaller in C, D and Ethan in A and B. The resonance is caused by the sudden changes in energy at rf cavities. These do not change the horizontal coordinates of the particle, but do change its equilibrium orbit, with respect to which betatron motion is measured. Thus betatron motion is excited. In addition, the betatron motion leads to a change in path length. The synchrotron phase at an rf cavity depends on the time taken to travel between cavities, which in turn depends on the length of the orbit. Thus synchrotron motion is also excited by the betatron motion. Synchro-betatron resonances driven by dispersion in rf cavities can be circumvented by placing the cavities symmetrically around the ring in a proper arrangement [10]. This is the method used in SPEAR [9]. Specifically, the rf cavity superperiodicity must not be equal to the integer nearest the horizontal tune or to any submultiple thereof. In the Booster, the integer nearest liz is odd (5) so rf cavities have been placed in diametrically opposite pairs. In the Accumulator, the integer nearest liz is 4 so the three cavities have been placed with 3-fold symmetry. While this strategy cancels the effects of the horizontal synchro-betatron satellites of the integer nearest the tune, those of the next nearest integer are not cancelled. However, they are of such high order (m ~ 17 in both A and B) that their effects are negligible. The afore-mentioned strategy of rf cavity placement makes the lowest order horizontal synchro-betatron resonances impotent provided that the cavities have no amplitude or phase errors with respect to each other. Also, vertical synchro-betatron resonances are not driven provided there is no vertical dispersion. Realistically, however, there will be amplitude and phase errors and there will be a small amount of vertical dispersion due to quadrupole misalignments, etc. Calculations have been made based on an rms voltage error of 10% and a vertical dispersion of 0.1 m. (In fact lower voltage errors should be attainable.) Besides effects due to non-linear synchrotron motion, the calculations also include both the 2118 modulation of the transverse tune due to space charge and the Ills modulation due to uncorrected chromaticity. Emittance growths, shown in Table 2.1.4, arise because a particle on a sideband will gain betatron amplitude and so, because of nonlinear transverse space-charge forces, be driven off resonance. The growth applies to the beam as a whole because the tune shift and the synchrotron tune decrease as the beam gains energy and so the sideband can sweep through the whole beam. For m > 5, growth is negligible because it occurs too slowly compared with the rate at which the sideband sweeps through the beam. 2-12 Table 2.1.4: Emittance Growth Due to Synchro-Betatron Resonances in the Booster m Horizontal Vertical 1 13% 7.5% 2 4.5% 2.5% 3 12% 6.7% 4 5.0% 2.8% 5 7.0% 4.0% Choice of Working Point The betatron tunes V& and Vy chosen must avoid serious resonances. In the case of the Booster we require Vx ~ 5 to drive the transition energy to higher values, and we choose Vy ~ 7 because this leaves us in the region of (vx, v y ) space which is fairly clear of low-order intrinsic resonances (indicated by heavy lines). The only resonances of this nature are those radiating from (vx , vy) = (6,6). The problem is to accommodate not just the 'working point' (vx , vy) of a lone particle, but the spread in tunes caused by different space charge defocusing at different points in the beam. This produces a necktie- shaped area whose length is determined by the maximum shifts (6.vx = -0.13,6.vy = -0.17) which occur soon after injection. This is the moment of tightest fit, because the synchrotron tune also has its maximum value then (vs = 0.046) pushing out the synchro-betatron satellites to their furthest extent from the parent resonances Vx = 5 and Vy = 7. Keeping Vx and Vy just below the 1/4-integer resonances, and below the diagonal, the horizontal m = 1,2 sidebands and the vertical m = 1 sideband can be avoided. However, most of the beam must sweep through the horizontal m = 3 sideband. An overall emittance growth of about 10% is expected but this will not lead to beam loss because the growth is confined to the core of the beam and not the halo. The chosen working point (5.24, 7.22) avoids crossing the lowest order sidebands. 'Fast' single particle instabilities can be defined as those whose growth is fast compared with the synchrotron motion. For fast instabilities, the instantaneous transverse tune, which in-cludes contributions from both chromaticity and space charge, becomes relevant. Tracking studies (Section 2.1.4) confirm that the uncorrected chromatic tune shift (±0.025, verti-cally) is too large to be accommodated along with the space charge tune shift in the region bounded by the diagonal, the integer, and the horizontal 1/4-integer. Vertical chromatic-ity must be corrected from its natural value of -1.4 (= ({w/v)/(fJp/p))to be no larger than about -0.7. The strength of the diagonal Vx - Vy = -2 resonance is determined in part by the solenoidal field from the tuners of the rf cavities. To minimize this effect, the cavities will be placed in back-to-back pairs. An analytic study[ll] was undertaken to determine whether the 2-13 effect is nevertheless large enough to require the cavities to be redesigned. Each tuner gives an integrated field of 350 G-m on the beam axis at top energy. (This is the relevant case because the tuner field rises more quickly than the beam's momentum.) We find a stopband width of 0.0012 for a single tuner. By comparison, the width due to the 1 mrad rms quadrupole tilt assumed in the tracking studies is 0.0070. Therefore, the effect of one uncompensated tuner is negligible. Effects due to solenoid aberrations have also been estimated and found to be negligible[ll]. In the case of the Driver, the working point (13.23, 14.19) is chosen in a similar region of tune space. Although the space-charge tune shift is only -0.10, the natural chromatic tune shift is large (±0.045). It turns out to be difficult to avoid the integer, the 1/4-integer, and the diagonal 'fast' resonances (each of which has considerable width), because the chromatic shift actually causes a positive instantaneous tune shift for particles with the largest synchrotron amplitude. As in the Booster, it is therefore desirable to reduce the natural chromaticity by a factor of 2. In the Driver, all the rf cavities are placed in dispersionless straights. Synchro-betatron resonances due to accidental dispersion are less important than in the Booster because, owing to a smaller synchrotron tune and smaller tune shifts, resonances up to m = 3 vertically and up to m = 5 horizontally can be avoided. 2.1.4 Tracking Studies In order to ensure beam stability in the rings, extensive tracking studies have been per-formed using the computer code DIMAD, which was modified to include simulations of synchrobetatron and space-charge effects. These studies, backed up by analytical work, allowed us to determine the uniformity necessary in the magnetic fields of the lattice elements, the orbit excursions arising from limited alignment precision, the effects of sex-tupoles, and the effect of synchrobetatron coupling. In the following sections we will outline the different effects studied. Closed Orbit Distortions Misalignments and magnet errors cause distortions of the closed orbit that were esti-mated by tracking. The machine elements were misaligned randomly with a Gaussian distribution truncated at twice the rms value. Test particles were tracked trough this model and the closed orbit was determined after setting the orbit correctors. Five runs were performed with different starting values for the random generator and the results analyzed statistically to find the maximum orbit excursion to be expected after correction and the maximum strength required for the orbit correctors. This procedure was carried out for each lattice. Misalignments and the resulting maximum orbit excursions and corrector strengths are summarized in Table 2.1.5. 2-14 Table 2.1.5: Misalignment, Orbit Excursions, and Corrector Strength Small rings Large rings units Displacement of elements rms 250 250 J.lm Roll error of elements rms 1 1 mr 8B/Bo rms 0.1 0.1 % Noise of monitors rms 250 250 J.lm max uncorrected orbit, hor. 50%c.1. 8.8 24.1 mm max uncorrected orbit, vert. 50%c.1. 9.3 14.4 mm max uncorrected orbit, hor. 95%c.1. 11.2 51.2 mm max uncorrected orbit, vert. 95%c.1. 14.7 24.5 mm max corrected orbit, hor. 50%c.1. 2.3 0.9 mm max corrected orbit, vert. 50%c.1. 1.0 1.4 mm max corrected orbit, hor. 95%c.1. 3.5 1.5 mm max corrected orbit, vert. 95%c.1. 1.2 2.2 mm max corrector strength, hor. 50%c.1. 4.4 0.2 mr max corrector strength, vert. 50%c.l. 0.3 0.2 mr max corrector strength, hor. 95%c.l. 5.7 0.2 mr max corrector strength, vert. 95%c.1. 0.4 0.2 mr In the small rings, the horizontal orbit excursions will be corrected by backleg wind-ings on the dipoles, while small orbit-correcting dipoles are provided at each defocus-ing quadrupole for correcting the vertical orbit excursions. For accelerating polarized beams the vertical orbit correctors will also be used to correct the imperfection res-onances (see Sec. 2.1.5). In the large rings, both the horizontal and the vertical orbit will be corrected by extra orbit-correcting dipoles. One horizontal orbit corrector will be provided at each focusing quadrupole and one at each defocusing quadrupole. Chromaticity Correction In order to minimize the tune spread, especially at transition, it has been usual to cancel the natural chromaticity e = p( dv / dp) of synchrotrons with sextupole correction magnets. However, these also cause aberrations limiting the acceptance of the machine. In our lattice designs, where transition is never crossed, it is no longer essential to make the chromaticity zero. Indeed, for controlling transverse coupled-bunch instabilities, the natural negative chromaticity is advantageous. Of disadvantage is the increase in tune spread, which is significant in our rings. With these considerations in mind, we explored the dynamic aperture for the machines with and without chromaticity correction. The results of these studies as outlined 2-15 in the following section indicate that the large tune shift indeed tends to reduce the acceptance in the machines at natural chromaticity, and partial correction of the chromaticity is needed to accept the full beam-stay-clear region. In the Booster the situation is somewhat more difficult than in the Driver as the space-charge tune shift is larger, and full chromaticity correction may be required. Fortunately the natural negative chromaticity is not required in this ring for beam stability. As these studies were done without correction of any nearby betatron resonances they do not prove the necessity of chromaticity correction, but space for sextupoles has been left in the magnet lattices and for each lattice a correction scheme without excessive geometric aberrations exists. In the small rings, chromaticity correction can be achieved with one family of six sextupoles in each plane. In the large rings, two such families per plane would be necessary to achieve full chromaticity correction while keeping the geometric aberrations small. Betatron Resonances In order to facilitate a cost-efficient design of the magnets for the five rings, it is important to have, amongst other things, a good estimate of the field uniformity required for each ring. While there are analytic approaches to this problem, usually based on a resonance model, the approximations in these formulae, especially the neglect of higher-order effects, makes it necessary to include relatively large safety factors in the design of the magnets. In addition, effects from synchrotron oscillations and space charge are usually not taken into account in a coherent way in the analytical treatment but rather dealt with separately. As the full problem is far too complicated for analytical treatment, particle tracking is the only practical approach to finding the limits of beam stability for a given set of magnet imperfections. We have therefore determined the stability limit of each lattice, its so-called dynamic aperture, by particle tracking. The field harmonics of the dipoles and quadrupoles as obtained from the magnet designs outlined in Sec. 3.1 of this report were simulated as multiple thin multipole elements associated with the respective magnets. In this way we created a rather complete model of the machine that includes misalignments, field uniformity errors, chromaticity corrections where desired, and synchrotron os-cillations. In the case of the Booster, the solenoidal stray field of the ferrite biasing system was also included (see Sec. 4.3). In order to simulate the effects of space charge two modifications were made to the tracking code. Simulation of the linear space-charge tune shift was included by mod-ulating the strength of the focusing elements according to the longitudinal position of the particle in the bunch. In this way, the modulation of the tune with a fre-quency equal to twice the synchrotron frequency and the corresponding variation of the working point is taken into account. This approach is expected to be somewhat pessimistic as the full space-charge tune shift is assumed for each particle indepen-dent of its transverse position. To estimate the effects of nonlinear space charge forces, a more sophisticated simulation deriving the space-charge forces for a beam of Gaussian shape is being implemented into DIMAD. This latter approach is still 2-16 under development.[12] • Small Rings The acceptance of the Booster lattice versus momentum spread of the beam is shown in Fig. 2.1.6 for the chromaticity corrected machine. The acceptance was determined by tracking particles at different emittances over 1000 turns and recording the highest particle still on a stable orbit for a given value of the momentum spread. As is evident from the figure the whole beam-stay-clear area lies within the acceptance of the ring, with the exception of a small area at a momentum spread of about 0.25%. The acceptance for the machine at half its natural chromaticity is shown with a dashed line; we attribute the dips in acceptance to crossing ofthe 211x -1I1i = 9 and the 311x +21111 = 30 resonances. For the reasons given above we expect the acceptance shown here to be a pessimistic estimate of the nonlinear dynamic aperture. 300 --~s.y=(O ,O) 240 - - ~SJ=( -3.6 , -5.1) ~ CIS M 180 S I S S 120 -.!:. ., .. w 60 stay-clear 0 .000 .001 .002 .003 .004 .005 6p/p Figure 2.1.6: Dynamic aperture of the Booster lattice. In order to assess the importance of synchrobetatron coupling in the cavities, simulations were carried out setting one or two of the cavities to zero voltage. No problems were revealed by these runs. Analytic calculations (Subsection 2.1.3) give an emittance growth of around 10%, but this figure is only relevant for the core of the beam and does not impact the stay-clear emittance.[13] The resonance spectrum of the lattice was determined by Fourier analysis of the tracking data. In Fig. 2.1.7 the resonance spectrum of the Booster lattice is shown, for an emittance of 12071" mm-mrad and dp/Po = o. Resonances up to 5th order can be identified. The lIx - 1111 difference resonance is prominent, indicating that the machine acceptance can be improved by skew quadrupoles if necessary. In the same way, the 3rd order resonances can be compensated by sextupoles. • Large Rings For the Driver lattice the dynamic aperture is given in Fig. 2.1.8 for natural 2-17 ~ ;:. -- Horizontal Invariant I r- .. - - Vertical Invariant -;:. H ~ ~ ;:. ;:. ;:. N r- N + -I .. ~ ~ H H ;:. ~ ;:. ;:. ;:. ;:. I ;:. r-N N I ,.,., "'It-+ I I H H I fl .. ;:. ;:. ;:. r-"'" N -.. ;:. r-I ,1 AAJ -I~~~ .L 1 1, .J .A . .tJ .0 .1 .2 .3 .4 .5 v Figure 2.1.7: Betatron resonance spectrum of the Booster lattice. 10 -- ~X.,=( -10.0,-9 .0) - -~x.,=(-21.8,-18 .3) Beam ....... .......... . -stay-clear o ~----.----.----,.----.----~ .000 .001 .002 .003 .004 .005 o pip Figure 2.1.8: Dynamic aperture of the Driver lattice. chromaticity (solid line) and half its natural chromaticity (dashed line) and the spectrum, in Fig. 2.1.9 for an emittance of 487r mm-mrad. Besides the Vx - Vy difference resonance a peak at 4vx also appears with noticeable strength. This resonance could be corrected by octapoles if necessary. Again, skew quadrupoles will improve the acceptance of the machine if this is found to be important. 2-18 ... ::. Horizontal Invariant I .. Vertical Invariant _ c- ::. Q) '1j ~ - --+oJ ... ..... .. ::. ...... p. ::. ... + S - ... -.:t ::. .. -::. N ... ::. ~ + ... ~ ::. ::. + .. N .. .. ::. I -.:t ::. -::. ~ I N -.. .. N ::. 7"1 ::. ... A N ~ ::. N )) .1 A A j 1 .0 .1 .2 .3 .4 .5 v Figure 2.1.9: Betatron resonance spectrum of the Driver lattice. 2.1.5 Polarized Beam The new optically pumped polarized ion source being commissioned at TRIUMF[14] has recently produced more than 20 /-LA of H- beam and allowed more than 5 /-LA protons with 50% polarization to be extracted from the TRIUMF cyclotron. As commissioning continues, the intensity and degree of polarization are expected to increase substantially. Essentially all of this beam could be accelerated to high energy. A polarized proton beam in a synchrotron or storage ring circulates with its polarization aligned vertically, along the direction of the guiding magnetic field. The polarization vector precesses in this field with a "spin tune" ,G where G = 1. 7928 is the anomalous magnetic moment and, the relativistic energy factor. Depolarization can occur if the protons experience horizontal field components, leading to precession of the spins away from the vertical axis, with a spatial frequency component the same as the spin tune. The Fourier spectrum of the horizontal fields always contains all multiples of the machine periodicity P, but due to the vertical betatron oscillations particles also experience horizontal fields oscillating at beat frequencies of the vertical betatron tune with the superperiodicity. The latter gives rise to the so-called intrinsic resonances that occur when the spin tune equals any of these beat frequencies. To first order ,G = kP ± Vy where k is any integer. These occur even in a perfect machine. The imperfection resonances occur when the spin tune equals any frequency component of the horizontal field components, ,G = k to first order, 2-19 but are only excited by vertical orbit excursions. Higher-order resonances do exist but are too weak to be dangerous for fast cycling machines. Each resonance is characterized by a width f, which can be evaluated by integrating the precession away from the vertical axis (depending on tune, magnetic field and betatron amplitude) around the closed orbit. For asymptotic passage through a resonance Froissart and Stora obtained an expression for the ratio of polarizations Pi (before) and P j (after): Pj I Pi = 2exp( _7r2 f2 12a) - 1. Here a = d, I dO is the speed of crossing. There are two extreme situations. In the first a weak resonance is crossed quickly so that f/a ~ 1 and Pj = Pi (no depolarization). In the second a strong resonance is crossed slowly so that f/a ~ 1 and Pj = -Pi (complete spin flip). Either extreme is acceptable in principle, since it leaves the beam fully polari~ed in the vertical direction. On the other hand resonances where f ~ a will produce IPj I < Pi and must be corrected by shifting them to one extreme or the other. For imperfection res-onances f must be reduced (or increased) by adjusting the orbit during resonance crossing to affect the specific harmonic of the field. For intrinsic resonances, the crossing speed can be affected by pulsing the betatron tune. The first-order resonance strength has been calculated for the Booster lattice using the program DEPOL.[15] By lowering the vertical tune to 4.2 it was found that only one intrinsic resonance of moderate strength, ,G = 0 + VII' lies within the acceleration range (Fig. 2.1.10,a). This resonance can be crossed with a tune jump as done routinely at the Brookhaven AGS, using four pulsed quadrupoles. Five imperfection resonances occurring at 0.523 GeV intervals will be corrected using the available orbit correctors,which will have sufficient bandwidth. For a well tuned system depolarization in the Booster is expected to be less than 10%. The situation in the Driver is much more difficult. Although in principle the number of intrinsic resonances can be reduced to only three, at ,G = 48k ± (VII - 4), there are 52 imperfection resonances to correct during acceleration, which represent a challenge just by their sheer number. The resonance spectrum is shown in Fig. 2.1.10,b. The number of intrinsic resonances has been kept small by retuning the vertical plane in the straight sections to an integer number of betatron oscillations, making the straight sections transparent in this plane. In this way the machine has the apparent symmetry of the arcs, P =12. The symmetry was increased further to P = 48 by removing the modulation of the beta functions in the arcs using the 4 independent power supplies. In this way the number of intrinsic resonances is reduced to 3 strong ones as given by the above formula, plus 4 weak resonances, which arise from a residual harmonic in the arcs due to the different length of the quadrupoles. These resonances can be further reduced in strength by careful tuning of the quadrupole families. Although ,t is lowered to 11, transition crossing should be 2-20 E* = 107T mm -mrad <y> = 0.38 mm vy = 4.18 99% Spin flip ----/ "-I \ 2 3 1 4 5 52-v 44+lIy 99% Spin ClifY 4-11 ----Y 1% Depolarization E* = 107T mmmrad <y> = 0 .55 mm 10 20 1 30 40 Figure 2.1.10: Spectrum of depolarizing resonances in the Booster (a) and Driver (b). The intrinsic resonances (!) are labelled according to their harmonics; the arrows point to the strength at 17(' mm-mrad emittance. The bars on the imperfection resonances indicate the 10' uncertainty of their strength. feasible with a polarized beam current on the order of 10 J.lA. Due to the fact that all intrinsic-resonance strength is concentrated in so few resonances these are quite strong, well beyond 99% spin flip. As resonance jumping would be im-possible, adiabatic spin flip would be used to cross these resonances. In order to increase the spin-flip efficiency at small betatron amplitudes the betatron tune can be pulsed so as to reduce the crossing speed, leading to about 95% spin flip integrated over the beam emittance. The resonance at ,G = 4 - Vy may also be crossed fast. In order to cross all 52 imperfection resonances, more orbit correctors may have to be installed. Depolarization in the Driver may be as low as 20%, resulting in an extracted polarization of about 50%, for 80% polarization injected. There exists, however, a much more elegant way of avoiding resonance crossing altogether by introducing a 1800 spin rotation about a horizontal axis in the machine, fixing the spin tune to 1/2 independent of energy. In this way deflections of the spin vector away from the precession axis are almost cancelled over two turns in the machine and depolarizing resonances are never crossed. Such a device, commonly referred to as "Siberian Snake" due to its invention in Novosibirsk, has been tested recently with success.[16] 2-21 For energies higher than a few Ge V, transverse fields have to be used for the spin rota-tors since the spin precession angle about these is proportional to ,G, larger by a factor ,G/(1 + G) than about the longitudinal field. This leads to the field strength varying only with the velocity f3 and the orbit through the rotator changing with energy, requiring rather large magnet apertures. It has been shown that a multi-twist helical magnetic field allows the design of snakes with relatively small orbit excursions, which can be optimized by choosing the number of twists in the helical field. [17] In order to keep the magnet apertures reasonable, a Siberian snake has been designed based on a helical field with three twists, but with 4 discrete dipole magnets per twist tilted by 450 about the beam axis.[18] This 14-magnet spin rotator is about 12 m long using superconducting 3 T magnets. Orbit excursions reach 8 cm at injection, requiring a magnet aperture of about 15 cm. A diagram of the snake is given in Fig. 2.1.11. As the velocity f3 of the beam varies by only 3% the magnets can be maintained ---""""-- - - - ...... _--"- - - -.::....-~~ Side view o view Figure 2.1.11: Diagram of the 3-twist helical Siberian snake. at a constant excitation independent of the energy. This creates only a negligible variation in the spin tune. With a snake the Driver will exhibit only a little depolarization (arising mainly from the cavities in the snake straight); the extracted polarization can be as high as 70%, for a beam of 80% injected into the Accumulator. Because of the focusing action of the many strong dipoles the snake must be carefully matched into the straight section of the Driver. This has been successfully achieved by a two-cell transformer on either side of the snake, maintaining the tune of the straight section (Fig. 2.1.12). Due to the snake being a fixed-field device the matching conditions vary throughout the acceleration cycle and require programming of the straight-section quadrupoles. Besides fixing the spin tune of the machine to 1/2, the snake also moves the stable spin direction in the ring into the horizontal plane. To match the vertically polarized beam from the Collector into the Driver, a 900 spin-rotating solenoid will be provided in the transfer line between the C and D rings. In the Extender a snake will be provided at the same azimuth as in the Driver in order to match the stable spin directions in both rings. In this way a spin rotator in the transfer line can be avoided. The change in the direction of polarization produced by the snake in the Extender can be used advantageously to 2-22 30 {3x 20 {3y TJx 10 " \ (m) 0 o 20 40 60 80 100 120 140 160 DISTANCE (m) Figure 2.1.12: Lattice functions and magnet arrangement for the straight section including the snake. provide longitudinally polarized beam in the extraction straight section. By adiabatically turning off the snake the polarization can be varied from longitudinal to vertical, providing both directions in the extraction straight section. This is important for experiments with internal targets in the Extender. The stable spin direction in the accelerators could be maintained vertical if two snakes with orthogonal precession axes were used, avoiding the need for the spin rotator in the C-D transfer line and also for the snake in the Extender. Unfortunately all snakes with a precession axis substantially different from the longitudinal axis appear to have very large orbit excursions and are impractical for injection energies below about 10 GeV. An alternative currently under investigation is the ramped partial Siberian snake proposed by Roser.[19] With a spin rotation of less than 1800 the orbit excursions at injection are significantly reduced, as is the field integral necessary. By ramping the snake with energy, optical matching of the snake would be much easier and, due to the rather low resonance strengths at low energies, the spin rotator in the C-D transfer line may be avoided. 2-23 2.2 Beam Transfer 2.2.1 H- Injection Introduction The purpose of the injection system in the A ring is to strip the 452 MeV cw beam of H- ions from the TRIUMF cyclotron and accumulate the resulting 451.5 MeV protons over 20 ms periods, painting them over a larger phase space volume in order to reduce the space charge tune shifts and the number of traversals through the stripping foil. Beam loss in the A ring is minimized if injection takes place into stable rf buckets rather than by allowing the beam to debunch and then be rebunched. An rf system is therefore provided at 46.1 MHz, the second harmonic of the cyclotron rf frequency. The ring circumference (215.66 m) is 4.5 times the 450 MeV cyclotron orbit length and the orbit time 1 J.LS. The half integer allows each bucket to receive charge every second turn. At 100 J.LA c.w. each Accumulator bucket will contain 3 x 1011 protons. For accumulation to be complete in 5x104 turns, the cyclotron must deliver at least 3x107 H- ions per bunch. Section 2.1.3 shows that loss from instabilities and proximity to resonances require tune shifts D.VXtY :::; 0.2 and D.v./v. :::; 0.2. This requires in turn, a 4fO' transverse emittance ~ 45 7rJ.Lm and a longitudinal emittance ~ 0.05 eV-s. The cyclotron H- beam emittance is much smaller, 3.5 7rJ.Lm horizontal, 27rJ.Lm vertical and 0.003 eV-s longitudinal. The stripping foil need only span the H- beam spot. Consequently the turn factor, or mean ratio of foil traversals to particle revolutions, may be reduced by displacing the injected H- beam and foil transversely and in momentum from the synchronous closed orbit [20,21] (Fig. 2.2.1). ... ' .• i . . '.~' . ->- ";~3!I~i1;;;¥.t .. -10 0 10 20 Y (mm) (a) ,......... > Q) ~O w <l -j180 -120 -60 0 ¢ (deg) (b) 180 Figure 2.2.1: Population of annular regions in phase space by charge exchange injection of a beam of small emittance displaced from the closed orbit. 2-24 The hollow distributions of Fig. 2.2.1 may be filled by altering the offset of the H- beam during accumulation or by moving the closed orbit from a position close to the injected beam on to the machine axis. The latter has advantages of further reducing the turn factor by adiabatically moving the stored beam off the foil and not steering the partially stripped neutral atom beam over the aperture. Angular scatter and energy loss in the stripping foil together with amplitude growth from collective instabilities or resonance effects will generate a halo of particles lying outside the specified accumulated emittance. A hands-on maintenance environment is taken to be :::;0.1 mSv Ih at 0.5 m following a cooldown period. Calculations and TRIUMF experience show that this would correspond to a distributed loss of 2 nA/m. However, the loss will be localized by means of 2 or 3 collimators. Experience shows that fields of 0.1 mSv Ih can be achieved within 14 m of a well shielded collimator intercepting 1 /-LA or less. The mean number of foil traversals per proton will be less than 200 to keep the halo component less than 1 /-LA. The combination, described below, of several such dynamic adjustments of closed orbit and H- energy coupled with a foil with two unsupported edges, has reduced the average number of foil traversals below 0.8% while satisfying both the constraints of beam stability during accumulation and subsequent acceleration and the limits on beam loss. The following advances have been made since the 1985 proposal was written: 1. The RAL and LANL spallation neutron sources have been operated at high intensity. These have given, and will continue to give, useful information regarding the lifetime of charge exchange foils and beam physics phenomena. 2. An emittance population, or "painting", scheme has been devised, using a foil with two unsupported edges ,. which reduces the average number of foil interceptions per proton to values similar to, or better than, the machines above. 3. A Monte Carlo tracking code has been further developed to include simple models of beam-foil interactions, transverse and longitudinal space charge, and collimation. 4. A more accurate calculation and a more useful parameterization of the Coulomb scattering cross-section have been made.[22] 5. Theoretical work on the stability of interim distributions generated during the in-jection process has been carried out.[23,24] The injection scheme described below avoids the generation of distributions hollow in longitudinal phase space. 6. The first steps towards an engineered design of an appropriately mismatched injection line and a new lattice insertion have been made. 2-25 Lattice Requirements Ideally [21 ] the injected H- beam should be achromatic and the lattice design should permit foil placement where (a"1 + 13"1') = 0 and "1 =f O. Dispersion at the foil automatically correlates particles of low momentum with a larger betatron amplitude.[25] Optimising injection to minimise the average number of foil traversals made by the cir-culating protons places constraints upon the lattice functions at the injection azimuth[26] and on the injected beam parameters. Specific constraints on the lattice functions at the injection azimuth are: ax = 0, a y = 0, "1~ = O. (1) To minimise the effects of angular scattering by the stripping foil, both the lattice and the injected beam J3x and J3y values at the foil should be as small as possible. However, to minimise the number of foil traversals the normalized dispersion parameter "1xl..[iJ; should be reasonably large, which requires a relatively large value of J3x in this case. Ideally: I laI3 = Vi - 2 JeixJ3iX I J3x "1x V}Jx !::"plp (2) where eix = injected emittance J3ix = beta-value of injected beam J3x = lattice beta-value at injection azimuth !::"plp = relative momentum spread of final beam bunch. An Accumulator lattice has been devised with a 7 m long injection straight satisfying (1) above, and for which: beta-values (no space-charge) beta-values (linear space-charge) normalized dispersion parameter (no space charge) normalized dispersion parameter (linear space charge) ideal normalized dispersion parameter (Eqn. (2)) f3x = 8.500 m, f3y = 3.000 m f3x = 8.518 m, f3y = 3.755 m 1]x/v'7Jx = 1.225 mt 1]x/v'iJx = 1.270 mt 1]x$x = 1.27(4511"), 1.6(6011") Apart from the vertical beta-function the lattice parameters are relatively insensitive to space-charge effects. The injected beam is mismatched into the Accumulator for both the dispersion and the x and y betatron parameters. A doubly achromatic beam is required at the injection azimuth with: 1]ix = 0 and 1]:x = 0 . The injected beam parameters are optimized for emittances of: eix = 3.5 7rmm mr and eiy = 2.0 7r mm mr to give: 2-26 aix = 0, (3ix = 2.97 m aiy = 0, (3iy = 1.34 m These values are obtained for beam phase-space ellipses which have beta values smaller than the matched values, but which still touch the painted emittance at one point. The limiting mismatch for small beta-values occurs when the beam phase-space ellipse touches the painted emittance at two points[21] but the reduced degree of mismatch allows more room within the phase-space acceptance for protons scattered by traversing the stripping foil. The achromaticity required in the injected beam parameters should make it relatively straightforward to maintain the mismatch parameters over the injection interval with the ramped injection energy. Details of the matching section at the end of the IA beam line can be found in Section 2.2.2. Injection System Layout The layout of the injection system within one of the three 7 m long straight sections is shown in Fig. 2.2.2 and Fig. 2.2.3. Lorentz stripping of 452 MeV H- ions restricts the maximum field of injection line elements to about 0.56 T. With the stripping foil positioned half way along the straight, there is sufficient straight section length to provide the long optical lever arms necessary to allow clearance of the upstream and downstream lattice elements. Four horizontally deflecting dc dipoles, HKI,2,3,4, arranged symmetrically either side of the stripping foil, produce a static localized closed orbit bump. This allows, in the dipole HK2 immediately upstream of the foil, the confluence of the H- injected beam with the proton beam circulating in the ring after the injection stripping. The H- beam is injected into HK2 by a dc septum magnet, DCI, which buttresses the dipole HKl. A similar dipole, DC2, buttresses the dipole HK4 and extracts any unstripped ions to an external beam dump. The dipoles HKI,2,3,4, are all of a septum design to minimise the separation required between the injected and extracted ions and the circulating proton beam. The four vertically deflecting dipoles, VKI,2,3,4, which produce the time-dependent localized vertical closed orbit bump, are placed symmetrically upstream and downstream of the dipoles HK1,2,3,4. With this arrangement of magnets the partially stripped HO beam can be allowed to drift in the straight section parallel to the circulating proton beam and then self extract between the downstream lattice dipole coils through a hole in the magnet yoke. The static injection point for the injected ions ensures that the unstripped and partially stripped ions can be conveyed with a small aperture beam line to external dump targets in shielded areas away from the Accumulator. Though some space exists between the HO beam and the circulating proton beam to ac-commodate the septum and vacuum wall for the dipole HK4, such an arrangement is not adopted as it is difficult to design the interconnections of the various vacuum chambers in the region. It may then be seen from Fig. 2.2.3 that the critical separation required for hardware is between the partially stripped HO atoms and the unstripped H- ions. 2-27 100 foo z ! j ... til .... Q SEPARATED FUNCTION a- INJECTION INTO ACCUMULATOR aORIZONTAL PLANE FOIL POSITION , a- INJECTION FOIL WIDTH aORIZONTALLY --~~--------------------------------i - --\ \\ / / ,r-1' ~' \\\ PROF:="E ~ DISPERSION ------------------- / / ' \ ---------------1 / I \ "-SWiPROFILE 60 1r mmlllr Figure 2.2.2: Separated function H- injection into the Accumulator - horizontal plane 50 SEPARATED FUNCTION H- INJECTION INTO ACCUHULATOR VERTICAL PLANE ---~---------------------------------------DISfAl'Cf .. tNe / / ~ -a--a-FOIL vlDTlI (A· 90 n _ mr) It- INJECTION DCI [l -EH3-~1 Figure 2.2.3: Separated function H- injection into the Accumulator - vertical plane 2-28 The magnet parameters for the dipoles are listed in Table 2.2.1. The pairs of horizontally deflecting dipoles up- and downstream of the foil have a separation between centres of 1 m, which results in a horizontal closed orbit displacement at the foil of 42 mm. The injection point is a distance Vc{3x - VCix{3ix = 16.35 mm further out. This gives a minimum clearance of 23 mm between the HO beam and the extracted H- beam into which to fit the two septa. The distance from the foil to the dipoles HK2 and HK3 can be arranged so that the resultant field from the ends of the magnets is sufficiently large to deflect the stripped electrons away from the foil onto an electron collector. Table 2.2.1: Magnet Parameters for Injection System Magnet DCl,2 HKl,2,3,4 VKl,2,3,4 Powered dc dc Programmable Length m 1.0 0.5 0.4 Vertical Aperture mm 10 50 50 Horizontal Aperture mm 70 140 140 Angular Deflection mr 116 42.3 15 Magnetic Field T 0.396 0.278 0.126 Ampere Turns A-turns 3124 11065 14070 The number of foil traversals may be reduced and an almost ideal longitudinal distribution generated by varying the energy of the incoming beam. Section 4.11 expresses a preference for a cyclotron extraction energy of 452 MeV H- ions (i.e., a proton energy of 451.5 MeV) and predicts a total energy spread of 1.3 MeV. The desired longitudinal emittance of the Accumulator can be painted by cavities providing an energy change from -0.65 MeV. to +0.65 MeV (Fig. 2.2.4). Suitable cavities providing a peak vol age of 350 kV and consuming 40 kW would be installed in the IA line (Section 4.3.6). This ramp, in conjunction with lattice dispersion at the injection point, simultaneously paints horizontal phase space. The vertically deflecting dipoles produce a maximum orbit bump at the start of injection and the bump is reduced stepwise to zero over the injection interval of 16 to 18 ms, with approximately 0.25 ms transition time between steps. The vertical closed orbit bump is programmed to obtain an acceptable transverse density distribution consistent with keeping the number of foil traversals to a minimum. The foil dimensions are as indicated in Fig. 2.2.2 and Fig. 2.2.3. The outermost dimensions are compatible with a "beam stay clear" recipe of 90 7rJ-lm plus, in the horizontal plane, an allowance for a relative momentum offset of 3.5 X 10-3• This takes into account the initial correlations of the longitudinal and transverse distributions but makes some allowance for any subsequent transverse coupling. 2-29 STRIPPING BEAM FROM CYCLOTRON ENERGY SWEEPING INTO BUCKET~~ _______ +3.5 MeV H 452 ~ ±0.65 MeV -2e --/ I -:::::::::--- -+2.6 "-" MeV 9· i_· ~-1.3 M~V -- --I--__ ..-J~ " P '\ ----~---. 451.5\MeV I - - _--1_ --__ *-"-t-"' I I \ 450.2 MeV BUNCH BUCKET~ \ \ J \ I \ I \ / '\ I " / " / " ,,-,," ............... _--.,...""" MeV t Figure 2.2.4: Longitudinal painting by modulating the energy of the incoming beam by ±0.65 MeV Foil Lifetime A foil thick enough to produce the equilibrium charge distribution would cause excessive scattering; 250 p.g/cm2 carbon is a compromise giving 99% conversion at 450 MeV. The foil lifetime is chiefly determined by radiation-induced alteration of the lattice, although at high temperatures this may be exacerbated by crystalline phase change. The TRIUMF cyclotron foil life corresponds to f'V 6 X 1019 protons/mm2• Aluminum oxide, pure metal, or thin carbon foils folded into a double layer may have a longer life. The Proton Storage Ring at Los Alamos has recently reported reduced scattering losses and extended life from a "postage stamp" carbon foil supported on thin carbon monofilaments. Simulations Foil lifetime requirements, scattering loss limits, etc, determine the average number of foil traversals. For a given incoming H- emittance the painted H+ emittance to provide this can be calculated from the geometric analysis of Ref. [20] and Ref. [21]. The actual foil traversals will be affected by other physical processes. In addition computer simulations have shown that, after accumulation, a few percent of particles lie outside, but close to, the boundary painted by the injection process. This penumbra is populated by small-angle scatter in the foil and incoherent coupled transverse and synchrobetatron oscillations. It may also be populated should the degree of mismatch (f3x) of the injected H- beam not be continually adjusted as the A-ring emittance is painted. A multiparticle three-dimensional tracking code, ACCSIM[27], has been written which includes many of these effects. It calculates the development of the ensemble during accumulation and yields foil-traversal information and details of beam loss. The munber of lost particles is small, by design, and a 2-30 better estimate of the distribution in the tails can be made by folding analytic descriptions of, for example, Coulomb scattering or energy loss with the foil traversal distribution, or single particle amplitude growth with details of the tune modulation given by ACCSIM. At the present time ACCSIM incorporates, with some degree of approximation • First order tracking of individual particles using a DIMAD lattice with chromatic effects and synchrotron motion. • Simulation of lattice perturbations by "thin" multipoles. • Energy loss, Coulomb and nuclear scattering in a stripping foil, collimator or internal target • Longitudinal and transverse space charge forces. The injected ensemble is drawn from specified distributions in transverse and longitudi-nal phase space. Beam parameters and the ring closed orbit may be varied to describe accumulation by multiturn phase-space painting. The output includes • Particle distributions in phase space and real space • Statistics and origin of particles lost by collision or lying outside a given acceptance • Figures of merit for the accumulated ensemble, for example, peak space charge tune shift, form factor, emittance and bunching factor • Beam-power distribution over the foil. Painting Program and Resulting Distributions Figure 2.2.5 shows the variation with time of the parameters painted, i.e., vertical closed orbit and incoming beam energy. Both betatron and synchrotron motion are used to reduce the number of foil traversals but in contrast to the 1985 proposal a foil with two free edges is used and the horizontal closed orbit is not varied. Following extraction the closed orbit sweep magnets and IA line cavities must be reset and stable before the start of the next accumulation cycle. A period of 4 ms has been allocated, but this may be shortened as engineering design progresses. This reset time, together with the elimination of 5 out of 45 cyclotron bunches and 90% H- extraction efficiency, means that the cyclotron must operate at 160 pA during the 16 ms filltime to provide 100 pA average current in the Accumulator. The scatter plots of Fig. 2.2.6 illustrate the distribution in phase space and in real space at an early stage and towards the end of accumulation. 2-31 Y(mm) Eylr1 (/Lm) 12- 60 ~~p~, 40 20 10 4 10 15 (P-~)/Ps 4 *103 ~m.~~ 2 10 15 Exhr (p.m) 60 40 20 10 5 10 15 TIME (ms) Figure 2.2.5: The shaded bands are the horizontal, vertical and momentum amplitudes painted as the Accumulator vertical closed orbit (Yeo) and incoming beam energy (To) vary with time. A deliberate correlation between horizontal and vertical transverse amplitude and between horizontal and longitudinal amplitude is seen. The former ensures that no particles are injected with both a large x and y amplitude; the latter reduces the horizontal width of the physical space required to retain beam. Foil interactions do not significantly modify the main distribution. The average number of foil traversals per proton is 55 (Fig. 2.2.7). Synchrotron motion, chromaticity and space charge modulate the tunes, and resonances in coupled motion (without amplitude growth) can alter the number of foil traversals up to rv 25% over narrow intervals (Fig. 2.2.8). In order to generate flat distributions, and lower the peak tune shift, the dwell time should be roughly proportional to the amplitude (transverse or longitudinal) being painted. This cannot be achieved for both horizontal and longitudinal planes because of the technique used of exploiting dispersion at the foil to paint both planes simultaneously. We have chosen to optimize the longitudinal line density. 2-32 -~6=0~-~40~-~20~~0~~20~-4LO~60 Xb (mm) 40 20 E 0 . 5 >--20 -40 -60 8 1 0 ;. -4 '" ... -8 -60 -40 -40 FOIL -20 0 20 40 60 X (mm) 0 -20 0 20 40 60 Y (mm) -4~~~~~~~~~~ -180 -120 -60 0 60 120 180 ~ (dog) (a) 10 /l x - 4 ·~·D:···· · ": . '--;:-.; .. ,.' :, ' _8~-L~~~~J-~~~ -60 -40 -20 0 20 40 60 40 20 E 0 5 >-- 20 -40 -60 8 1 0 ;. '" ... -4 -8 -60 -40 -20 -40 -20 Xb (mm) FOIL 0 20 40 60 X (mm) , 0 20 40 60 Y (mm) -4~~~~~~~~~~ -180 -120 -60 0 60 120 180 ; (dog) (b) _B~-L~~~~-L~~~ -60 -40 -20 0 20 40 60 40 20 E 0 5 >--20 -40 -60 8 4 ]:0 ;. -4 w ... -8 -60 -40 -40 Xb (mm) -20 0 20 40 60 X (mm) , .' . . \ :! -20 0 20 40 60 Y (mm) -4 ~--'-~...L........::::J=----'-~~~ -180 -120 -60 0 60 120 180 ~ (dog) ( ~ ) Figure 2.2.6: Distribution of accumulated beam painted by the program illustrated in Fig. 2.2.5: (a) after 4000 turns, (b) at the end of accumulation, 16,000 turns, (c) at the end of accumulation when the effects of longitudinal space charge have been included in the calculation. 2-33 .14 I I I I .12 - -(f) Q) . 10 U - -+-' ~ 0 CL.08 - -4-0 c: .06 - -0 +-' U 0 .04 - -~ LL .02 - -.00 ~ I I I o 100 200 300 400 500 Number of foil traversals Figure 2.2.7: Fraction of stored particles executing a given nwnber of foil traversals. (The average number of traversals is 55) 60 55 50 .-.J H c::l LL 45 :z: c::l (f) 40 t--H ::r:: 35 30 25 1.4 1.6 1 . B 2.0 2.2 2. 4 2.6 DE (ME V) Figure 2.2.8: The number of foil traversals in 1000 turns as a function of longitudinal amplitude for an injected H- beam with small emittance. The calculation includes the energy dependence of both Vs and V X ' 2-34 The longitudinal space-charge potential is computed from a smoothed numerical differ-entiation of the linear density of macro-particles. The transverse space charge potential is computed semi-analytically from the rectangular density distribution associated with the Lissajous motion of each macro-particle; the potentials are summed and a multi pole expansion fitted to the result.[28] The transverse tune shift for each particle is calculated from the latter (scaled by the linear density) and the turn-to-turn betatron phase advance modified in the tracking program. Figure 2.2.9 shows the distribution in tune shift of the ensemble at the end of accumulation. ~ 5.70 5.65 5.60 5.55 -f---,----,---r----,---,---t-3.55 3.60 3.65 3 .70 3.75 3.80 3.85 l/x Figure 2.2.9: Distribution in betatron tune space calculated for the ensembles of Fig. 2.2.6c [( x) after 4000 turns (+) after 16,000 turns]; the calculation ignored the effect of dispersion in reducing the mean particle density. Including this effect would drop !:1l/x by 0.04. The variation of tune with momentum is less than the space-charge tune shift but is present even at low intensities and increases as well as decreases the tune. This can enhance resonant behaviour in foil traversals and possibly feed some particles into the half integer resonance. The reference scheme does not paint the largest momentum amplitudes until late in the cycle when the space-charge tune depression is significant. The final tune shifts are significantly smaller than the imposed limit of 0.18 (see Section 2.1.2). In further studies it will be attempted to decrease the painted emittances, thereby further increasing the stay-clear safety factor. Figure 2.2.10 shows the fraction of particles lying outside a given emittance at the instant of extraction for the simulation discussed above. Not all particles executing non-linear or coupled motion will have their maximum values at this time. Particles whose emittance exceeds a certain value at any time during accumulation are tagged. These are the quan-2-35 tities to be considered when determining aperture or designing collimators. The studies are not yet complete and will be reinforced by dynamic aperture calculations using the second-order code DIMAD. ~,"'::" .............. , .... , .... .: 10 ............ ~AUSSIAN .: 10 , , , , ............ ~AUSSIAN w Cl Vi 5 o WITHOUT FOIL INTERACTIONS INCLUDING FOIL INTERACTIONS w Cl Vi 5 o ~ , , , , \ INCLUDING FOIL WITHOUT FOIL INTERACTIONS \ INTERACTIONS , , , , , , , , , , , , , O. 1 +-...,....-.-...,....-.-~.--.--.-....--+-.,..-.-..--.-..--+ o 10 20 30 40 50 60 70 80 ty (mm-mr) 0 . 1~~,-~,-~,-~.-+-.-~+-~.-~+ o 10 20 30 40 50 60 70 tx (mm-mr) Figure 2.2.10: Emittance distributions for an ensemble accumulated by the painting scheme of Fig. 2.2.5 alone and when including foil interactions. The Gaussian is for 4fO' = 457rJ.l m 2-36 80 2.2.2 Transfer Line Design Beam Transport from the TRIUMF Cyclotron to the Accumulator Momentum measmt. The beam of H- ions extracted from TRIUMF will have ,J.. emittances of Cx = 3.57r mm-mr, Cy = 2.07r mm-mr and an energy spread of ±0.6 MeV and consists of bunches 2.5 ns wide at 43.4 ns intervals. It is cavities transported from the Injector cyclotron to the Accumulator (A ring) in Stripper 4 the IA transfer line. This beamline produces the beam characteristics and provides the energy modulation required for the injection scheme described in section 2.2.1. Further, there is provision for beam energy Chopper 4 and emittance measurements of the beam from the cyclotron and for installation of a chopper. The purpose of the chopper, which is described in detail in section 3.3.2, is to remove selected bunches from the injected beam such that a gap is created in the circulating beam of the A ring. This gap, which is transmitted to succeeding rings, allows operation of the kickers which transfer beam from one ring to the next. The IA beamline consists of four main sections. The first, internal to the cyclotron vault, produces a doubly achromatic beam and prepares it for exit from the vault. A long straight section follows; its purpose is to transport the beam the appropriate distance from the accelerator building. This is followed by an 8-quadrupole doubly achromatic sec-tion which provides most of the bending required to direct beam to the injection point of the A ring. The last is a matching section between the transported beam and the A ring. Each section is briefly described below. An overall view of the transfer line is shown in Fig. 2.2.11. Upon extraction from the cyclotron the beam is made doubly achromatic by a 3-quadrupole 2-dipole system. There follows a series of quadrupoles which provide a parallel beam for a drift of I'V 12 m through the vault wall to the external part of the beamline. The long straight section external to the vault consists of a 4-quadrupole matching section. Emittance-measuring equipment is located at its end. Two negative identity sections follow. Each is 40.2 m long and is com-posed of two quadrupole doublets separated by 19.3 m. The chopper is located in the last of these. Figure 2.2.11: The complete 1A transfer beamline 2-37 E measmt . 4 r Chopped bunches are kicked up into a stripper foil located at the object point of the second identity section. There the centre of the kicked beam is 10 mm above that of the unkicked beam while the beam height is (nominally) 3 mm. (Vertical kicking was chosen because chopper design is easier and because a smaller beam profile could be attained throughout the straight section.) The protons emerging from the stripping foil are deflected to the right (looking along the beam) into a beam dump by the first dipole of the second achromatic section. The achromatic section is composed of 6 dipoles and 8 quadrupoles. It is doubly achromatic and has a transfer matrix of +1 in the horizontal plane and -lin the vertical. The total bend of this section is 99 degrees and permits an energy measurement at its mid-point. The final matching section consists of 2 quadrupole doublets, a dipole, a quadrupole triplet and 2 injection septa. Between the doublets is a 7.6 m straight section in which the rf cavities are placed. They provide the energy modulation required for A-ring injection. Because of this energy ramping, the 7 quadrupoles are varied with beam energy such that, in combination with the injection septa and dipole, an achromatic double waist is attained at the A-ring injection foil. There the beam is off-axis both vertically and horizontally; beam dimensions are x = ±3.22 mm and y = ±1.64 mm. 3 x 0 --..... E-1 U '-...-/ >--2 I , , , , -3 0 £ Measmt. J. 40 Chopper Energy Chopper Foil Measmt. .j. .j. .j. :' - - -,: O.5R 16 Injection Foil J. '; (cm/%) 80 120 160 200 Z (m) Figure 2.2.12: Beam profiles along the IA beamline Figure 2.2.12 shows the beam profiles along the beamline. The dispersion function is shown as a dotted line. H- ions not stripped by the injection foil are directed to a beam dump by 2-38 a 15-degree dipole. The section transporting unstripped H- to the beam dump is included in figure 2.2.12. Any HO atoms produced in the foil are also stopped in that dump. Beam Transport from the Accumulator to the Booster Beam transfer from the A ring to the B ring begins in a DOFOD straight section. Extracted beam is deflected down and brought back into a horizontal plane 1.1 m lower by a double reverse vertical bend system. A 60-degree horizontal bend section, similar to the A-ring configuration, follows. At its end a second double reverse vertical bend takes the beam down another 1.1 m to the B-ring beam plane. The transfer line is designed to accept horizontal and vertical emittances of 907r mm-mr; it is capable of matching into the B-ring acceptance of Cx = Cy = 1207r mm-mr. I I I I +K1 Sl -o \ \ -K2 \ r\ I\+- 82 - 1\ -B1 ~\ 1\ \ \ r-- r-- r-- r-- - --2 - \ -\ '-- '-- '-- '-- - -+ l' -l- l' + f + t S2 K2 -3 I I I I 30 40 50 60 70 80 Z (m) Figure 2.2.13: Beam transfer from the A ring to the Bring Figure 2.2.13 shows the layout of the transfer line. The arrows above the ordinate axis indicate F quadrupoles by a i and D quadrupoles by a L. Quadrupole polarities in the hor-izontal section of the transfer line are identical to those of the corresponding quadrupoles of the A and B rings. The quadrupoles in the vertical dog-leg sections are D quadrupoles. Extraction from the A ring is begun by two kickers, one which kicks 2 mr located upstream of the first D quadrupole and the other which kicks 8 mr immediately downstream of it. A septum 1 m long with a field of 0.89 T is located downstream of the F quadrupole to bring the total vertical bend to 18 degrees. This is followed by a quadrupole and an 18-degree 2-39 reverse-bending dipole to bring the beam into the horizontal plane. A similar system is used for injection into the B ring. In this case, however, the two kickers are replaced by a single kicker of 10 mr because there is insufficient space downstream of the B-ring D quadrupole to install a second kicker. All dipoles and quadrupoles in the transfer line are of the types used in the A ring. The beam is matched to the required values of /3x, ax, /31J! ali' "7x and "7~ in the Bring. In the vertical plane the beam is dispersionless-that is, "7y = 0 and "7;= o. Figure 2.2.14 shows the beam profile along the transfer beamline. 8. 6. 4. / 2. O. -2. -4. -6. 30 € = € = 901T mm-mr I \ R111 (em/%) / \ x y / R36 (cm/%) .' / \ 40 50 60 70 Z (m) \ \ 80 Figure 2.2.14: Beam profiles in the A ring to the B ring transfer beamline Beam Transport from the Booster to the Collector Because of the large separation between the Booster and the Collector, this transfer line is the longest (......,200 m) of the KAON complex. It is somewhat further complicated in that the Collector lies 4.5 m above the Booster and injection takes place from above (in order to allow the future installation of a higher energy ring above the Extender). Consequently the beamline has been subdivided into four sections. The first consists of Booster extraction and a vertical rise of 5.7 m. This is followed by a short section containing a 10-degree horizontal bend. The third is a long FODO section which transports the beam to the Collector. Injection into the Collector constitutes the final section. The beamline is designed to carry an emittance of 3271' mm-mr and to be matched into the Collector whose acceptance is 4071' mm-mr. 2-40 Extraction from the B ring takes place in the vertical plane. A 7.7-mr kicker magnet is located immediately downstream of the first D quadrupole in a DOFOD cell. Two septa, one thin and one thick, are placed downstream of the F quadrupole. These are followed by a 7 -quadrupole reverse (vertical) bend system which carries the beam to a height of 5.7 m above the beam plane of the Booster. This system rises at an angle of 10 degrees and is dispersionless in the vertical plane at its exit. Following the vertical section is a 3-quadrupole 10-degree horizontal bend system which is tuned to produce an achromatic beam in the horizontal plane. The dipole is followed by a quadrupole doublet which brings the beam to a double waist. Thus, at the location of the waist, the beam is dispersionless in both horizontal and vertical planes. Figure 2.2.15 shows the vertical section of the B to C transfer line. (The appearance that elements are not perpendicular to the beamline is caused by the different vertical and horizontal scales, not design.) Figure 2.2.16 shows the beam profiles along it. Next a 4-quadrupole matching section is used to match into an FODO array which is identical to that of the Collector straight. It carries beam some 100 m to the Collector injection section. Initially the injection line is 1.2 m above the Collector. Although it would appear obvious that beam could be injected into the Collector with a single vertical bend, it was not possible to find a solution which met all matching requirements and allowed installation of 'real' quadrupoles. Consequently injection into the Collector is achieved with a double reverse-vertical bend system arranged so that the injected beam is dispersionless in the vertical plane. Six quadrupoles in the horizontal straight section of the transfer line match vertical and horizontal phase spaces to those required by the Collector lattice. Injection into the Collector is accomplished with a septum located upstream of a D quadrupole in the Collector straight and one 4 mr kicker located upstream of the following D quadrupole. Figure 2.2.17 shows the injection section of the B to C transfer line; beam profiles along it are shown in figure 2.2.18. Beam Transport from the Collector to the Driver Extraction of beam from the Collector is achieved with one 4 mr kicker located downstream of a D quadrupole in the Collector straight section and a septum placed downstream of the following D quadrupole. The septum is part of a reverse vertical bend system. A 6-quadrupole array for phase space matching and another reverse vertical bend section follow. The Driver injection septum is part of the latter and is placed upstream of a D quadrupole in the Driver. One 4 mr kicker is located upstream of the following D quadrupole. The transfer line is designed to carry an emittance of 407r mm-mr and to be matched into the Driver whose acceptance is 487r mm-mr. Figure 2.2.19 shows the elements of the transfer beamline. Beam profiles along the transfer line are shown in figure 2.2.20. 2-41 7 6 5 4 ~ ~3 >- 2 o -1 7.5 5.0 2.5 0 .0 -2.S -S.O -7.5 10 20 30 40 50 Z (m) Figure 2.2.15: The vertical section of the B to C transfer line I I 10 " J \ I , R16 I \ \ 20 I , I \ I " 30 SR36 (cm/%) € = € = 321T mm-mr x y 40 50 60 Z (m) 60 70 Figure 2.2.16: Beam profiles along the vertical section of the B to C transfer beamline 2-42 6 .5 I I B1 6.0 - I-B2 83 5.5 - [\ ,.... ~ r--.. E '-../ 5.0 "" S1 K1 - 1\ ~ .-->- ...... "" 4.5 '-4 .0 - ~ .j, t .j, t ,j. t ,j. t ,j. 3 .5 I I 100 125 150 175 Z (m) Figure 2.2.17: Injection into the Collector 6 € = E = 3211' mm-mr • y 3 I '-0 - - ./ "- "- - - - --"- ;-"-/ / I -3 I /' \ / (em) \ / \ / 5RJ6 (cm/%) -6 100 125 150 175 Z (m) Figure 2.2.18: Beam profiles along the injection section of the B to C transfer line 2-43 >-5 .5 I I K1 S1 r-4.5 \ ""- S2 K2 r-3.5 - 1\ -'--+ l' + l' + l' .j.. t ,j. t ,j. 2.5 I I 30 60 90 120 Z (m) Figure 2.2.19: Beam transfer between the Collector and Driver 6. 3. O. -3. -6. 30 - - - --€ x € y / \ \ 5R36 (em/%) - - --_./ \ /-/ / 401T mm-mr -x (em) 60 90 120 Z (m) Figure 2.2.20: Beam profiles along the C to D transfer line 2-44 For operation with polarized beam a 1.5 m long, 4.5 T solenoid is installed at the mid-point of the horizontal section of the transfer line. This is necessary because the vertical polarization of the Collector must be rotated into the horizontal transverse plane for stable acceleration in the Driver. In this operating mode phase space matching is still possible. A slight mismatch in vertical dispersion occurs, however; design is such that rJx = rJy = 0, rJ ~ = -0.006 and rJ; = 0.011. Beam Transport from the Driver to the Extender Transfer from the Driver to the Extender differs from the other transfer lines in that it takes place in the horizontal plane, the horizontal separation being 3.5 m. The high energy of the extracted beam combined with the (relatively) short inter-quadrupole spacings of the Driver straight sections makes extraction somewhat difficult. The transfer beamline is designed to match a 671" mm-mr emittance from the Driver to a 671" mm-mr emittance in the Extender. Figure 2.2.21 shows the elements of the transfer beamline. 5 I I K2 - - f-4 / v - - f-3 V V S4,5,6 - 82 4 / t 83 -V - B1 4 -K1 V r- ./ o '-t ~ tSl,2,3 t + t ~ t + t ~ t ~ -1 I r I 20 60 100 140 Z (m) Figure 2.2.21: Beam transfer between the Driver and Extender One 2.5 mr kicker is placed downstream of the first F quadrupole of the Driver straight. Beam is extracted with three septa located downstream of the next F quadrupole. The first of these septa is thin (9.5 mm), 1.2 m long and is pulsed to provide a field of 1.2 T. The second and third septa are 1 m and 1.6 m in length and produce fields of 1.4 T and 1.5 T respectively. Extracted beam passes between the poles and coils of the follow-ing D quadrupole. There follow a dipole, two quadrupoles, a second dipole and another quadrupole. 2-45 Because the quadrupole spacing in the Driver and Extender straights is different, the re-mainder of the transfer line differs from that described above. A quadrupole, a single dipole and a quadrupole doublet prepare the beam for injection into the Extender. Injection septa are identical to those of the Driver and are followed by one 2 mr kicker. Beam profiles along the transfer line are shown in figure 2.2.22. 5.0 e = € = 67r mm-mr )( y 2.5 0.0 -2.5 5R16 (cm/%) -y (em) -5.0 20 60 100 140 Z (m) Figure 2.2.22: Beam profiles along the D to E transfer beamline The first elements of the fast extraction beamline to the neutrino area are identical to those of the D to E transfer line. A more detailed description of that line is given in the KAON Science and Experimental Facilities report. Slow Extraction from the Extender It is proposed that the slow extracted beam from the Extender be split into three or four parts, each feeding a separate section of the experimental hall. A study of the extraction mechanism is detailed elsewhere[29]. This section presents an outline of the extraction hardware developed in that report. Slow extraction is accomplished with a 3-element configuration consisting of a preseptum and an electrostatic septum, each of which deflect in the horizontal plane, and a mag-netic septum which bends vertically. The preseptum is located downstream of the first F quadrupole in the long straight of the Extender. It consists of aIm length of 33 /-lm 2-46 carbon wires which are spaced at 5 mm intervals. Bulging caused by electrostatic forces is cancelled by the use of a symmetric field. The electrostatic septum, located downstream of the next F quadrupole, has a length of 5 m and a field gradient of 50 k V / cm. It is assumed to be constructed of 50 pm molybdenum strips. Because it is massive along the beam direction, the shadowing effect of the preseptum is required. The magnetic septum, a Lambertson magnet, is located downstream of the next (centre) F quadrupole of the Extender. At its entrance the circulating and extracted beams are separated by 3 cm. It is 5 m long and run at a field of 0.8 T. Following the magnetic septum are three quadrupoles and a reverse-bending vertical dipole. These elements are followed by two horizontally-bending dipoles with a quadrupole located between them and one following them. When brought back into the horizontal plane the extracted beam is 75 cm above the Extender. Immediately downstream of the second quadrupole is installed a second preseptum which begins the beam splitting to the exper-imental areas. Optics of the common section produce a point-to-point focus between the first to the second preseptum. The slow extraction system is illustrated in figure 2.2.23. 1.5 I I I I I VB1 !. J.HB1 J.HB2 - PS2 -1.0 ES1 MS1 / - -0.5 - /' / ,; PS1 / >- 0.0 /" - /' -0.5 - r-t ,J. t ,J. t J. t J. t -1.0 I I I I I -60 -40 -20 o 20 40 60 Z (m) Figure 2.2.23: Extender slow extraction system In order to estimate the losses in this extraction channel a file of particles was generated usmg a Monte Carlo simulation of resonance extraction. The extraction channel was 2-47 simulated to the second preseptum in a REVMOC[30] calculation. (REVMOC is a TRIUMF Monte Carlo program which simulates energy loss, multiple and nuclear scatterings, nuclear absorption and interactions with apertures in a beam transport system.) Recorded losses were 8 ppm between the preseptum and the electrostatic septum, 5 ppm inside the latter, 16 ppm between the electrostatic septum and the magnetic septum and less than 1 ppm in its yoke. Beam losses between the magnetic septum and the second preseptum are less than 1 ppm. Total beam loss in this extraction section is estimated to be less than 50 ppm. This value does not include any losses occurring in the non-extracted beam caused by non-linear motion in the Extender. Further details of beam splitting and delivery to the experimental areas are given in the KAON Science and Experimental Facilities report. 2.2.3 Slow Extraction The challenge in designing the slow-extraction system for the KAON Factory is to reduce the extraction losses, typically on the order of 1 % in today's slow-extraction systems, to the 0.1 % level. This provided the motivation for the design of the racetrack lattices as outlined in Sec. 2.1. The extraction system itself includes an electrostatic pre-septum in addition to the usual electrostatic and magnetic septa. Being quite short, this pre-septum can be made very thin and separates the extracted beam cleanly from the circulating beam. The extraction dynamics were studied both analytically and by using a simulation code in order to identify the factors that increase extraction losses.[31,32] The extraction system makes use of the horizontal third-integer resonance at 3vx = 40, ex-cited by 2 sextupoles of about 40 T /m2 strength located opposite each other in the straight sections, so as not to excite any odd-harmonic resonances. The phase-space trajectories of the particles become distorted and the phase space separates into stable and unstable regions, between which, on the so-called separatrices, the particles move outwards to be extracted by the septa. A beta function of 100 m was chosen at the septa in order to achieve a stepsize of 10 mm without making the ring too sensitive to misalignments. A plot of the horizontal phase space including the separatrices is shown in Fig. 2.2.24. The pre-septum will be aIm long double-gap device with a gap of 20 mm for the extracted beam and 80 mm for the circulating beam. Very thin wires of a low-Z material, 33 pm carbon fibres, are used in order to minimize the effective thickness presented to the beam and to keep scattering angles low. In order to prevent transverse forces on the wires both gaps will have the same electric field of 12.5 kV /cm, giving a differential kick of 0.08 mr. This small kick is sufficient to create a gap of 1-2 mm at the electrostatic septum located one cell downstream. The main septum will be 5 m long with a single gap of 20 mm and 50 pm molybdenum strips (as in the deflector for H- extraction from the cyclotron) 2-48 5 3 - -1 X -3 f3 = 98 m a = 4.2 7J=5m € = 0.1771' mm-mrad 6 X (mm) Figure 2.2.24: Horizontal phase space for the Extender at the onset of slow extraction. The inset gives an enlarged view of the extracted beam. operating at 100 kV and giving a kick of 0.8 mr. The gap of about 30 mm created by this device at the magnetic septum is sufficient to provide a 0.8 T field over a length of 5 m in order to extract the beam. For reasons of reliability a Lambertson septum with a 20 mm separating yoke is used, thus avoiding a current sheet in a critical position. In Fig. 2.2.25 the extraction straight section is shown including the extraction sextupoles and the septa. Provided all septa are accurately aligned, our simulations show that more than 90% of the septum hits will occur at the pre-septum. The effective thickness presented to the beam by this septum is the sum of the thickness of the first wire and the projection of the beam divergence over the length of the septum. It is therefore necessary to keep the pre-septum short and the beam divergence small; for a fixed size of the extracted beam the latter is equivalent to keeping the extracted emittance small. A 0.5 m long preseptum with 10 pm thick tungsten wires was initially proposed; with a divergence of the extracted beam of 20 prad, corresponding to an emittance of 0.271" mm-mrad, about 0.2% of the beam would hit the septum. Almost all of these particles would be lost as the scattering angles in the tungsten wires are rather large, about 80 prad rms. The pre-septum proposed in this report presents significantly less material to the beam, consequently the rms scattering angle is only about 4 prad and more than half of the scattered particles are extracted in a controlled way. Since the vertical beam divergence at 2-49 80 {3x 60 {3y r]X 40 (m) 20 m 0 n 395 ,ES ,MS o nom / \ / \ \ / \ / \ / \j v \ 447 498 550 DISTANCE (m) Figure 2.2.25: Extraction straight of the Extender. PS is the pre-septum, ES the main electrostatic septum and MS the magnetic septum. the preseptum is about 0.6 mrad, no significant vertical blowup is caused by the preseptum either. Although with this device the number of septum hits is increased to about 0.4%, the losses are kept at 0.2%. As the electric field is lower the length is increased to 1 m. The theoretical limit for the extracted emittance is extremely small, about 10-4 71" mm-mrad in our case. However, the emittance in any real system is dominated by imperfections, most notably the stability of the quadrupole power supplies. Ripple and noise produce beam divergence both by creating a modulation dv in the machine tune and thus modulating the stable phase-space area, and by creating modulations d{3 and da in the lattice functions, which in turn create a variation in the position of the separatrices. The extracted emittance can be estimated to be dx ( , d{3 ) fcirc dv fextr = -73 x "2 - xda + 2 8 v dB' where dx is the half-width of the extracted beam and dv/dB = L/(271"{3c)dv/dt describes the speed of change in v. To keep this value at or below 0.271" mm-mr, the stability of the power supply must be about 10-4 or better. The slow-extraction system in the Extender will support both chromatic and achromatic extraction. In the former, the chromaticity of the ring is cancelled by sextupoles in the arcs, and the extraction process is initiated by moving the tune of the machine towards the resonance using the quadrupoles in the straight sections. The circulating beam becomes unstable starting from large betatron amplitudes and is peeled off by the septa. In this 2-50 way, the full momentum range of the circulating beam is extracted at all times. In the chromatic extraction scheme the machine is set to a chromaticity of about -10 horizontally. The stable phase space area becomes a function of the momentum of the particle and so particles at lower momenta are extracted first. In this way the momentum bite of the extracted beam can be reduced and, as a consequence of the synchrotron oscillations of the particles, so can the lengths of the extracted bunches. While the reduced momentum bite is desirable in primary-beam experiments using e.g. polarized protons, short bunches are important for particle separation in the secondary channels at lower energIes. In order to maintain low divergence in the chromatic mode, the dispersion at the septum has to fulfill the following relationship: e = 6A (Jp (sin8s + Q cos 88 ) + l1'focoS8s ) , where A is the sextupole strength normalized to the beta function and 88 its azimuthal angle relative to the septum. Associated with chromatic extraction is an increase in extracted emittance due to syn-chrotron oscillations: and it is this relationship that limits the reduction in momentum spread in our case to about 60%. In order to simulate the slow extraction, a computer code SLEX has been developed at TRIUMF. This code allows simulation of the particle's motion in 4-dimensional phase space (x, x', 1>, op!Po) including the effects of power-supply noise and ripple, and chromaticity of both the tune and the lattice functions. Both chromatic and achromatic extraction have been investigated. The code does not take into account third and higher-order aberrations that tend to deform the separatrices; work is under way to include this and also vertical betatron oscillations. Tracking in DIMAD indicates that third-order aberrations can be cancelled very effectively by octapoles at the symmetry point of the straight sections. Simulation results for chromatic and achromatic extraction are given in Table 2.2.2. Fig. 2.2.26 shows the intensity distribution for achromatic extraction, for 10-4 relative stability of the power supply, resulting in a duty factor of about 55%. The duty factor is limited by the power supply stability, indicating that 10-5 is necessary if the duty factor is to reach the 85% predicted for perfect stability. Chromatic extraction is performed over a wider tune range due to the need to scan the full momentum width of the beam; as a consequence the influence of ripple and noise is somewhat less and the duty factor is about 2-51 64%. The longitudinal phase space is shown in Fig. 2.2.27 for both chromatic extraction together with the ellipse occupied by the circulating beam. Using chromatic extraction a reduction of 50-70% can be achieved in both quantities simultaneously, and a minimum extracted bunch length of 1 ns can be achieved. The extracted horizontal phase space is shown in Fig. 2.2.28; the emittance of the smallest ellipse around the beam is 0.1911" mm-mrad. Scattering in the pre-septum was simulated in separate runs[32] and the resulting emittance, given in Table 2.2.2, indicates that the tails created by scattering are small. (see Sec. 2.2 of this report). Table 2.2.2: Slow Extraction Simulation Achromatic Chromatic extraction extraction units Tune range tl.1I 0.011 0.032 Power-Supply noise dllnoi8e 10-4 10-4 Chromaticity of II e 0 -10.7 " of f3 (df3/dp)po 667 667 m " of a (da/dp)po 0 0 Synchrotron tune 118 0.0016 0.0016 Duty Factor 55±4 64±4 % Septum Hits 0 1 (of 2000) Fraction not extracted 1.8±0.3 3.7±0.4 % Extracted Emittance w/o scattering in septum (100%) € 0.1911" 0.1711" mm-mrad with " (87%) € 0.0611" mm-mrad with " (98%) € 4.211" mm-mrad Momentum Bite dp/po 0.21 0.088 %FWHM Bunch Length d¢ 2.4 1 ns 2-52 3000 D.F. = (55±4)% ....... fIl 2500 .., ·a ::s .ci 2000 '"' $ 1500 >. .., .v; s:: 1000 III .., oS 500 0 0 soo 1000 1500 2000 Turn II Figure 2.2.26: Intensity distribution for achromatic slow extraction. 0.10 ........ /" ........... ~ 0 .05 > Q) S -0.0 \ ) Po '-..... ./ 'ti -0.0 ----0.10 -0. 15+-r--,-,--,-,rr-r-.-." ... -ro,--r--r-T--.-,--,-,r-.-r-.-...... + -60 -40 -20 o 20 40 60 rpC) Figure 2.2.27: Longitudinal phase space of the extracted beam for chromatic slow extrac-tion. - 0 .8 +-.lL.-.1........I--'---'--'-""""'---.1........I--'---'--'-...I!O..1........I--'---'--'-""""'---+-f c .. = O.17n mm -mrad - 1.0 ~ '"0 -1.2 to h 8-1.4 '--"' X -1.6 - 1.8 -2.0 +-..,...-,.....,-,--.,--,-..,...-,.....,-,--.,--,-..,...-,.....,-,--.,--,-..,...-+-35 40 45 50 55 X (mm) Figure 2.2.28: Horizontal phase space of the extracted beam. 2-53 References [1] E.D. Courant and H.L. Snyder, Ann. of Physics 3, 1 (1958); L.C. Teng, NAL Report FN-207, (1970); L.C. Teng, Part. Accel. 4, 81 (1972); S. Ohnuma, NAL 15 Note 105, (1980), and J.I.M. Botman, R.C. Gupta, and M.K. Craddock, High Transition Energy Magnet Lattices, IEEE Trans. Nucl. Sci., NS-32, 2308 (1985). [2] R.V. Servranckx et al., Users Guide to the Program DIMAD, SLAC Report 285 UC-28 (A), May 1985. [3] F.W. Jones, DIPLOT: A Post-processor for DIMAD, TRI-DN-87-22. [4] R.V. Servranckx, U. Wienands, and M.K. Craddock, In Search for a Booster Lattice, TRI-DN-89-K48. [5] R.V. Servranckx, New Lattices for the C,D, and E rings, TRI-DN-88-3. [6] R.V. Servranckx and U. Wienands, Lattices for the Extender Ring of the KA ON Factory, TRI-DN-89-K58. [7] M.C. Crowley-Milling and J.J. Rabinowitz, IEEE Trans. Nucl. Sci. NS-18, 1052 (1971). [8] A. Piwinski and A. Wrulich, DESY 76/07 (1976). [9] SPEAR Group, IEEE Trans. Nucl. Sci. NS-24, 1863 (1977). [10] T. Suzuki, Method to Suppress Synchrobetatron Resonances by by Proper Arrangement of RF Cavities, TRI-DN-84-61. [11] R. Baartman, Optical Effects of Tuner Solenoids TRI-DN-89-K46. [12] G.F. Wellman, Transverse Space Charge Simulation in DIMAD, TRI-DN-89-K84. [13] R.A. Baartman, Synchrobetatron Resonances Driven by Dispersion in Rf Cavities: A Revised Theory, TRI-DN-89-K40. [14] P. Schmor et al., Recent Results from the TRIUMF Optically Pumped Polarized Ion Source, High-Energy Spin Physics 8th Int. Symposium, Minneapolis, Minnesota. AlP Conf. Proc. No. 187, Particles and Fields Series 37, 1989, p.1210. [15] E.D. Courant and R.D. Ruth, BNL Report 51270. [16] A.D. Krisch et al., Phys. Rev. Lett. 63, 1137 (1989). [17] E.D. Courant, Proc. 8th Int. Symposium on High Energy Spin Physics, Minneapolis, MN, p. 1085 (1988). 2-54 [18] U. Wienands, Proceedings of the 1"t European Accelerator Conference, Rome, Italy, p. 905 (1988). [19] T. Roser, Proc. 8th Int. Symposium on High Energy Spin Physics, Minneapolis, MN, p. 1442 (1988), and L.C. Teng, ibid., p. 146l. [20] J. R. Richardson, Some Suggestions on the Extraction of a Continuous H- Beam from TRIUMF and the Injection thereof into an Accumulator Storage Ring TRI-DN-83-30. [21] C. W. Planner, A Preliminary Investigation of Charge Exchange Injection for the TRI-UMF KA ON Factory using Simultaneous Stacking in the Longitudinal and Transverse Phase Planes TRI-DN-84-63 [22] M. Butler, S. Koscielniak, Scattering Cross-Sections for 9 Be, 12C, 16 0 and 27 AI. TRI-DN-89-K27. [23] R. Baartman, F. W. Jones, S. Koscelniak and G. H. Mackenzie, Stability of Beams Hollow in Longitudinal Phase Space, Proc. PAC Meeting, Chicago, March 1989. [24] R. Baartman and S. Koscelniak Stability of Hollow Beams in Longitudinal Phase Space Proc. of Int. Conf. on High Energy Accelerators, Tsukuba, 1989. [25] Proceedings of High Energy Physics in the 1990's (Snowmass) World Scientific, p.438, (1989) [26] C. W. Planner, G. Rees, G. Mackenzie, A Separated H- Injection System in a Modified Accumulator TRI-DN-89-K98 [27] F. W. Jones, G. H. Mackenzie and H. Schonauer, ACCSIM - A Program to Simulate the Accumulation of Intense Beams, Proc. of Int. Conf. on High Energy Accelerators, Tsukuba, 1989. [28] H. Schonauer, Addition of Transverse Space Charge to ACCSIM Code, TRI-DN-89-K50. [29] L. Criegee, A Study of Slow Extraction at the KA ON Factory, TRI-DN-89-K78. [30] C. Kost and P. Reeve, REVMOC: A Monte Carlo Beam Transport Program, TRI-DN-82-28. [31] U. Wienands and R. V. Servranckx, Proceedings of the 1"t European Accelerator Conference, Rome, Italy:, p. 269 (1988). [32] L. Criegee, A Study of the Slow Extraction at the KAON Factory, TRI-DN-89-K78. 2-55 MAGNET AND KICKER SYSTEMS Chapter 3 3 MAGNET, POWER SUPPLIES AND KICKER SYSTEMS 3-1 3.1 Magnets ....... 3-1 3.1.1 Introduction 3-1 3.1.2 AC Magnets 3-3 3.1.3 DC Magnets 3-11 3.1.4 Magnetic Measurements 3-12 3.1.5 Magnet Installation and Services 3-12 3.2 Magnet Power Supplies 3-15 3.2.1 Magnet Excitation 3-15 3.2.2 Storage Rings . . . 3-15 3.2.3 Synchrotron Rings 3-15 3.2.4 AC Makeup Power 3-19 3.2.5 DC Bypass Chokes . 3-20 3.2.6 Quadrupoles and Other Magnets 3-20 3.2.7 Control 3-22 3.3 Kicker Magnets . 3-23 3.3.1 Introduction 3-23 3.3.2 1 MHz Chopper 3-23 3.3.3 Magnetic Kickers and Pulse-Forming Networks 3-25 3.3.4 Diagnostic Kickers 3-29 3.3.5 Abort Kickers ... 3-30 1 3 MAGNET, POWER SUPPLIES AND KICKER SYSTEMS 3.1 Magnets 3.1.1 Introduction The KAON Factory will require the design, procurement and installation of about 1900 magnets with the number of each type distributed as shown in Table 3.1.1. This section deals with the accelerator ring and transfer line magnets only. Parameters for the ring magnets are listed in Table 3.3.2 and are the latest values specified. During the Project Definition Study (PDS) the parameters were updated as the lattices were improved so some values quoted for prototype designs or in the references may differ from those in the table. Magnets have been categorized as ac and dc although they will almost all be of laminated Table 3.1.1: Magnet Quantities Dipoles Quadrupoles Sextupoles OeD's Total I-A Transfer 11 40 12 63 A-Ring 24 55 24 48 151 A-B Transfer 6+2S 8 16 B-Ring (ac) 25 52 24 24 125 B-C Transfer 5+28 37 12 56 C-Ring 96 140 48 136 420 C-D Transfer 2+2S 8 12 D-Ring (ac) 96 140 48 128 412 D-E Transfer 2+6S 6 14 E-Ring 96 132 48 136 412 Switchyard and Primary BL's 31+4S 72 40 147 Secondary Beam Lines 35 87 6 128 TOTALS 445 777 198 536 1956 Quantities do not include spares Septa are indicated by +NS, where N is number needed 4 Skew quadrupoles have been included in the total for each ring All sextupoles and OeD's listed may not be needed for initial operations. construction. The design effort both for physics and engineering has concentrated on the ac magnets because these represent a new technology for TRIUMF. During this year visitors from other laboratories and from industry have spent time reviewing both our designs and our design methods and they have also contributed some of the reference designs for specific magnets. Dipoles and quadrupoles have received the most attention but all magnets have been looked at briefly to establish their size, power- requirements and costs. Preliminary considerations of the magnet designs are summarized in the proceedings of the magnet workshop held at TRIUMF[1] and a recent magnet conference[2]. 3-1 Ring A dc B 50 Hz C de D 10 Hz E dc Table 3.1.2: Magnet Parameters for the Rings DIPOLES Effective Vertical Max. Pole Length Aperture Field Width (m) (m) (T) (m) 1.00 0.092 0.882 0.24 2.99 0.108 1.118 0.28 1.00 0.070 0 .834 0.18 4.89 0.100 1.381 0.28 3.90 0.052 1.731 0.25 Quantity 24 25 96 96 96 One extra B-ring magnet is included for power supply purposes Ring Type A FandD B F D C FandD SS D FandD SSl SS2 E F D SSl SS2 Effective Ring Length (m) A F 0.2 D 0.2 B F 0.2 D 0.2 C F 0.2 D 0.2 D F 0.2 D 0.2 E F 0.2 D 0.2 Effective Ring Length (m} A 0.18 B 0.18 C 0.18 D 0.18 E 0 .18 QUADRUPOLES Effective Bore Pole Tip Length Radius Field (m) (m) (T) 0.3 0.067 0.284 0.46 0.071 0.604 0.36 0.071 0.890 0.2-0.4 0 .065 0.64 0.2 0 .050 0.52 0 .94-1.69 0 .074 0.98 1.09-1.24 0 .052 0.98 0.68-0.94 0.061 0.98 1.4-1.8 0.043 0.521 0.82 0.062 0.99 1.1-1.6 0.035 0.842 0 .7-1.3 0.049 0.942 SEXTUPOLES Bore Pole Tip Radius Field (m) (T) 0.073 0 .021 0.073 0 .015 0.073 0.079 0.073 0 .056 0.086 0.103 0 .058 0.050 0 .086 0.834 0 .058 0.407 0.070 0 .553 0 .058 0.407 ORBIT CORRECTION DIPOLES Maximum Vertical Field Aperture (T} (m} 0.01 0.12 0.028 0 .072 0.016 0.102 0.127 0.102 0 .127 0.102 3-2 Max. Gradient (Tim) 4.25 8.51 12.53 9.78 10.39 13.24 18.84 16.06 12.12 15.96 24.06 19.22 Maximum Gradient (T/m2 ) 3.937 2.760 14.89 10.44 13.93 14.94 112.8 121.0 112.9 121.0 Bend Angle (mr} 0.53 0.40 0 .22 0 .22 0 .22 Quantity 51 24 24 96 40 96 12 28 48 48 12 20 Quantity 12 12 12 12 24 12 24 12 24 12 Quantity 24 24 136 136 136 As the cost of dipoles scales rapidly with the pole gap and width, an important design consideration is to minimize these quantities. This work has not been completed as yet although an attempt has been made to optimize the design of the Driver dipole. Allowance for the beam pipe is 1 cm for the dc magnets and 2 cm for the ac magnets, which require a ceramic vacuum chamber. Quadrupole magnet apertures were estimated from 'beam stay clear' values (see Section 2.1.2). In most focusing quadrupoles the maximum horizontal beam excursion is larger than the bore radius. The racetrack rings each require ten families of quadrupoles with different parameters. The designs have been consolidated into at most four types per ring and as the parameters become finalised there will be opportunities to reduce the number of different designs further. 3.1.2 AC Magnets The Booster and Driver ring magnets have dc-biased ac excitation and are designed as ac magnets. This requires that eddy current and core losses be carefully considered and evaluated in the conceptual design of each type of magnet. It will also be important that magnets of different types in each ring, e.g. dipoles, quadrupoles and sextupoles operate with the same level of saturation so that they will track together during the acceleration cycle. The Booster magnets are the most demanding, as they operate at a repetition frequency of 50 Hz and therefore prototypes of a dipole and a quadrupole were designed and procured during the PDS year. These prototypes were designed for dual frequency excitation with a 33.3 Hz rise and a 100 Hz fall. This requirement is not now part of the design specification as the excitation is to be a simple sinusoid at 50 Hz. These magnets are described below; in both cases recent changes mean that these prototypes must be redesigned slightly to conform to the current magnet parameters listed in Table 3.3.2. The ac magnets will have a laminated steel construction and the laminations will be epoxied together to reduce vibrations and increase mechanical stiffness. The coils will be subdivided to reduce eddy current losses, and transformations to eliminate or reduce circulating eddy current components will be made at the end terminations. The steel procured for the magnet construction will be categorised by some parameter such as coercivity, which will be recorded for each batch produced. The laminations for each magnet will be selected so that a mean value of the parameter is achieved for all magnets of each type and possibly for all the magnets in anyone ring. Prototype Booster Dipole This magnet is shown in Figs. 3.1.1 and 3.1.2 and its parameters are listed in Table 3.1.3. It has a curved profile and is made from single-piece laminations[3,4]. Some design decisions were made because the magnet was a prototype and the procurement time was limited. 3-3 Figure 3.1.1: Prototype Booster dipole magnet arrangement Figure 3.1.2: Cross-section showing lamination cooling arrangement 3-4 An example of this is that the coils are made from an array of square hollow conductor rather than from a stranded indirectly cooled cable, which would have reduced the copper eddy current losses. This decision was taken because the delivery time of the conventional conductor was much shorter. The magnet steel is type M17 fully processed non-oriented electrical steel which has an insulating coating and nominal core losses of 3.9 W /kg at 60 Hz and 1.5 T. Studies have shown that the use of a more expensive oriented steel is not cost effective because the magnet profile requires that there are regions where the flux lines are perpendicular to the rolling direction[5]. The magnet has curved pole corners shaped to reduce saturation effects at different levels of excitation and it also has a shaped end profile to maintain the effective length constant at the injection and extraction energies. Table 3.1.3: Prototype Booster Magnet Parameters Physics Energy Range Field Rise Frequency Field Fall Frequency Maximum Pole Tip Field Minimum Pole Tip Field Effective Length Pole Gap Field Uniformity Good Field Width Number Required Power Supply Maximum Current Maximum Voltage/Magnet (peak) Maximum Inductance No. Magnets/PS Cell Maximum Voltage/PS Cell (peak) 450-3000 MeV 33.3 Hz 100 Hz Dipole 1.05 T 0.277 T 3.18 m 10.68 cm B/Bo ~ 1 x 10-4 ± 5.0 cm 25 5000 A 3 kV 5.75 mH 5 15 kV Quadrupole 0.70 T 0.175 T 0.8 m 13.2 cm CN/C2 < 2 X 10-3 6.6 cm 48 1600 A 0.4 kV 1 mH 24 10 kV The prototype was designed for a peak field of 1.05 T and a return yoke flux density of 1.25 T. The current specification calls for a peak field of 1.118 T and it has been shown that the higher field can be achieved with some reduction in the field uniformity [6] as indicated in Fig. 3.1.3. The reduced region of uniform field is greater than the beam size at extraction and is acceptable. 3-5 .6B x10 4 BOOSTER DIPOLE B I eMS 1 2 3 4 7 2 3 4 5 1.0ST 6 7 1 .1 aT 8 1.1ST Figure 3.1.3: Field uniformity of Booster Dipole vs. peak field The prototype magnet coils have been delivered to TRIUMF and the steel is scheduled to be completed by March 1990 with assembly and first field measurements taking place in April. Prototype Booster Quadrupole A prototype of a Booster quadrupole with parameters as specified in Table 3.1.3 was designed and is being fabricated[7]. This magnet section was based on an existing HERA design which meant that the tooling was available and the lamination costs were reduced. The magnet is shown in Fig. 3.1.4 with the quarter section and coil detail in Figs. 3.1.5 and 3.1.6. The coil for this magnet is indirectly cooled using a square stainless steel cooling tube sandwiched between rectangular magnet wire. Eight conductors are connected in parallel to form one turn and the ends are transposed to reduce circulating eddy currents at the terminations. The whole array is impregnated with polyester resin. Studies show that in order for this magnet to track the dipole, close attention must be paid to the actual steel stacking factor and to the B-H properties of the steel. Fabrication of the cores for this magnet has shown that the length changes during the curing cycle with about 4% shrinkage being recorded; this is attributed to epoxy loss during curing. This length change will be allowed for on the production magnets, and the weight of steel per unit length will be tightly specified, rather than being given as a minimum which must be attained. The coils have presented some fabrication problems and will not be delivered until March 1990. 3-6 ...-.------711.5 • • • • • • • • • • • I I r-r- If-,-I I I I- II-~ r I I 1 , I I I~ 0- I I I I I 1 1 1 T I~ I- II-I I I L.....I..- I I--'-I I • • • • • • • • • • Figure 3.1.4: Prototype Booster quadrupole magnet arrangement 30 20 10 O-+-----.----------~-L----~ o 10 20 30 PROTOTYPE BOOSTER QUAD QUARTER SECTION Figure 3.1.5: Prototype Booster quadrupole quarter section 3-7 WIER GIDlOO YIHAP .. rr PAtOKE (Room VIRAP .r- ClXlIN(; TlI£S 2 4 V 0 11011 ~ DID/DID i 2 4 \.. OlNlLClOO INSllAI ION ~ w.. TlW IRE OJNlJCTOO Figure 3.1.6: Coil cross-section for Booster quadrupole showing indirect cooling arrange-ment Driver Magnet Designs The Driver ring dipole has been studied in concept only, and no prototype has been designed as yet[8,9]. The dipole aperture is now larger than in the 1985 proposal and so it was necessary to make the design as compact as possible to keep the costs and weight from escalating to unacceptable values. The dipole quarter section is shown in Fig. 3.1.7. The pole profile was varied to determine the minimum acceptable field uniformity as given by (B / Bo - 1) at injection and extraction excitations. Figs. 3.1.8 shows the results of these calculations. Geometry 41 gives the most compact magnet with a field uniformity of 2 x 10-4 over 6.2 cm at injection and 1 X 10-4 over 3.4 cm at extraction (dimensions are half apertures). It is proposed to make the magnet from low-carbon steel laminations 1.5 mm thick, which reduces the steel cost and the amount of handling in stacking the individual laminations. The use of thicker low- carbon steel does result in higher core losses, but even with a 9% larger air gap the total losses in the current design have only increased by 10%. The magnet will be curved because this minimises the weight and the stored energy, which affect the costs of the installation equipment and the power supply. The Driver dipoles will be the major procurement contract for the KAON project and it will be essential to make a prototype prior to production magnets. These magnets will probably be the critical path item for the whole project. The ring requires quadrupoles of three different apertures. The F and D quadrupoles will have a bore radius of 7.4 cm, a pole tip field of 0.98 T and their lengths will be varied to achieve the specified strength. These magnets have their coils out of the median plane to avoid possible radiation damage from spilled beam. Reference[lO] describes the current design, for which the cross section is shown in Fig. 3.1.9. It was necessary to make the 3-8 30 20 E u -10 10 - GEOMETRY "1111 ~ GEOMETRY 1141" 20 30 X (cm) 40 Figure 3.1.7: Quarter section of Driver Dipole showing reduced size for Geometry 41 2r------r---.----,---r--..,.---r------. (XI0-4 ) 0 2 0 4 (0 ) CD ......... 6 @ INJECTION CD ( Bo = .167T ) 8 10 0 2 3 4 5 6 7 2 x(cm) (XI0- 4 ) 0 2 4 0 CD 6 ......... (b) CD 8 @ 10 (8 0 = 1.35 T) 12 0 2 3 4 5 6 7 X (em) Figure 3.1.8: Calculations of the field uniformity of the Driver dipole for various geometries 3-9 coil profile stepped, which will complicate coil fabrication. The magnet design has been shown to track with the dipole and the harmonics are within tolerance. A study with TOSCA has shown that the effective length changes by about 1 cm between injection and extraction energies for two different end profiles, each with a 2 cm bevel, so further studies are recommended before the concept is finalised[ll]. These studies were based on an early specification with a 6.8 cm bore radius so the results will have to be scaled to the current design. 1lO_2SCI J .RES J6I11 /89 ":58:50 10. 35. 30. 25. 20. PAGE 5: CONT POTE alAs-res LA9E~ ELEIMlJ.IO srm-XY SDJj-H Static: SoLutIon Mrsil SSi EL_ts i5 ReqlQnS 10.000 50 .000 PE20 Figure 3.1.9: Current design of Driver quadrupole showing field contours The straight-section quadrupoles are of two designs having bore radii of 5.2 and 6.1 cm. The pole-tip fields will be similar to those of the end section magnets and again the lengths will be adjusted to achieve the required strength. No specific design calculations have been made on these magnets. Other AC Magnets Sextupoles and OCDs (orbit correction dipoles) are specified for both the Booster and Driver rings. The pole-tip fields needed in the Driver sextupoles are high 0.8T and 0.4 T for the F and D magnets, respectively. The effective lengths are only 0.2 m, and with the bore apertures of 17 and 12 cm, calculations using TOSCA will be necessary. All sextupoles and OCDs listed in Table 3.1.1 may not be necessary for initial operation at the beam current. The OCDs will be simple frame magnets which will provide a bend angle of up to 0.5 mr. In most cases the windings will be water cooled to keep the magnets compact[12]. Some of these magnets will be designed for operation at several hundred Hz so that corrections can be applied during the accelerating pulse. Space is very limited in the ac rings so the horizontal steering will be applied by separate windings on the dipole magnets. 3-10 Skew quadrupoles have been specified for all rings; the parameters do not appear to be demanding although the length/bore ratios are less than unity, so they will require design using a 3D code. Preliminary specifications are also available for orbit bump magnets in the Driver ring. 3.1.3 DC Magnets The magnets in all transfer lines and storage rings will be excited by dc current. Some transfer line magnets will be based on the preceding ring designs and will be included in the order for the ring magnets, e.g. the B-C transfer line quadrupoles could well be the same as and be procured along with the Booster ring quadrupoles. The storage ring magnets will be made from low-carbon steel laminations, thicker than the ac laminations, and they will not be epoxied together. The thickness will depend on the punch-press capacities and lamination size. At European laboratories laminations up to 5 mm thick have been used for small magnets. They are produced by 'fine-blanking' rather than stamping. There are a number of reasons for laminating dc magnets. Each lamination is the same to high accuracy so variation between magnets of a particular type due to machining differences is eliminated. The amount of steel in each magnet can be accurately controlled by weighing the laminations so that the magnet is assembled to a weight limit rather than to a length limit. Variations in steel magnetic properties can be reduced by measuring a parameter such as coercivity and mixing steel from various batches to achieve an average value throughout all magnets of the same type. Using these techniques magnets for synchrotrons and storage rings can be made identical to within 1 part in 103 for all of their major parameters. Coils for dc magnets will be from square hollow conductor of cross-section large enough for 1000 A operation. This reduces the number of turns required and the complexity of the cooling circuits. Coils with these characteristics have been designed at TRIUMF many times in the past decade. All coils will be made from copper. Only conceptual designs for costing and sizing purposes have been made for these magnets. The Extender dipole has a field of 1.73 T and a vertical aperture of 5.2 cm, which is set by vacuum pumping considerations. This magnet has been briefly reviewed using the finite element code PE2D but the work is incomplete at this time. The required parameters can be met but more work is needed to reduce the pole width and magnet size. The A-ring and C-ring dipoles are similar in length and field values but they have different apertures. There are a number of septa required for the transfer lines[13,14] . These have been briefly reviewed and prototypes will be made for the most demanding of them. Octopoles and sextupoles in the E ring and its extraction line have recently been specified but are not as yet designed. 3-11 3.1.4 Magnetic Measurements All magnets will be measured to ensure compliance with the physics, mechanical and electrical parameters specified. It is proposed to do this at TRIUMF or in a rented facility nearby. In general Hall probe measurements will be made on the dipole magnets and Morgan coils will be used to survey the multi poles. Ac dipole magnets will be surveyed using an HP 3458A multimeter connected to an IBM 386 computer; about 20 measurements per pulse will be taken and stored for analysis[15]. It is anticipated that most measurements will be done on a comparative basis using the prototype or first production magnet as a standard. Several testing stations will be operated in parallel as the magnets will be surveyed as they are delivered, alignment fixtures will be installed, and the magnets will be removed for installation in the tunnels. A measurement rate of up to 25 per week will be necessary at the peak periods. The data collection and storage will be a major project and it is expected that several people will be required just for data management. Some preliminary measurements have been made to prove out the system and to investigate noise reduction techniques. Care is required at connection interfaces and to avoid ground loops. Using a BHT710 Hall probe with a sensitivity of 12.0 mV /kG, a signal of 60.0 mV is obtained in a 5.0 kG field. This signal is amplified 10 times by a highly stable voltage preamplifier resulting in a 600.0 m V output. The absolute accuracy of the HP 3458A is a function of the voltage range, and integration time chosen for the measurement. It has been shown to be capable of achieving the required accuracy. 3.1.5 Magnet Installation and Services Installation and Support Provision is required for supporting up to 3 rings of magnets and other accelerator compo-nents, one vertically above the other. In the present design the synchrotron rings, Booster and Driver, are on the lowest level and therefore can be supported from the tunnel floor by concrete or steel pedestals. The other rings or beam transfer lines are supported from a cantilevered structure as shown in Fig 3.1.10. Calculations have shown that a beam section of square hollow tubing 203.2 mm width and 12.7 mm wall thickness is suitable for this structure. 2-D static loading, 3-D modal analysis and 3-D seismic loading have been investigated. Deflections are less than 1 mm in the worst case under load, stresses are not serious and the frequency response of the structure appears acceptable. The dipole magnets are designed to be self supporting and will be located on 3 or 4 mounting pads with height and horizontal adjustments. Quadrupoles and other magnetic elements will be supported from a base plate. 3-12 MAIN RING TUNNEL o 1000 2000 111111111111111111111 mm 500 1500 Figure 3.1.10: Cross section of Main Ring tunnel showing magnet installation and support arrangement The installation of the magnets and other heavy components will be done using a specially designed magnet transporter. The Driver dipole magnet weighs 26 tons and is the heaviest element to be handled. The Be transfer line elements in the main ring tunnel are located 3.4 m above the tunnel floor and this sets the height requirement for the transporter. A local consulting company was given a contract to investigate the various installation techniques used at other accelerator laboratories and recommend a design for the KAON Factory[16]. They proposed two towed fork-lift type vehicles, one optimized for each tunnel and with a design based on the Fermilab transporter built by Elwell-Parker. Figure 3.1.10 shows the arrangement of the MT -30 version of this device. Operation of the magnet transporter once it is towed into position is done from a control pendant so that removal and positioning of magnets in hot areas can eventually be carried out by operators located behind local shielding. Electrical and Mechanical Services An effort has been made to standardize the currents used in the accelerator ring dipoles and quadrupoles to reduce the number of different conductors required with currents of 500 A 3-13 and 1000 A used for dc magnets, 3600 A for the Booster dipoles and 4500 A for the Driver dipoles. A study has been made of the optimum conductor configurations for providing these currents from the power supplies in the service buildings to the components. This study investigated conductor configuration and support systems, routing, voltage drop, heat load to the tunnels, radiation resistance and costs. A single conductor 1000 kcmil cable with EPR insulation emerged as the choice for the 1000 A service with different insulation jackets for the 1 k V to 15 k V voltage requirements. Similarily a single 500 kcmil cable was selected for the 500 A service. The higher current requirements are met by a water-cooled copper bus made from 75 mm diameter copper pipe on porcelain insulators insi~e the tunnel and by six parallel runs of 500 or 750 kcmil conductor in the ducts leading to the tunnel. Low-active LeW systems will be used to cool the magnets and other devices in the tunnels. These systems will be closed loop recirculating water systems with a deionization-polishing system to continuously maintain low levels of conductivity to the 10 M! level. Quick disconnect couplings will be used to connect magnets to the cooling water headers. 3-14 3.2 Magnet Power Supplies 3.2.1 Magnet Excitation The Booster and Driver ring magnets require dc-biased ac excitation with repetition rates of 50 Hz and 10 Hz, respectively. A change has been made with respect to the original proposal in that the Booster excitation is now a simple 50 Hz sinusoid. The Driver excita-tion remains dual frequency with a 6.7 Hz ramp followed by a 20 Hz reset. The magnets are powered in series by a resonant system which is described later in this section. Dipoles and quadrupoles of the same type are powered independently. The A,C and E ring magnets are powered by regulated dc supplies, also with magnets of the same type powered in series. A design study was undertaken to define the power supply system for magnet excitation for the five accelerator rings. This study concentrated on an approach which led to the standardization of power conversion equipment. 3.2.2 Storage Rings A,C and E ring dipoles and quadrupoles are designed for nominal 1000 A excitation, with the most stringent requirement existing for E ring dipoles which require 0.001% regulation. Input ac line variations almost preclude the use of thyristor based supplies as current regulators for E ring dipoles. The study showed that an approach based on high frequency link power conversion equipment was both technologically feasible and economically competitive with 12- or 24-pulse SCR equipment. Economics further dictate that as opposed to building a multiplicity of small supplies, building a smaller number of high power units results in lower cost per watt as well as reducing the requirements on the control system by reducing the number of control points. The basic building block was selected to be a 450 V dc 1000 A constant current supply regulated to 0.001%. This unit is large enough to power all A ring dipoles and is used in conjunction with series voltage sources to power the C and E ring dipoles. A full power prototype was constructed and performance has proven to be satisfactory. For quadrupoles the regulation requirement can be relaxed to 0.1 % though the basic building block could remain the same without adversely effecting cost due to the overall quantities involved. 3.2.3 Synchrotron Rings Typical Booster and Driver dipole current waveforms are shown in Fig. 3.2.1. To minimize ac input line disturbances resulting from these loads, resonant magnet excitation has been 3-15 I max 4500A I Bias 2850A J min 1200A ~ toms ~( tOms ~ I I I '< 20ms )' (a) BOOSTER DIPOLE CURRENT CYCLE (50Hz) I max 5553A 1 Bios 3110A I min 667A Acceleration I Reset t ~--------~--~--~ I I 75ms k25ms I ), ), (b) DRIVER DIPOLE CURRENT CYCLE (10Hz) Figure 3.2.1: Typical Booster and Driver dipole current waveforms 3-16 selected as proposed by J. A. Fox[17] and successfully employed at a number of labora-tories. The dual frequency D ring excitation is as proposed by Walter Praeg[18]. During acceleration, the low frequency component results from the parallel capacitance of Cl and C2 resonating with the effective cell inductance. The high frequency reset component is achieved by the switching out of 8/9 of the resonant capacitance Cl once per cycle via SI during the reset interval, resulting in a rise to fall ratio of 3 to 1. This mode of operation has been verified using NINA dipole magnets in our magnet test stand.[19] A typical Booster resonant ring configuration is shown in Fig. 3.2.2 with the dc bias inserted in a modified resonant cell to minimize the ac current component flowing in the C ti-- res Cres 1 VIRTUAL GND Figure 3.2.2: Arrangement for resonant powering of the Booster dipoles bias supply. Only the centre point of the bias supply is grounded with virtual grounds occurring at the centre of the dc bypass chokes as well as at the cell boundaries. This serves to minimize the maximum voltage to ground around the ring. Field parameters are dictated by the physics requirement for these rings and translate to ampere-turns for the magnets. System complexity and reliability is a function of total device count and stress level. An optimization of these various parameters shows that there are certain key decisions which define system parameters. A predominant factor is the peak voltage to ground within the resonant system which is determined by the maximum di/dt of the inductive components. It was felt that a maximum value for this is 15 kV peak. A second factor is the maximum current which is achievable without having to parallel thyristor devices in the capacitor disconnect switch. This current is 8/9 of the sum of the ac peak current flowing in the magnets and the dc bypass chokes. These parameters effectively establish the desired total magnet and dc bypass choke inductance per resonant cell and with the number of magnets required yield the minimum number of cells. The resonant cell parameters for the Booster and Driver rings are shown in Table 3.3.1. 3-17 Table 3.2.1: Resonant Circuit Parameters B D Ring Energy .45-3 Ge V, 50 Bz 3-30 GeV, 10 Bz Dipoles 1. No.of resonant cells 5 10 2. No.of magnets/cell 5 10 3. Magnet inductance/cell 25mB 90mB 4. Choke inductance /cell 25mB 150 mB 5. Capacitance fixed/cell 680 uF 704 uF 6. Capacitance switched/cell 5600 uF 7. DC bias current 2850 A 3110 A 8. Peak magnet current 4500 A 5553 A 9. Peak choke current 4500 A 4576 A 10. Peak switch current 2170 A 11. Peak voltage to ground 7.8 kV 14 kV 12. Magnet peak stored energy 253 kJ/cell 1.39 MJ / cell 13. Choke peak stored energy 253 kJ/cell 1.57 MJ/cell 14. Cap. fixed stored energy 83 kJ/cell 276 kJ/cell 15. Cap. switched stored energy 244 kJ/cell Quadrupoles Focusing 1. No.of resonant cells 1 2 2. No.of magnets/cell 25 24 3. Magnet inductance/cell 100 mB 278 mB 4. Choke inductance/cell 100 mB 278mB 5. Capacitance fixed/cell 202.8 uF 413 uF 6. Capacitance switched/cell 3295 uF 7. DC bias current 1000 A 1000 A 8. Peak magnet current 1600 A 1718 A 9. Peak choke current 1600 A 1718 A 10. Peak switch current 1276 A 11. Peak voltage to ground 9.41kV 12.6 kV 12. Magnet peak stored energy 128 kJ/cell 410 kJ/cell 13. Choke peak stored energy 128 kJ/cell 410 kJ/cell 14. Cap. fixed stored energy 8.96 kJ/cell 130 kJ/cell 15. Cap. switched stored energy 115 kJ/cell Quadrupoles Defocusing 1. No.of resonant cells 1 2 2. No.of magnets/cell 25 2 3. Magnet inductance/cell 100 mB 172.5 mB 4. Choke inductance /cell 100 mB 172.5 mB 5. Capacitance fixed/cell 202.8 uF 738.5 uF 6. Capacitance switched/cell 5,908 uF 7. DC bias current 1000 A 1000 A 8. Peak magnet current 1600 A 1768 A 9. Peak choke current 1600 A 1768 A 10. Peak switch current 1365 A 11. Peak voltage to ground 9.41kV 8.3 kV 12. Magnet peak stored energy 128 kJ/cell 270 kJ/cell 13. Choke peak stored energy 128 kJ/cell 270 kJ/cell 14. Cap. fixed stored energy 8.96 kJ/cell 102 kJ/cell 15. Cap. switched stored energy 91 kJ/cell 3-18 The dc bias power supply for the Booster dipoles was selected to be a 12 pulse SeR supply rated at 600 V dc, 3500 A regulated to 0.01 %. An 80 V dc, 3000 A prototype was built and tested successfully in a test stand for the Booster dipole prototype. The same supply has been selected as the current regulating element for the Driver dipoles, acting in series with a number of voltage sources. 3.2.4 AC Makeup Power The ac energy loss in the resonant systems is made up via distributed pulse forming networks which provide a power pulse in the centre of the acceleration interval via primary windings of the dc bypass chokes as shown in Fig. 3.3.3. Typical waveforms are illustrated in Fig. 3.2.4. This method was proposed by J .A. Fox[17} and was experimentally verified on the magnet test stand. The charging time constant of the pulsed source is chosen to minimize the pulsed power effects on the ac side of the power supply. DC BUS + I I L F/2 I I L ___________ .J PFN Lm 2 Lm 2 Figure 3.2.3: A typical Driver resonant cell with pulse forming network 3-19 (.) (b) (c) (d) Figure 3.2.4: Waveforms during PFN charging 3.2.5 DC Bypass Chokes Investigations were carried out to determine the best form of dc bypass choke. The ideal approach would be to follow the NINA and DESY models with large combined chokes to approach a unity coupling coefficient between cells. This proved to be impractical both from a complexity point of view as well as the resulting physical size of the required chokes. Various competing designs were developed and they indicated that a unit choke for each Driver cell would weigh 90 tons which approaches the anticipated handling limits [20,21]. It was therefore decided to take the separate choke approach for both the Booster and Driver. Primary windings are connected in parallel to the pulse forming network for good electrical coupling. The separate choke approach provides the additional advantage of economIC spares. 3.2.6 Quadrupoles and Other Magnets Focusing and defocusing quadrupoles each have their own independent resonant networks which are slaved to the dipole resonant system which acts as the master. Tracking of the three systems is achieved by the control of the pulse forming networks of the quadrupoles. By varying the timing of these pulses as required, it is possible to achieve phase lock to the master. By independently varying the energy transferred per pulse independent amplitude control is obtained. Straight section quadrupoles, sextupoles and OeD magnets will be driven from programmable supplies. 3-20 Table 3.2.2 sununarizes the power supply requirements for the accelerator ring and transfer line magnets. Table 3.2.2: Power Supplies for Accelerator Magnets Magnet Type Rating Quantity Regulation Volts Amps % OCD 60 12 48 0.1 60 200 184 0.1 100 200 264 0.1 Transfer line 20 500 30 0.1 40 500 4 0.1 150 500 2 0.1 20 1000 36 0.01 A & C quads 30 1000 4 0.01 40 1000 13 0.01 50 1000 6 0.01 75 1000 4 0.01 150 1000 7 0.01 E quads 350 1000 5 0.01 A,C,E dipoles 450 1000 22 0.001 B & D quads 450 1000 18 0.001 B & D dipoles 600 3500 9 0.1 PFN 2500 80 4 0.1 2000 500 1 0.1 2500 3200 1 0.1 Septum* 6.4 4700 6 0.01 1000 100 4 0.01 25 2600 2 0.01 47 2600 2 0.01 10 4400 1 0.01 10 5000 2 0.01 33 9550 2 0.1 6.5 8375 2 0.01 13.2 5000 2 0.01 682 * Does not include voltage drop in bus. Some septum magnets will be pulsed 3-21 3.2.7 Control All magnet power supplies are to have their own dedicated controller which communicates to the central control system via coaxial cable. The controller contains all of the digital and analog circuitry required for either local or remote operation of the power supply. These units will be based on existing G64 based controllers currently used at CERN where they have been in successful service for a number of years. 3-22 3.3 Kicker Magnets 3.3.1 Introduction Kickers magnets will be required for all ring-to-ring transfers in the 5 accelerator rings in the KAON Factory. Table 3.3.1 indicates the parameters of these kicker magnets. The circulating beam will be present in a series of bunches in the Accumulator ring (A ring) at the synchrotron rf frequency of 46 MHz. The field in the kicker magnets must rise from 1 % to 99% of full strength during the time interval of gaps created in the beam so that the beam can be extracted fully with minimum losses. The uniformity and stability of the kick strength are required to satisfy ± 1% accuracy criteria so that the beam emittance is not increased significantly by the kicker magnet field. By suppressing 5 bunches in the injection line a 1 MHz chopper produces a gap of 108 ns in the beam in the A ring, to allow sufficient time for the field in the magnetic kickers to be established. The total time period for the 40 bunches in the A ring is 867 ns. Thus the A-ring extraction kicker must have a rise time of less than 108 ns and a flat top of 867 ns. The 5-bunch gap length and the total time period for the 40 b~am bunches will shorten as the beam is accelerated, with the gap for the D-ring extraction kicker field rise-time, at 30 Ge V, approximately 80 ns. The C ring will collect beam serially from 5 complete B acceleration cycles with the dura-tion of the C injection kicker flat-top 659 ns and the C extraction kicker a factor 5 longer. Since the E ring is longer than the D ring by 5 beam bunches the rise time of the E injection kick will be 10 beam bunches long, as opposed to the 5 beam bunches in the other rings. In general the rise time for injection kickers can be longer than the required fall time and similarly the fall time of extraction kickers can be longer than the rise-time. The following sections describe these devices in more detail. 3.3.2 1 MHz Chopper The extracted beam from the cyclotron consists of 100 p.A of 452 MeV (1.026 GeV Ic) H-ions with a repetition rate of approximately 23 MHz and a bunch width of about 5 ns. The Accumulator ring is capable of storing 45 bunches at 46 MHz with adjacent bunches filled in successive revolutions. This would leave only about 10 ns between bunches, which ' is insufficient time for the field in a magnetic kicker to be established. A 1 MHz beam chopper will be installed in the cyclotron to A ring beam line (IA line) to deflect selected beam bunches in the H- particle beam to create a 5-bunch gap. These bunches will be directed to a stripper foil and separated from the H- beam in a downstream dipole magnet and diverted into a 12.5 p.A beam dump. 3-23 Table 3.3.1: Injection and Extraction Kicker Magnet Parameters Kicker A A B B C C D D D E E Location Extr 1 Extr 2 Inj Extr Inj Extr Inj Extr Fast Inj Abort Momentum (GeV/c) 1.026 1.026 1.026 3.825 3.825 3.825 3.825 30.92 30.92 30.92 30.92 Kick angle (mrad) 1.76 8.44 10.0 7.7 4.0 4.0 4.0 2.5 2.5 2.0 2.0 Horizontal [H] or Vertical tV] V V V V V V V H H H H x Aperture (mm) 79 115 146 135 76 76 86 37 37 56 56 y Aperture (mm) 73 73 81 70 68 68 76 24 24 32 32 Frequency (Hz) 50 50 50 50 50 10 10 10 10 10 10 Flat Top (ns) 867 867 867 659 659 3625 3625 3522 3522 3522 3522 Rise Time (ns) 108 108 10ms 82 82 82 25ms 80 80 lOps 160 Fall Time (ns) 1 ms 1 ms 108 10ms 82 1ms 82 25ms 25ms 160 lOps Fill Time maximum (ns) 79 79 79 53 53 53 53 51 51 132 132 Fill Time actual (ns) 49 78.2 64.6 47.3 53 53 53 51 51 132 132 'LVPFN (kV) 54.0 54.0 86.0 291.0 131.1 131.1 146.4 374.5 374.5 175.2 175.2 Number Modules 1 1 2 5 3 3 3 7 7 4 4 VPFN (kV) 54.0 54.0 43 58.2 43 .7 43.7 48.8 53.5 53.5 43.8 43.8 Pulse Current (A) 1080 1080 1720 4656 874 874 976 1070 1070 876 876 Flux Density in Aperture.(mT) 17.2 11.8 14.8 43.3 14.5 14.5 14.3 56.0 56.0 35.5 35.5 Characteristic s/c Impedance (n) 25 25 12.5 12.5 25 25 25 25 25 25 25 Actual Magnetic Length (m) 0.35 2.45 2.314 2.27 3.535 3.535 3.579 4.607 4.607 6.0 6.0 Insertion Length (m) 0.9 3.0 2.904 2.904 4.165 4.165 4.209 5.397 5.397 6.67 6.67 Available Length (m) 0.9 3.0 2.904 2.904 7.77 7.77 7.147 7.0 7.0 11.9 11.9 3-24 The first 20 beam bunches from the cyclotron, numbered 1 to 20, (see Fig. 3.3.1) will not be deflected by the chopper but will be transferred to the A ring. Bunches numbered 21, 22 and 23 will be deflected using an electric field established between a set of deflection plates with kick-strength rise and fall times of less than 39 ns (fill time and pulse rise time combined) and a flat top duration of more than 92 ns. Then 20 more bunches, numbered 24 to 43, will be allowed to pass into the A ring. Bunches 44 and 45 will be deflected by the chopper, using an electric field with a flat-top duration of more than 49 ns. The process starts allover again at bunch number 46 which is stored along with bunch number 1. The resulting beam in the A ring will have 5 beam bunches missing, which will increase the normal single pulse gap by 5 times the pulse period to approximately 108 ns. The required deflection angle of 1 mrad at 452 MeV, for the 5 bunches to be removed, will be achieved with a set of plates 2.5 meters long and 50 mm apart with a voltage difference of 15 k V. A novel design concept was developed for providing the required voltage pulse at the plates[22] which minimizes power dissipation in the driving circuits. The concept involves the storage of electrical pulses in a very low loss 50 n transmission line that has a one way propagation delay of approximately 1 I1S and is connected to the deflection plates at one end. The deflection plates will be centre fed to reduce their fill time[23] and this requires that their impedance be 100 n to match the impedance of the 50 n storage cable. The plates will be open circuit at each end. A schematic diagram of the beam chopper is shown in Fig. 3.3.2. At the opposite end of the storage cable to the deflection plates there will be two tetrodes. One of the tetrodes will be used as part of the charging circuit (charger) and will serve to reshape the leading edge and to refill the flat-top portion of the pulse. The other tetrode circuit (clipper) will be used to reshape the falling edge of the pulse. Reshaping is necessary to clip the effects of dispersion [24] , due to losses in the storage cable, and unavoidable stray capacitance components[25]. 3.3.3 Magnetic Kickers and Pulse-Forming Networks The design of the pulse generator proposed for the injection and extraction kickers will be based on that of the CERN PS division[26,27], which uses a quasi-resonant power supply that can charge a pulse-forming network (PFN) up to 80 kV a few ms before the thyratron switches are triggered. Quasi-resonant power supplies are chosen as they can be used to minimize the time duration of the charge on the PFN, and the corresponding voltage across the thyratrons, to maximize switch dIj dt at turn-on while avoiding excessive faulty shots (i.e. spontaneous turn-ons) of the thyratron switches. In addition the duration of the high-voltage stress on the PFN is minimized. Three stage deuterium-filled thyratrons will be used for the fast switching and will be mounted in impedance-matched oil-filled coaxial housings to minimize reflections. 3-25 I" -I 39n5 92n5 39n5 39n5 49n5 39n5 KICK STRENGTH 1 1 1 1 ! ! ----{ f----1 1 I 1 1 1 1 I: / BEAM BURST ""- 1 : 1 1 1 NUMBER '" l J20' 11 I 23~ '[24 ~15 ~2 431 ._-I.~_---K.._ - ....... -'----. J L.. - -1 f- _.J L '-----A.._....D...----..I 3 DEFLECTED BEAM BURSTS 2 DEFLECTED BEAM BURSTS Figure 3.3.1: Timing of deflector voltage pulses to alternately deflect 2 and 3 beam pulses. 15 ~. -'o n 3.ON LONG X .78M OIA. VACUUM ~NISTER. 100 OHM S'TRIPUNE 2.~M LONG. 1 MICROSEC LONG (3OOM)--1 50 OHw LJJW COSS CASI£ ( • • 9. 12.7 CM DIAWElER) I<ICK RISE AHD FAll.. JInS (IoWCIMUM) ALl9NAlE FlATTOP 48nS AND 92nS (MINIMUM) + 18.2l<V -rCHARGER Figure 3.3.2: Schematic diagram of 1 MHz chopper. 3-26 Prototype studies have been completed at TRIUMF with a pulse generator and PFN that were borrowed from the CERN PS division. The quasi-resonant power supply and the thyratron grid pulser were upgraded to permit operation at 50 Hz at a PFN voltage of 80 k V and pulse durations of 700 ns. This design is similar to the design requirements for the A extraction kicker and the C injection kicker. Fig. 3.3.3 shows a typical pulse from this generator when operating at 50 Hz into a resistive load. The current rise-time (5% to 95%) is about 30 ns. It can be seen in the figure that there are two small current pulses which occur during a period of 80 ns before the main pulse. These are caused by displacement current flow in the thyratron, as each gap turns on consecutively. Methods to minimize the effect of this displacement current will be the subject of future engineering studies. The 30 u :·1 I' • .1 i ' , \i.,:\ kV Ilif' id .1\.. o -~..I"!'" .. .. ...... . -IOkV--. Figure 3.3.3: Operation of CERN pulse generator at 50 Hz into a resistive load. injection and extraction kicker magnets will be of the transmission line type, based on the design of the CERN PS division[28,29]. These transmission line kickers consist of ferrite C-core sections sandwiched between high-voltage capacitance plates: each of these ferrite C cores, together with its ground and high voltage capacitance plates, is termed a cell. The magnetic field rise time of a transmission kicker magnet results from a superposition of the the rise time of the pulse from the pulse generator and the propagation time of the pulse through the magnet (fill time)[30]. 3-27 High kick strength and short rise-time are conflicting requirements for a magnetic kicker. Computer code PSpice[31] has been utilized to study the transient response of several of the proposed kicker magnets in some detail[32,33]. The predicted time response for a 25 n terminated Accumulator extraction kicker magnet, represented as being connected in a realistic electrical circuit, is shown in Fig. 3.3.4. In order to achieve high reliability it 105~------ ~------~-------~------~---- --~------~------~------~------~ , , , , 100 .;. ' .... a98.3% , , , . + : I 90 ~--- --- . ------~-- - - ---~- -----~- -- -- --~--- - ---~--- - ---+-- ---- -.;-- ------o .3us 0 . 4us 0 . Sus 0 .6us 0 . 7us 0 . Sus 0 . gus 1 .Ous 1. ius 1. 2us Tlme Figure 3.3.4: Normalized flat-top kick strength for a 24-cell module of the Accumulator extraction kicker is recommended that the magnetic kickers for the KAON Factory should be designed to operate with a 'conservative' PFN voltage of about 50 kV to 60 kV. The requirement for a short rise time can be satisfied by sub-dividing a kicker magnet into several modules, each with its own pulse generator. A large kick angle is achieved by installing the appropriate number of independent modules through which the beam passes sequentially. In the A, C, D, and E rings the kicker magnet impedance will be 25 n. Space limitations in the Booster ring lattice do not permit a 25 n kicker impedance. The Booster injection kicker will be a terminated 12.5 n magnet, so as to double the kick strength per unit length, for a given voltage, relative to a similar 25 n magnet. The Booster extraction kicker will be a short-circuited 12.5 n magnet, so as to increase the current 4 fold, relative to a similar 25 n terminated kicker. Further work is required to 3-28 identify the effect of short-circuiting the kicker magnet upon beam instabilities. In the Driver synchrotron the beam size is reduced as the energy is increased. It is pos-sible to reduce the number of modules, resulting in a considerable cost saving, by taking advantage of the reduced beam emittance for the extraction kickers. This will be done by installing the extraction kicker aperture outside the normal acceptance of the beam dur-ing the injection and acceleration phase. A bump magnet will be used during extraction to move the beam over into the kicker aperture. The extraction kicker magnet will thus require an open aperture with the ground conductor placed so as not to interfere with the bumped beam. All of the other kickers will have a ground conductor closing the aper-ture, which has the effect of both improving the magnetic field uniformity and minimizing magnet fill-time. Results of magnetic field mapping, performed for two different kicker magnets borrowed from CERN, were in good general agreement with predictions obtained using computer software PE2D[34]. 3.3.4 Diagnostic Kickers -In addition to the injection and extraction kickers there is also a requirement for a vertical and a horizontal diagnostic kicker in each of the 5 rings. The rise and fall times of the diagnostic kickers do not have to be as short as the times required for the injection and extraction kickers; in addition the deflection angles are not as large. The parameters for the diagnostic kickers are shown in Table 3.3.2. The diagnostic kicker pulse generators will operate at a lower PFN voltage than the injection and extraction kicker magnets. A version of the DESY design[35] will be used for the diagnostic kickers: these are lumped-inductance magnets rather than transmission-line magnets. A single stage thyratron is mounted in air with no special provisions to reduce the stray inductance. The PFN impedance will be 25 n for the A, B, and C rings and 12.5 n for the D and E rings. 3-29 Table 3.3.2: Diagnostic Kicker Magnet Parameters II Kicker Location II Momentum (GeV Ic) 1.026 3.825 3.825 30.92 30.92 Kick angle (mrad) 0.2 0.3 0.2 0.2 0.1 x Aperture (mm) 70 80 77 86 49 y Aperture (mm) 70 80 77 86 49 Rise and Fall Time (ns) 200 200 200 200 200 PFN Volts (kV) 13 18 18 25 25 Impedance (ohms) 25 25 25 12.5 12.5 Magnetic Length (m) 0.05 0.35 0.23 0.7 0.2 Insertion Length (m) 0.45 0.75 0.63 1.1 0.6 Number 1 Vertical 1 Vertical 1 Vertical 1 Vertical 1 Vertical Modules 1 Horizontal 1 Horizontal 1 Horizontal 1 Horizontal 1 Horizontal 3.3.5 Abort Kickers An abort kicker is required for the E ring to empty the circulating beam remaining after the slow extraction process. The E-ring abort kicker will be a duplicate of the E-ring injection kicker with a 160 ns rise time, even though there will be no beam gap during the abort period. Kicker magnets will not be used in any of the rings to provide a fast abort during a fault condition. Such kickers would have to be ready to operate at any time and it is anticipated that they could actually cause more spilled beam, due to spontaneous triggering of the thyratrons. Instead pulsed magnets are used in the transfer lines to direct beam to a beam dump in the event of a failure occurring in the next ring. 3-30 References [1] A.J. Otter and A. Strathdee, editors, Proceedings of the KAON PDS Magnet Design Workshop, Vancouver, October 1988, TRI-89-1. [2] A.J. Otter et ai, Magnet Designs for the TRIUMF KAON Factory Proposal, Proc. 11th Magnet Technology Conf. Tsukuba 1989 [3] A.J. Otter, A Prototype for the Booster Dipole TRI-DN-89-K18. [4] P. Schwandt, An Improved Magnetic Design for the Booster Ring Dipoles, TRI-DN-89-K32. [5] P. Schwandt, Comparison of Realistic Core Losses in the Booster Ring Dipole Magnets for Grain-Oriented and Ordinary Lamination Steels, TRI-DN-89-K31. [6] M. Harold, Increasing the Field in the Booster Dipole TRI-DN-89-K88. [7] P.A. Reeve, Preliminary Design of Prototype Booster Quadrupole, TRI-DN-89-K34. [8] P. Schwandt, Magnetic Design of the Driver Ring Dipole - a Second Look, TRI-DN-89-K37. [9] P. Schwandt, A New Magnetic Design of the Driver Ring Dipoles, TRI-DN-89-K73. [10] David Orrell, Designs for a Quadrupole in the Driver Ring, TRI-DN-89-K80. [11] D. Orrell, TOSCA Modelling of Driver Quadrupole, TRI-DN-89-K99 [12] A. J. Otter, A Conceptual Design for a D-ring Orbit Correction Dipole, TRI-DN-89-K57. [13] M. Harold, Summary of Extraction Channel from the DRing, TRI-DN-89-K89. [14] M. Harold, Extraction Channel from the Booster, TRI-DN-89-K90. [15] C. Haddock, P.A. Reeve, D. Evans and D. Livesey, Prototype Booster Magnet Mea-surements, TRI-DN-89-K60. [16] J. Canova et al., A Magnet Mover for a Proposed KAON Factory at TRIUMF, RSI-968 September 1989. [17] J. A. Fox, Resonant Magnet Network and Power Supply for the 4 Ge V Electron syn-chrotron NINA, Proc. IEEE. 112, 1107 (1965). [18] W. F. Praeg, Dual Frequency Ring Magnet Power Supply with Flat-Bottom, IEEE Trans. Nucl. Sci. NS-30, 4 (1983). [19] K. Reiniger, The Generation of a Reference Design For TRIUMF KA ON Factory Booster Magnet Excitation, Proc. of 1989 Particle Accelerator Conference, Chicago, (1989). 3-31 [20] H. Sasaki, Design of the Energy-Storage Choke for the Booster Ring Magnet System II, TRI-DN-89-K47. [21] D. Pavlic, A Review of Energy Storage Chokes for Rapid Cycling Synchrotrons, TRI-DN-89-K39. [22] Wait G. D., Barnes M. J., C. B. Figley A 1 MHz Beam Chopper for the KA ON Factory, TRI-DN-89-K70. [23] D. Fiander, G. Wait, M. Barnes Advantages of Center Feeding the Deflector Plates for the 1 MHz Chopper System, TRI-DN-89-K45. [24] M. Barnes, W. Roberts, G. D. Wait, Attenuation and Dispersion of Pulses in Low Loss Coaxial Lines. TRI-DN-89-K69. [25] M. J. Barnes, G. D. Wait, Influence of Stray Capacitance and Delay of Storage Line Upon the Performance of the 1MHz Chopper. TRI-DN-89-K76. [26] P. Pearce, D. Fiander, A Proposal for a Pulse Generator for Bunch by Bunch Transfer from the C.P.S. to the S.P.S .. MPS/SR/Note 71-37. [27] D. Fiander , P. Pearce, FAK Pulsed Resonant Power Supply Including Voltage Mea-surement and Servo Control. MPS/ AE/Note 74-25. [28] P. Faugeras, The Full Aperture Kicker Magnet for the CPS. Part 1: Magnet Design and Computation of its Performance. CERN/MPS/SR 71-6. [29] K. Metzmacher, L. Sermeus, AC Injection Kicker for AA Complex. PS/BT/Note 87-9. [30] M. J. Barnes, G. D. Wait, Kicker Magnet Fill-Time and Parameters. TRI-DN-89-K86. [31] Microsim Corporation, PSpice version 4.01 released January 1989. [32] M. J. Barnes, G. D. Wait, Analysis of the Transient Response of Magnetic Kickers for the KA ON Factory. TRI-DN-89-K75. [33] M. J. Barnes, G. D. Wait, Effect of Thyratron Displacement Current Upon Kick Rise-Time in Magnetic Kickers. TRI-DN-89-K85. [34] H. J. Tran, Magnetic Field Measurements and Calculations of Kicker Magnets. UBC Technical Report (1989). [35] J. Riimmler, New Stripekicker in the Injection Chain of HERA. Proceedings of the KAON PDS Magnet Design Workshop, Vancouver, October 3-5, 1988. pp 80-86. TRI-89-1. 3-32 ACCELERATING SYSTEM Chapter 4 4 ACCELERATING SYSTEM 4-1 4.1 RF Voltage Program 4-1 4.2 Beam Loading ... 4-5 4.2.1 Introduction 4-5 4.2.2 Stability of Beam-Cavity Interactions 4-7 4.2.3 Transient Beam Loading. 4-7 4.2.4 Compensation Teclmiques 4-8 4.2.5 Compensation Systems for Each Ring 4-13 4.2.6 Summary and Conclusion 4-16 4.3 Radiofrequency Systems 4-19 4.3.1 Introduction 4-19 4.3.2 Booster Cavity 4-19 4.3.3 Driver and Storage Ring Cavities 4-22 4.3.4 Power Amplifiers . . . . . . 4-23 4.3.5 Higher Order Cavity Modes 4-25 4.3.6 Summary of RF System Requirements 4-27 4.3.7 Conclusion ............... 4-29 4 ACCELERATING SYSTEM 4.1 RF Voltage Program The rf system in a synchrotron serves two purposes. It is of course primarily to accelerate the beam, but also serves to bunch and stabilize it in longitudinal phase space. The requirements of acceleration determine only the product V sin ~s, where V is the rf voltage per turn summed over all the cavities and ~s is the synchronous phase. Since V sin ~s is proportional to the rate of change of the dipole magnetic field, which in rapid cycling synchrotrons varies as a cosine with time, V sin ~s varies as the sine: it starts from zero, rises to a smooth maximum at mid-cycle and drops to zero again at the end of the acceleration period. The balance between V and ~s is set by the additional requirements of providing sufficient bucket area and limiting rf costs. The bucket area increases as v17 but decreases to zero as ~s increases from 0 to 7r /2. We have required the bucket height to be at least 25% larger than the beam's momentum spread. This keeps the beam sufficiently far from the bucket's rim (the separatrix) that it is not overly sensitive to rf perturbations such as noise in voltage or phase. There are, however, compelling arguments for keeping the bucket area as small as possible. These are as follows. (1) For a given V sin ~s, a smaller bucket area requires a smaller V allowing fewer rf cavities. (2) The longer the bunch, the lower is the beam charge density. In rings A and C and at the beginning of the acceleration cycles in Band D, we would like the bucket area to be just larger than the longitudinal emittance, E1, for then the bunching factor is largest and the space charge tune shift is reduced. (3) A lower density is beneficial in terms of both longitudinal and transverse collective instabilities. (4) Since the longitudinal focusing force is sinusoidal rather than linear, a longer bunch gives a greater spread in longitudinal oscillation frequencies thus making the beam less coherent and therefore less sensitive to longitudinal instabilities. In" the Accumulator, no acceleration is required and so ~s must be zero. There are economic advantages to limiting V and since the beam coming from the cyclotron has a very small longitudinal emittance (El ~ 0.002 eV-s), one can consider using a very small rf voltage. There are two reasons why one cannot do this. First, for the Accumulator to make use of longitudinal phase space in stacking, it is necessary that the longitudinal acceptance be very large compared with the longitudinal emittance of the cyclotron beam. Second, the circulating current in the Accumulator builds up to 2.1 A and such a large current itself induces a large voltage in the rf cavities. The desired gap voltage should be at least comparable with the induced voltage or the stability of the rf system is lost. Using these considerations, we have found an optimum for €l at about 0.048 eV-s. This requires a bucket area of 0.10 eV-s and an rf voltage of 510 kV. 4-1 .8 .7 .6 60 MHz .5 .4 .3 .2 .1 .0 O. 2. 4. 6. 8. 10. t (msec) Figure 4.1: RF parameters during the Booster acceleration cycle. At the start of the acceleration cycle in the Booster, <P8 = 0 and V = 590 kV (not quite the same as in the Accumulator because ,t is different). Over the first 10% of the ramp (see Fig. 4.1), the rf voltage program is designed to avoid increasing either the transverse tune shift (due to a too short bunch) or the synchrotron tune. This is to minimize the number of betatron and synchro-betatron resonances which are crossed by the beam. From there on, the important constraint becomes that of keeping the ratio of bunch height to bucket height no larger than 0.79. Towards the end of the acceleration cycle, however, maintaining this ratio constant would require a drastic reduction in V . This is because for a given <P8' bucket area changes as jV,IT] rv v',3V (because in our case transition is well above top energy). Hence, at the end of the Booster cycle where again <Ps = 0, we would need a voltage of V = (,inj/,ext)3Vinj = 26 kV. This is much too small compared with the voltage induced in the cavities by the beam. A more comfortable lower limit is 360 kV. For this choice, the bucket area will grow by a factor of 4 near the end of the cycle (see Fig. 4.1). The basic requirement in the Collector is for a constant rf voltage of 1.8 MV, 5 times that in the Booster at extraction, the factor of 5 coming from the ratio of their radii. In the Driver, the Keil-Schnell criterion for avoiding fast longitudinal instability is violated 4-2 unless the beam's longitudinal emittance is increased by at least a factor of 4 (for impedance divided by mode number of ZII/n =8 Q). It is convenient to do this in the Collector before the beam is transferred to the Driver. The C ring receives a batch of bunches from the B ring once every 20 ms. After 5 batches are accepted they are transferred to the Driver. There is a wait time of 20 ms before the transfer during which all 5 batches circulate concurrently, and this time can be used for controlled emittance dilution. The basic technique is to perturb the total rf voltage by adding an auxiliary high-frequency cavity so as to mimic phase and amplitude noise. The method has been in use at the CERN PS for several years[l] and at the Brookhaven AGS[2]. Tracking studies have demonstrated that the necessary 4-fold increase in tOl can be achieved in 20 ms in the Collector with two 850 MHz cells (=14 times the fundamental rffrequency) run at 140 kV each and modulated at 4 times the synchrotron frequency (see Section 5.2.1). .8 .7 .6 .5 .4 .3 .2 .1 .0 o. 15. ,,6p/p (%) " "" ---30. 45. t (msec) f rf -60 MHz ~--r30 GeV 60. 75. Figure 4.2: RF parameters during the Driver acceleration cycle. The rf voltage required in the Driver starts at 1.8 MV and rises to a peak of 2.55 MV, while <Ps reaches a peak of 53° (see Fig. 4.2). At the start of the cycle, this program keeps the transverse tune shift and the synchrotron tune from going thru maxima and thereafter 4-3 keeps the ratio of bunch height to bucket height from exceeding 0.81. Towards the end of the cycle, there is again the problem of not being able, for reasons of beam loading, to bring V down to a level which would keep the bucket area constant. A reasonable lower liinit for V is 850 kV. In the Extender, the nominal rfvoltage is given by the need to match to the Driver, namely, 850 kV. However, from the point of view of the experimenters, it is very desirable to have a controllable bunch length; from as short as possible (e.g. for coincidence experiments) to debunched beam (e.g. for rare decay experiments). In principle, the bunch length in this ring can be controlled by varying the rf voltage. However, for shortening the bunches, this technique is not very useful since the dependence is weak; a twice larger voltage shortens the bunch by only 16 %. A better technique is to tune the trim quadrupoles to lower It to the point of instability. The minimum bunch length at a given intensity will depend upon the longitudinal impedance of the machine (see Section 5.1.1). To operate in debunched mode, the rf voltage cannot simply be turned down because the momentum spread will become too small and a microwave instability will occur. However, in this mode, it is possible to operate above transition. With It = 10, debunched operation is stable. The rf program will be to run at 300 kV until the bunches have rotated 90° in longitudinal phase space (0.5 ms), and then to bring the rf voltage to zero while retuning the rf cavities to resonate at a frequency midway between revolution harmonics. 4-4 4.2 Beam Loading 4.2.1 Introduction The principal task is to maintain synchronism between the beam and the rf-system. The secondary task is to tailor the beam bunches. To do this we must independently control the peak accelerating voltage V (t) (summed about the ring) and the phase of the bunch centre <Pb(t) with respect to the rf-system carrier frequency. However, the amplitude and phase controllers are not independent from one another because of beam loading. This impairs our ability to follow a standard voltage and phase program and to compensate for beam variations. The cross-coupling is, in fact, due to two effects : geometrical and dynamical. image current V gap _ -e:- voltage ---¢v angle Fig.4.2.1 : Phasor diagram, showing current vectors and accelerating voltage Geometric Coupling To the accelerating cavity, the generator current and the beam image current are indistin-guishable - and so the 'drive' signal is their phasor sum: (1) as shown in Fig. 4.2.1. IT and <PT are the total current modulus and phase angle respectively. Likewise for the beam current Ib, <Pb and generator current Ig , <pg. A phase modulation of the generator component results in phase and amplitude modulation of the total current, and likewise for an amplitude modulation. This is the geometric cross-coupling. Before proceeding to the dynamic coupling, firstly the steady-state beam-loading conditions must be defined. 4-5 Steady State Conditions Close to resonance the cavity behaves like a lumped parallel resonance circuit, and so when driven at angular frequency W r! the relation between voltage and current is : (2) Here R is the shunt resistance, Q the cavity quality factor, fres the resonance frequency. V is the voltage per accelerating gap, and 'ljJ is called the tuning angle. The drive frequency is fr! = h x fa where fa is the revolution frequency, and h the harmonic number. If the cavity is detuned to compensate the reactive component of beam current, then the rf generator will see a purely resistive load. Then 'ljJ = <PT and <pg = <pv, in which case the generator current amplitude and power is minimised: Ig = la + Ib sin <Pb with la = V / R. The total power is the sum of two components: dissipation Pa = L V 2/2R summed over cavities, and beam power Pb = Ib sin <Pb V /2. The required detuning (!::J.fT = fres - fr!) is: 2Q Ib !::J.fT x -j = tan'ljJ = I cos <Pb res a (3) Away from the optimal tune, the cavity must also deliver reactive power Pro Dynamic Coupling Now we give the response of the cavity voltage (amplitude and phase) to modulations of the total current vector. Introduce the Laplace transform variable s /27r, and define the denominator D = (fres +A2Qs)2 + (2Q!::J.fT)2. Thence the cross-coupling modulation transfer function, from h to V, is (4) and that for direct coupling is Gc(s) = (V /IT ) [Jres(fres + 2Qs) + (2Q!::J.fT)2)fD (5) If the cavity is detuned (!::J.fT f:. 0), then since G s( s) f:. 0 there are frequency dependent cross-modulations from amplitude to phase and vice versa. This is the dynamic cross-coupling. The KAON Factory with its high beam current, and in particular the Booster with its high synchrotron frequency, operates in a regime where both the geometric and dynamic couplings can seriously impair the control of the rf-system. 4-6 4.2.2 Stability of Beam-Cavity Interactions The steady state conditions are necessary but not sufficient for a stable cavity voltage. Because of feedback through the beam current to the cavity voltage an instability (mode indices n = 0, m = 1) can occur in which the bucket area for coherent oscillations is reduced to zero. This case was first treated by Robinson[3], for beam and cavity in isolation. Below transition the stability criteria are: Ib 2 cos <Pb -<---10 sin 21/; sin(21/;) > O. (6) The second condition is to set the drive frequency below the cavity resonance frequency, when below transition energy (0 ~ <Pb ~ 90°). For the special case <pg = 0 this reduces to IPbl < Po : the power delivered to the beam must be less than the cavity dissipation. The effect of adding phase,radial and quadrupole,amplitude loops is to alter the coherent oscillation frequency and damping rates below threshold, but not the threshold current for instability[4]. However, the control loops couple to the cavity tuning loop and thereby may produce a small increase in stability[5]. So far, only the effect of the beam-current fundamental (at frJ) has been described, but there are many other Fourier harmonics. 4.2.3 Transient Beam Loading Even if stability is established according to the preceding discussion, the magnitude of the response to large Ib variations may cause unacceptably large variations in the cavity voltage or phase. There are two types of transient beam loading problem. Injection/Ejection Transients Injection of prebunched beam, or fast extraction of a fraction of the beam, both produce sudden changes to the drive vector hi and require injection of large amounts of reactive power (P r) until the cavity tuner reacts to re-establish the optimal detuning. Periodic Transients A bunched beam with empty buckets or unequal bunches has a rich Fourier spectrum. Each component can induce a voltage if there is cavity impedance at the appropriate frequency, and this will produce a periodic modulation of the accelerating voltage. A small 4-7 modulation can be beneficial in promoting Landau damping of coupled bunch modes, but a large perturbation is certain to cause unacceptable emittance growth. Coupled bunch Mode Dampin n=1,2,3 ; m=1,2,3 Fig.4.2.2 : Beam-cavity interactions with feedbacks 4.2.4 Compensation Techniques The first step to reduce beam loading effects is to make the ratio Ib/Ig as small as possible; by minimizing the number of cavities, and increasing the power per cavity to the maximum feasible. This means high Q, matched by high peak gap-voltage and high dissipation Po. Fewer cavities will also lower the overall cost of the rf system. The requirement that beam power be less than cavity excitation power can be circumvented with the addition of feedforward of the beam-current, or feedback of the cavity gap voltage. The two methods are complementary, not competitive. Feedback is used primarily and supplemented with feedforward as necessary. The techniques are shown as loops b, d, and e in Fig. 4.2.2. There is also the possibility to suppress transient beam loading effects. And, of course, some type of feed-forward must be used to compensate for 'injection transients' which occur when beam suddenly appears in a previously empty ring. 4-8 Low-Level Feedforward An rf signal proportional to Ib is sensed upstream of the cavity and added in the low level drive chain, (loop b in Fig. 4.2.2) with such a phase and amplitude that it gives a contribution to Ig equal to the beam rf current Ib, but with opposite phase. The sum-ming point is the rf-modulator, between veo and drive amplifier. Only the way the Ig vector is controlled by the amplitude and phase feedback loops is changed, resulting in a reduced cross-coupling from the vector geometry. Though the signal corresponding to Ibexpj(71"/2 - ¢>b) does not need to be synthesised exactly, the increase in the instability threshold is limited to an order of magnitude, by gain and phase tolerances. For varying carrier frequency, the pick-up to cavity delay must be continuously adjusted, and the variations in gain and phase of the rf power amplifier corrected. This is not a closed loop system, and so adjustment can be quite critical. Fast Feedback around the Power Amplifier This is the most powerful scheme known, and two orders of magnitude compensation have been obtained[6]. Due to the straightforward implementation, this is the method of choice for the KAON Factory. Fast feedback can: (i) undo the geometric and dynamic cou-pling, (ii) reduce periodic transients, (iii) reduce injection transients. The corresponding limitations are: (i) sufficient gain (ii) band-width (iii) power and small delay. A signal is sensed from the voltage appearing at the accelerating gap, and injected into the rf drive amplifier. The effective cavity shunt impedance as seen by the beam (and phase and amplitude loops) will be reduced by a factor (l+H), where H(w) is the feedback open loop gain, (loop d in Fig. 4.2.2). The modulation cross-couplings are also reduced by this ratio, and the threshold for beam loading instabilities will therefore be raised by the same factor. The penalty is high power dissipation in the feedback components. The steady state conditions are of course unchanged, though the demand value generator current Ig' is reset when the loop is closed. A schematic is shown in Fig.4.2.3. Let us define Ao = Al X f3 x 9 where 9 is the tube conductance, Al and f3 are gains, and Z(w) the cavity complex impedance which depends on detuning angle 'ljJ. There will be a loop delay T through cabling and amplifiers, and this introduces a phase change /::l.¢> = /::l.¢>o + 271"(J - Ires)T which must be an integer multiple of 271" radian. Let bw = 271"(J - Ires). With the loop closed, the apparent impedance becomes: Z*(w) = Z(w, 'ljJ) 1 + exp(jbwT)Ao(w )Z(w, 'ljJ) (7) 4-9 r g f---i del ay 1-----. T2 loop delay is T=Tl+T2 delay T1 v Z(w) cavity Ig tube FigA.2.3: Layout of high-level Fast feedback The open loop gain is H = Ao x R. When H ~ J2 tan 'Ij; the effect of the cavity disappears: the modulation transfer functions become insensitive to the amount of detuning. The required feedback half bandwidth of Ao(w) is H x 6.fc, with 6.fc = fres/2Q the cavity half band-width. Loop stability depends on the delay, and limits the gain to H = AoR:::; 2 f Q T 7r rJ X (8) The feedback technique is very attractive, as it can reduce the effective impedance of the cavity not only at the carrier but also over a large bandwidth. However, this bandwidth must not include any of the higher order mode resonances of the cavity. There is no need for critical adjustments if the loop delay is small enough. It is this consideration that has motivated TRIUMF R&D for a solid state driver amplifier (see section 4.3.). For a varying carrier frequency, one must adjust the delay of the return path to keep the 1800 phase condition. Broad-band compensation If one wishes to compensate several more harmonics than just the fundamental, both of the last two compensation techniques are limited by the criticality of arranging the correct gain and phase (of the injected signals) at each of the revolution harmonics over a large band-width. However, this problem can be overcome by including periodic filters in the signal paths which provide well-defined phase and gain at revolution harmonics, at the expense of increasing the delay to exactly one revolution; which slightly reduces the efficiency in compensating injection transients. These variants[6] are described below. 4-10 One-Turn-Delay Comb Filter Feedback If a sufficiently short delay to implement the previous scheme cannot be achieved, then adding a comb filter to the feedback path (loop e in Fig. 4.2.2) makes rf feedback around the power amplifier possible over many revolution harmonics. The ideal comb filter has transfer function G(w) = ?o , 1 - exp(J bw x Trev) with bw' = 27r(f - /rf) . The maximum gain depends on the comb filter characteristics, which are limited by the digital technology used to realize them. The periodic response of the feedback network keeps the system stable in spite of the added delay. One-Turn-Delay Feed-forward Compensation An rf signal proportional to Ib is added in the high level drive chain, (loop c in Fig. 4.2.2) and thereby cancels the beam image current. The single turn delay produces a periodic response in the transfer function whereby the criticality of gain and phase adjustments is reduced. Adding a one-turn-delay feedforward compensation in a fixed frequency ring like the Col-lector or Extender is straightforward. This type of compensation is also required in the Driver where the radio frequency varies (over 2.5 MHz) and so the required delay changes. This can be implemented (Fig. 4.2.4) with a two-path filter combined with a digital delay. PU CAVITY BEAM MIXER ANTIALIASING...I fc·nX'REV LOW PASS FILTERS ~-e::;;:;::;?~~~ DELAY 20 - • "_x fRF ('0 ~ 8 MHz) 27 MHz h Fig.4.2.4 : One-turn-delay feedforward compensation filter for variable RF Two balanced mixers are driven at h x /0, but with a 90° phase shift. In this way the input signal is translated down in frequency where the filtering action takes place in the two channel filters. The anti-aliasing filters band-limit the signal to roughly ±10/0 • After 4-11 filtering, the signal is translated to its original frequency, by coherent up-mixing. The variation of time delay in the FIFO memory is adjusted automatically, by clocking off a multiple of the revolution frequency. Coupled-Bunch Mode Damping In the C,D,E rings there is considerable cavity impedance at h + 1 (and less so at h + 2, h + 3) due to the detuning. This may drive the modes n = h - 1, m = 1,2 (and n = h - 2, h - 3) unstable (below transition). The growth rates may be harmful. These modes can be damped by an active damping system[9] working through the cavity (loop f in Fig. 4.2.2) which introduces a large artificial coupling impedance that changes sign at (and between) revolution harmonics. The real part is positive at the upper synchrotron side-bands fr! +pfo+mfs and negative at lower sidebands. This is achievable with a notch filter, perhaps implemented with 6 channels. The ideal notch filter has transfer function G(w) = 1 - exp(j6w' x Trev). Alternatively, if (as in the Booster) a broadband all-mode beam damper (see section 5.2) is built to combat instabilities driven by the parasitic cavity resonances, then there is no need for coupled-bunch mode damping through the cavity. Quadrupole Mode Damping According to Sacherer's formulae[7] the impedance of the detuned cavity damps all n = a modes: m = 1,2,3 etc. This picture is, however, strongly modified by the phase, amplitude and tuning feedback loops. While the dipole (m = 1) mode will be strongly damped by the phase loop, both the quadrupole (m = 2) and octupole (m = 4) modes may be unstable when Landau damping is lost. A simple feedback loop (loop g in Fig. 4.2.2) feeds amplitude variations of Ib into the cavity amplitude loop with an appropriate phase shift (90°) such that both modes can be damped. Tuning Loop When a tuning loop is needed, its response can also be made insensitive to the beam loading if the normalized reactive power delivered by the amplifier is used as input to the tuning feedback loop[ll]. The desirable tuner response depends on several factors, including sustainable reactive power and beam stability, but it is expected that response up to 5 kHz should be adequate. In all cases where feedback or feedforward is used, the fast suppression of the initial tran-sient depends on the capability of the generator to provide the required change in Ig. The magnitude of the maximum required Ig can be reduced by anticipated detuning: the cavity is detuned by half of the expected detuning required to reactively compensate the injected beam load. 4-12 4.2.5 Compensation Systems for Each Ring The parameters for each ring and a listing of the feedback and feedforward loops required are shown in Table 4.2.1 (see end of Subsection 4.2.6.) The cavity quality factor includes the loading down due to tuner and power tube. The sections following summarize the characteristics of the accelerating system for each rmg. The Accumulator Compensation System The beam-loading ratio is tan'IjJ = 9.9. To make the transfer functions for (cp, A) mod-ulations about (h X fo) insensitive to detuning requires a minimum gain a = v'2 tan 'IjJ. In order to give some margin for future improvements in beam current, we shall take a = 2 tan 'IjJ; giving an rf feedback gain H = 20. No problem will be encountered in ob-taining the required gain and band-width (92 kHz) to ensure stability of phase, amplitude, tuning and synchronization feedback loops. Forty-five identical bunches give a Fourier spectrum containing only frequencies P x fr! where P is an integer. The 5 bunch gap for kicker rise breaks the periodicity of 45 buckets, and gives rise to 'extra' Fourier components at harmonics Pfr! +nfo with n = 0,1,2, ... 44. Because the revolution harmonics are widely spaced, the lines h ± 1 etc. lie well outside the cavity band-width. This has two consequences. The modulation of the cavity voltage is small, and the gain to reduce it any further is prohibitively large (because of the delay limitation). The induced voltage at the nearest side-frequency pair is L\ Vh±l = !Z!h±l!Ib!h±l = 2.8 kV per turn, giving an amplitude/phase modulation index of L\~V = V44 -::- V46 = 2.18% (±1.10). V V (9) Taking into account all the higher harmonics, we get twice the amplitude above; but it still remains within acceptable limits. There will be no injection beam-loading transients since the ring is filled gradually. Further, little or no reactive power is required since the fast tuning loop will track the relative beam-loading. A coupled-bunch mode feedback loop will not be needed to damp dipole mode driven by sidebands of h + 1, since the bunches are well damped by Landau damping throughout most of the cycle. 4-13 The Booster Compensation System Because of adiabatic damping of the bunch phase-extent, the cavity voltage is kept low toward extraction; and this dominates the beam-loading parameters during the cycle. Sufficient gain of the power amplifier rf feedback loop (H = 46) can be obtained to ensure the stability of the required phase, amplitude, tuning, radial, and synchronization feedback loops. Since the revolution harmonics h ± 1 are outside the cavity band-width, there is little point increasing the gain to H = 133, the delay-limited value for Tdelay = 70 ns. For most of the cycle, the periodic transients from the five-bunch gap induce modulations ~ V/V of ~ 3%, which is acceptable. Close to extraction this modulation rises to 7% which is marginal. Consequently, there is probably no need for secondary compensation, such as comb-filter feedback or one-turn-delay feedforward. Wide-band solid state power amplifiers with the necessary band-with to include the tun-ing range plus the feedback bandwidth have been demonstrated (see section 4.3). Conse-quently, non-tracking amplifiers are used. Sufficiently small delay to allow 4m cabling for shielding purposes have been demonstrated at TRIUMF and elsewhere. Further, means[8] are available for controlling the loop phasing accurate to ±10° over the full range of cavity frequency swing without introducing further delays. The injection beam loading transient is damped by the above mentioned feedback loops and the required power is well within that available. Landau damping due to space charge is lost in the latter part of the cycle and a coupled-bunch mode feedback system is probably required to damp the n=1,44 mode induced by the detuned cavity. The mode is driven by side-bands of (h + l)fo. A simple feedback loop for the quadrupole mode (n = 0, m = 2) damping may also be required because of instability associated with the tuning loop. Collector Compensation System The Collector ring differs in several ways from those just described. The revolution har-monics are spaced 5 times closer together, and several fall within the cavity bandwidth. The Fourier spectrum changes five times during the 100 ms cycle, as each batch of 40 bunches is transferred from the Booster. With 5 equal batches present and 5 kicker gaps, there are strong Fourier components at frequencies h ± 5, h ± 10, etc. With missing batches or part way during the cycle, there are also strong components at h ± 1, h ± 2, etc. Sub-stantial measures are required to reduce the cavity voltage modulations arising from all these beam-current harmonics. To reduce the apparent impedance at the first four side-frequencies about fr f requires the 4-14 maximum possible feedback gain and bandwidth consistent with the delay limitation. A 65 ns delay implies a gain H = 200 and a feedback half bandwidth of 1.2 MHz (which includes 4.5 revolution harmonics). With missing batches the voltage modulation due to each of the first two side-frequencies is ~ Vh±l = IZ*Ih±lIIblh±l = 31 kV summed about the ring. Including the contributions from frequencies up to h ± 4 we find a modulation index ~ V/V = 8.4%. To this must be added the modulation which arises from the kicker gap harmonics at h ± 5, h ± 10 which lie outside the feed-back band-width. It is clear that a single turn delay feed-forward will be required. When all batches are present, the cavity voltage modulation is entirely due to the kicker gaps, and is found to be ~ V/V = 4%. The rf feedback gain (H = 200) required to stabilize the phase and amplitude loop can readily be obtained. Since the maximum detuning (0.152 MHz) places the cavity resonance midway between the hand (h + 1) revolution harmonics there is a n = 1,44 coupled bunch instability. This can be damped by a single channel filter and phase-shifter feeding the RF cavity, or as part of a dedicated stripline wide-band damper. A simple quadrupole damping loop (n = 0, m = 2) may be needed for tuner-related instability. The Driver Compensation System The power amplifier rf feedback gain (H = 56) to ensure stability of the required phase, amplitude, tuning, and radial loops under the beam loading is readily obtained. However, this is insufficient to compensate transients. There are two causes of periodic transients: (i) missing or unequal batches giving lines at h ± 1, h ± 2, etc. (ii) kicker gaps giving lines at h ± 5, h ± 10, etc. Since the revolution harmonics are closely spaced, and several harmonics lie within the range of the delay limited feedback bandwidth, the gain is pushed to the maximum value of H = 110 for Tdelay = 70 ns .. However, the cavities do not have sufficient stored energy to suppress the kicker-gap periodic transients. They lie outside the fast-feedback bandwidth and would give rise to a voltage modulation of !::J. V/V = 15% which is unacceptable. So a one-tum-delay feedforward system with comb-response at h ± 5, h ± 10 about frJ (ie 7.2 MHz bandwidth) is required for each cavity. This will also serve to compensate for missing batches. These measures are sufficient to cope with injection transients and periodic transients from missing batches and kicker gaps. The extreme detuning in the Driver (0.4 MHz), forces a minor complication on the FFB system. Strictly, it is the product of cavity resonance frequency and loop delay (JresTdelay) which should be an integer. However, if the detuning is small, then the phase-adjustment can track with the carrier (Jrf) without serious error. In the Driver, the peak detuning is 23% of the swing in frc, and it is essential for the FFB loop phase-adjustment to track fres. In the Driver, there is a point near extraction where fT = fo ,i.e. the cavities are tuned to 4-15 the n = 1,224 coupled-bunch mode. It is likely that n = 2,223 is also excited, by sidebands of (h +2)10' Even though the coupled-bunch mode growth rates are slow, a coupled-bunch mode feedback system must be allowed for, since Landau damping is lost during most of the cycle. For this reason, quadrupole mode (n = 0, m = 2) damping may be required as well. The Extender Compensation System The Extender differs from the Collector and Driver in one important respect: the harmonic number is h = 230, due to the by-pass sections which increase the circumference. There are four kicker gaps of length five and one kicker gap of length ten buckets. Breaking of the five-fold symmetry results in strong Fourier components at side frequencies h ± 1, h ± 2 even when there are no missing or unequal batches. Sufficient power amplifier rf feedback gain (H = 100) can easily be obtained to ensure stability of the required phase, amplitude, and tuning loops. To reduce the periodic tran-sients, the first step is to increase the gain and feedback bandwidth beyond that required for loop stability and up to the delay-limited values. When all five batches are present, and with a gain of H = 196, the modulation contributed by side-frequencies h ± 1, ... h ± 4 is ~ V IV = 2.6%. The kicker gap frequencies at h ± 5, h ± 10 contribute a further modulation ~VIV = 4.8%. Even though the bunches are not damped by Landau damping due to the inductive wall effect, the growth rates from the detuned cavity are so slow (6.6 s) that a coupled-bunch mode feedback system may not be needed, unless there are troublesome higher-order-mode resonances. 4.2.6 Summary and Conclusion In all five rings in the chain, the fundamental component of the beam current exceeds that of the generator current required to drive the cavities by a factor 2: 10 at some point in the operating cycle. By any measure, this represents heavy beam loading; and places great demands on the rf-control system to maintain stability. The driving rf vector is controlled by amplitude, phase, radial and synchronization feedback loops, while the tune is controlled by a tuning loop. To ensure stability of these loops under the heavy transient and steady state beam loading it is essential to equip the final power amplifiers with fast rf feedback with sufficient gain to reduce the apparent Q of the cavity-amplifier system as seen by the beam and the feedback loops mentioned above. The principal limitation in the FFB loop is caused by the feedback path delay. However, sufficiently short delay can be obtained in all cases. To ensure that degradation of the 4-16 longitudinal emittance is minimized, further beam-loading compensation is required. In those cases where a sudden injection of an already bunched beam occurs (B, C, D & E rings), the combination of phase, amplitude, tuning, and power amplifier feedback reduces the injection beam loading transient to a duration short compared with the synchrotron period. The five-bucket kicker-gaPes) cause periodic transient beam loading. In the Accumula-tor and Booster, the stored energy in the cavities is sufficiently high (the off-resonance impedance sufficiently low) to attenuate the resulting voltage modulation to an acceptable value. Missing batches from the Booster produce additional periodic transients, as does the ten-bucket kicker-gap in the Extender. The power-amplifier fast feedback bandwidth can be made sufficiently wide to reduce the modulations to a marginally acceptable value (C and E rings). In the case of the Driver neither the cavity impedance nor the power amplifier feedback attenuates the periodic transients sufficiently, and an additional one-turn-delay feedforward compensation with sufficient bandwidth must be added to reduce the modulation to an acceptable limit. Since compensation of the reactive beam loading requires detuning of the cavity, lon-gitudinal coupled-bunch modes with mode numbers just below the rf harmonic number (n = h - 1, h - 2) are excited, often with very fast growth rates. Although one-turn delay feedforward reduces these by an order of magnitude, coupled-bunch mode feedback damp-ing may still be required in all rings except the Accumulator. If these are the only modes excited in the C and E rings, then damping through the cavities is possible with little cost except for some signal processing and narrow-band power amplifiers. Because of cavity parasitic resonances, the Booster and Driver will be equipped with wide-band dampers to treat all 45 and 225 modes respectively; and so extra channels into the cavities may be unnecessary in this case for these rings. The wide-band damper is described in section 5.2. Although the detuned cavity impedance itself would stabilize the n = 0, m = 2 quadrupole mode, the combined action of the coupled amplitude, phase and tuning loops may drive this mode unstable, so a simple quadrupole damping loop would be required when Landau damping is lost due to space charge or inductive wall effect. In summary, the cavities, amplifiers, and control systems required in the five synchrotron rf systems proposed here are quite complex and would push the limits of existing technology. For instance, the beam current in KAON would be '" 6 times greater than in the FNL[10] 50 MHz system. However, the requirements of each ring are not greatly in excess of the best performance that has been attained in various machines operating at Fermilab and at CERN. Hence, the overall system is entirely feasible. 4-17 Table 4.2.1 Summary of RF Systems and Feedback Systems Ring Accumulator Booster Collector Driver Extender Carrier Frequency [MHz] 46.11 46.11-60.75 60.75 60.75-62.53 62.53 Harmonic number 45 45 225 225 230 Revol'n frequency [MHz] 1.025 1.025-1.350 0.270 0.270-0.2779 0.2719 Max. Detuning [MHz] .0461 0.1958 0.1519 0.4107 0.1351 Cavity Q (bate) 5000 3000-5000 5000 3000 5000 R/Q per cavity 100 35 (Q vs fo linear.) 100 100 100 VrJ Volts/cavity [kV] 172 30.4-62.5 100-154 46.4-142 142 Max. h cos ¢>b/ fo 9.88 32.2 25.0 39.4 21.6 Min. FFB Gain 14 46 35 56 31 Suggested Gain 19.8 64.4 200 110 196 Cavity ~ BW (bare) [kHz] 4.61 6.08-7.69 6.07 10.1-10.4 6.253 FFB ~ BW [MHz] 0.092 0.495 (track fo) 1.214 1.146 (track fo) 1.226 Max. allowed delay [ns] 65 70 65 70 65 Single turn delay FF no not likely likely yes v.likely Coupled-bunch mode FB no yes likely yes likely Injection compensation no yes yes yes yes fsynch (bare) [kHz] 41.2 47.3-7.0 7.8 8.53-0.44 0.44 Phase loop yes yes yes yes yes Amplitude loop yes yes yes yes yes Radial loop no yes no yes no Synchronization loop yes yes,2 no yes yes Quadrupole loop no yes TBD yes TBD Number of Cavities 3 12 12 18 6 Length (em) 225 123 171 169 166 Number of Amplifiers 3 12 24 36 6 Total Volts/ring [kV] 514 365-750 1300-1846 836-2550 847 Po / ring [k W] 88.8 31.7-185 284.6 64.7-602 119.6 Pb/ring [kW] - 234-926 - 1984-6350 -Pr/ring [kW] - 1725 2600 7770 1300 Tuning range [MHz] .0461 14.84 0.152 2.191 0.1351 Fast Tuning yes yes yes yes yes Max. tuning angle [deg] 84.2 88.2 87.7 88.6 87.3 4-18 4.3 Radiofrequency Systems 4.3.1 Introduction The specifications for the radiofrequency accelerating systems are given in the previous section along with the requirements for compensating for the high beam loading. This sec-tion deals with the design of the cavities and amplifiers which are based on the technology developed at Fermilab, LANL, AECL and CERN. The system required is near the limits of operation attained for frequency range, voltage, and beam loading at these laboratories. The rf parameters for each ring are listed in Table 4.2.1, Section 4.2. 4.3.2 Booster Cavity The rf cavity for the Booster synchrotron requires a frequency swing from 46 MHz to 61 MHz at a repetition rate of 50 Hz. In most synchrotrons, ferrite tuning of rf cavities has been used to provide the necessary change in frequency. The basic principle involves changing the permeability of a ferrite material to vary the inductance of the resonant cir-cuit by applying an external magnetic field. The external magnetic field can be applied either parallel or perpendicular to the rf magnetic field. The initial reference design[12] was based on the Fermilab booster cavity which is a double gap drift tube cavity with par-allel biased ferrite tuners. At the same time LAMPF was developing a single gap cavity using perpendicularly biased yttrium-garnet ferrite to vary the frequency from 50 to 6q MHz for their AHF proposal.[13] Measurements on the LAMPF cavity[14] indicated that our lower frequency of 46 MHz could be reached while still maintaining a high magnetic Q. Fig. 4.3.1 shows the results of this measurement on the LAMPF protoype cavity. A comparison[14] was made between a parallel biased nickel-zinc ferrite design and a perpen-dicularly biased yttrium-garnet ferrite design and on almost all counts the perpendicular bias was favoured. The main advantages of this microwave ferrite are its high magnetic and electric Q's and the absence of non-linear rf loss mechnisms[15] which are very troublesome with parallel biased Ni-Zn ferrite. Much higher gap voltages should be possible, reducing the number and cost of the cavities and even more important, the gap impedance driving coupled-bunch instabilities. [However in the perpendicular bias case the ac magnetizing circuit[16,17] is much more complicated and special care must be taken to minimize the induced eddy current losses when designing the magnetic circuit, tuner cavity and cav-ity support structure]. Because of these considerations, development of a parallel biased cavity was suspended, following successful completion of signal level tests on a full-scale prototype with air-dielectric tuners, and all efforts were switched to demonstrating the feasibility of perpendicularly biased cavities. 4-19 4 .0 3.5 3 .0 ~ -.....:l -~ 2.5 < ~ :::;; c:t:: ~ P-. 2 .0 1.5 1.0 40 Sept.4 1987 g TRIUMF meAsurements • LANL meAsurement. Ii 3000 x l/427 f' 1 + 610/H 45 60 FREQUENCY 65 60 ( MHz. ) 65 Figure 4.3.1: Results of TRIUMF's measurements on the LAMPF prototype cavity The dc-biased prototype cavity constructed and tested at LAMPF[18] was kindly made available to TRIUMF and has now been almost entirely rebuilt at TRIUMF with a com-pletely redesigned tuner section for ac-bias operation. A cross sectional view of the ac-biased ferrite tuned cavity is shown in Fig. 4.3.2 and a more detailed view of the tuner portion of the cavity is shown in Fig. 4.3.3. The return yoke consists of twelve magnet sectors held together by an aluminum clamping plate and a set of tie rods. The sectored design provides room for the entrance and exit of the stranded cable which requires a large bend radius and further provides easy access for the water jacket cooling lines. To save on the cost of dies, each sector is made from three separate rectangular shaped lami-nated blocks which are then tapered by cutting and grinding. The coil consists of twelve turns of stranded cable made up of 82 strands of #9 heavy formvar-insulated magnet wire surrounding a copper tube for water cooling. The cable is wrapped with fiberglass tape. To minimize the risk of shorts due to the vibration of individual strands, and to provide insulation against the high voltage to ground, it was decided to vacuum impregnate the voids between the individual strands as well as the coil structure itself. The tuner cavity can be considered as consisting of two rf membrane end walls, the outer conductor which is the cooling jacket for the ferrite, the ferrite rings interleaved with beryllium oxide rings and the centre conductor which is a tapered conical shape. To reduce eddy current losses the rf membrane is made from 0.5 mm stainless steel and plated with 0.0127 mm copper for the rf currents. To further reduce the eddy current losses 48 radial slots are cut into the rf membrane. The water jacket consists of an annular cylinder of copper 159 mm high and 16 mm thick with machined water cooling channels. The jacket has the potential of generating large eddy current losses but an insulating break has been incorporated to min-imize this effect. An insulating break has also been incorporated in the membrane cooling wheel to minimize eddy current losses. The centre conductor is a 2.0 mm thick stainless steel tube plated with 0.0127 mm copper for the rf currents. Although the thicker material 4-20 INPUT CAVITY .,,,_,,<.u;.,,,,nNG GAP Figure 4.3.2: Ferrite tuned cavity for the Booster synchrotron ''(RRnE toOUNC JACKO ... ...-..... 0 .5 1&1 5.5. $f1([1 Cu PlAlID Yrml 41 fW)1Al. Slots (["IlR COHOUCIOR _......,. $IOOC COOlING CtRCUIJ COlI. AUJ.. Q.ANPU«i PlAlE • ,~ ,...........fI( ROO H£J'I¥[(H S£CIORS ,/ ('WeER stOCK COOUHC ClAC'UII 0\1l[1I COHDUCIOR __ S.S c ... PlATtO C(Hf(R COHOUCTOR S s. t" PlArro Figure 4.3.3: Cross sectional view of the ac-biased tuner for the Booster 4-21 increases the eddy current losses it is required to support the vacuum load. However the centre conductor is water cooled to take care of the heat generated by the additional eddy current losses. One of the consequences of this type of ferrite tuner is that it produces a magnetic field on the beam axis[19]. The effect of this magnetic field is greatly reduced by operating two of the cavities back to back such that the magnetic fields produced on the beam axis by each tuner are in anti phase and cancel each other for first order effects[20]. 4.3.3 Driver and Storage Ring Cavities The cavities in rings A, C, D, and E will be similar to those designed and built by AECL in Chalk River for the HERA 52 MHz rf System[21] which is based on the Fermilab Debuncher Ring rf system design (Fig. 4.3.4). The cavity can be considered as two half-wave resonators placed back to back or to a TM-OI0-like cavity loaded by an intermediate cylinder SUPPORT STAND Figure 4.3.4: Cavity for the A, C, D and E rings based on the 52 MHz rf system designed and built by AECL for HERA supported at the zero field symetry point in the centre of the cavity. To minimize multi-pactoring problems the inside of the cavity will be titanium coated. The inner conductor, intermediate cylinder and portions of the outer cavity wall are water cooled. Doubled wall 4-22 construction for the intermediate cylinder and inner conductor allows cooling tubes to be attached to the cylinder walls in the annular volumes, providing a cooling system with no water to vacuum joints. For the A, C and E rings where the carrier frequency is fixed and only a small detuning range is required, the tuners will be an AECL design as shown in Fig. 4.3.5a. The tuner design is based on ferrite rings perpendicularily biased in the longitudinal direction and coupled to the cavity via an inductive coupling loop. The num-ber of tuners and ferrite rings will differ for each system depending on the detuning range required. The ferrite rings are interleaved with beryllium oxide rings and surrounded with a water cooling jacket. In the D ring where the carrier frequency has a 1.8 MHz tuning range, the tuner design will be similiar to the tuner used at Los Alamos[22] on their main ring prototype ferrite-tuned cavity which uses ferrite rings perpendicularily biased in the radial direction as shown in Fig. 4.3.5b. The water cooled bias coils are placed between the ferrite rings providing cooling for the ferrite as well as the coils. The current in the coils are set up such as to produce magnetic flux lines which go radially outward through one of the ferrite rings and return radially inward through the other ring. The rf walls between the middle bias coils and the ferrite rings are slotted radially to permit the high frequency components of the bias field to rapidly penetrate through the wall and into the ferrite. This configuration reduces the energy stored in the bias magnet with a corresponding reduction in the bias power supply requirements. 4.3.4 Power Amplifiers All of the rings will use 4CW150000E amplifier tubes and only the circuitry will vary slightly from ring to ring. The gap voltages per cavity are limited by either the voltage breakdown at the coupling point of the tuners or the power density in the ferrite material. The amplifiers will be driven by the same solid state driver which was developed at TRI-UMF for the prototype Booster rf system. The main requirements for the driver amplifier are an output power of 2.4 kW, a power gain of 35 db and a group delay of less than 50 ns. Fermilab presently uses 14 tetrodes connected in parallel for their driver ampli-fier. At present the tubes have become very expensive and must be replaced with great frequency and therefore the possibility of using solid state devices for a driver amplifier was investigated. Using the Motorola TMOS-FET solid state devices and push-pull am-plifier design[23], three different amplifier designs were built and tested. In order to take advantage of the shorter time delay, the greater gain, the lower price and best operating reliability, the decision was taken to combine twelve 300 W rated units operating at a conservative level of 200 W each through a ferrite loaded transmission line transformer. An overvoltage and overcurrent fast protect system has also been developed to protect the solid state devices. The rf driver chain has been tested out to full power into the properly loaded and matched input cavity of the rf amplifier. 4-23 INTERLEAVED FERRITE AND BERYLIA RINGS FERRITE PERPENDICULAR BIASING SOLENOID ( A, C, AND E RING TUNER ) FERRITE RINGS ( D RING TUNER ) Figure 4.3.5: Tuner designs for the storage ring and Driver cavities (a) based on the AECL design for the HERA cavities using ferrite rings perpendicularily biased in the longitudinal direction. (b) based on the LANL design for the Main ring cavity using ferrite rings perpendicularly biased in the radial direction. 4-24 4.3.5 Higher Order Cavity Modes A major problem to be overcome for the proposed rf cavities is the excitation of higher order modes by large beam currents. The modes that are of most concern have resonant frequencies that extend to 1 GHz. The magnitude of the voltage that is excited for a particular mode is directly related to its shunt impedance. It is essential to reduce the shunt impedances of these modes to less than 1000 n without significantly affecting the fundamental mode. A mode damping scheme that has been investigated is based on coupled transmission lines[24]. The technique investigated is to introduce into the cavity a separate damping transmission line that is terminated at one end by a resistive load as shown in Fig. 4.3.6. The geometry of the damping line is selected so that the voltage induced at the resistor by the magnetic and electric coupling cancel at the fundamental frequency resulting in no attenuation while at the frequency of the higher order modes the induced voltage will not be zero and these modes will be attenuated. Theoretical analysis and measurements on a simple stripline resonator[24] show that this scheme is effective in attenuating higher order TEM modes without significantly affecting the fundamental mode. Before investigating such a scheme it was necessary to establish a reliable measuring technique to determine the shunt impedances of the higher order modes. Three methods have been investigated[25] with the preferred method being the one which measures the transmission coefficient across the accelerating gap. The cavity is in effect in series for this measurement and therefore at the resonant frequency there is a sharp dip in the magnitude of the transmission coefficient which is a function of the shunt impedance. The graph plotted in Fig. 4.3.7 shows the effect of a coupled transmission line mode damper[26] installed in the LAMPF prototype cavity. All of the higher order TEM modes up to 1 GHz have been damped to less than 1000 n impedance, as required. The problems of incorporating such a mode damper at full power have still to be investigated. 4-25 10' S 10· LLJ U 10' Z -<t l-V) Vi LLJ 10' a::: 'Z :::> 10' ~~ I V) 10· 10-' -0 TO TUNER INOUCTIVE COUPLING STRAP ( 1 OF 3 SHOWN ) AMPLIFIER 50 0 RESISTOR ( 1 OF 3 SHOWN ACCELERATING GAP Figure 4.3.6: Arrangement of HOM damper in Booster rf cavity 10' ~ Slot DAHPER I "STALLED NO DAHPER LLJ U Z 10' <t l-V) Vi LLJ 10' a::: ~ :::> 10' I Vl 10· 10- ' 200 400 600 800 1000 0 200 -00 GOO 8CJ(J FREC'UnJCY (fvlH:;:) FREQUENCY (','Hz) Figure 4.3.7: Higher Order Mode damping in the Booster rf cavity 4-26 1(J(J0 4.3.6 Summary of RF System Requirements IA Line Two cavities powered in anti-phase will be used to modulate the energy of the H- injected beam to the A ring. Each cavity will operate at 46 MHZ with a maximum voltage of 350 kV with a ± 900 phase swing. It is proposed to use two of the A ring type cavities. Since no fast tuning is required on this cavity the voltage is limited only by the electric field gradients at the gap and therefore 350 k V should be readily attainable. The peak power required is 122 kW but the average power is only approximately one third of the peak power. Accumulator Three of the AECLjHERA-type cavities, modified for the lower frequency of 46.1 MHz will be required to provide the total rf voltage of 750 kV. Very little power will be required to develop this voltage since only cavity dissipation is required. A single amplifier will be used in each of the cavities. Booster Twelve of the perpendicularly biased tuned cavities developed at Los Alamos, modified for ac 50 Hz operation and increased tuning range, will be used to provide the necessary rf voltage and beam power for the Booster ring. The required installed rf power is mostly determined by the active power delivered to the beam, although the required transient reactive power required for suppression of periodic beam loading transients is an important consi deration. Collector Twelve of the AECLjHERA-type cavities, modified for the higher frequency of 61 MHz will be needed to develop the required voltage of 1860 kV. However 24 amplifiers are needed to provide the continuous reactive power for periodic transient beam loading compensation and therefore two amplifiers per cavity will be required. This is accomplished by replac-ing the balancing capacitor in Fig.4.3.4 by a tetrode tube amplifier identical to the final amplifier shown. Driver The number of cavities required in this case is determined primarily by the 6 MW peak beam power requirements. Eighteen of the AECLjHERA-type cavities, modified for the higher frequency and using the LANL radial- biased tuner, will be used. Each cavity will require two amplifiers to provide the cavity losses plus the beam power requirements. Sim-ilar to the Collector ring, the balancing capacitor will be replaced with a second amplifier. 4-27 Extender Six modified AECL/HERA type cavities will be used to provide the required 850 kV of rf voltage. Although the continuous active rf power requirement is low, the rf power requirement is mainly determined by the amount of transient reactive power required for the fast suppression of the injection beam loading transient. 4.1.7 Conclusion The cavity and amplifier requirements in the rf systems proposed here are complex and will push the limits of existing technology. However the requirements of each system are not greatly in excess of the best performance that has been attained in various machines operating at Fermilab and at CERN. 4-28 References [1] D. Boussard, The PS 200 MHz rf system, CERN-SPS/ARF/78-6 [2] R.Reece, L. Ahrens, W. Van Asselt et aI, Operational Experience and Techniques for Controlled Longitudinal Phase Space Dilution in the AGS, Chicago PAC 1989. [3] K.W. Robinson, CEAL-1010, (1964) [4] S.R. Koscielniak, A General Theory of Beam Loading, TRI-DN-89-K25. [5] F.Pedersen, Beam Loading Effects in the CERN PS Booster, IEEE Trans.Nuc.Sci. Vol.22 No.3 p.1906 (1975). [6] D.Boussard, Control of cavities with high beam loading, CERN SPS/85-31 (ARF). [7] F.Sacherer: A Longitudinal stability criterion for bunched beams, Proc. 1973 PAC, San Francisco, p.825. [8] D.Boussard, Variable phase shifter for the fast rf feedback in the Booster, TRI-DN-89-K82. [9] B.Kriegbaum and F.Pedersen, Electronics for the Longitudinal Active Damping Sys-tem for the CERN PS Booster, IEEE Trans. Nuc. Sci, Vol.NS-24, No.3, p.1695 (1977). [10] J.Griffin, A numerical example of an rf accelerating system, AlP Fermilab Summer School, 1981. [11] F.Pedersen, IEEE Trans. Nucl. Sci NS-32 (1985) p 2138. [12] R. L. Poirier and T. A. Enegren, Status of RF Development Work on a Ferrite Tuned Amplifier Cavity for the TRIUMF KAON Factory, IEEE Particle Accelerator Con-ference, Washington, D.C. p 1499. (1987). [13] W. R. Smythe et al., RF Cavities with Transversely biased Ferrite Tuning, IEEE Particle Accelerator Conference, Vancouver, B.C., p 2951. (1985). [14] R. L. Poirier et al., Parallel bias vs Perpendicular Bias of a Ferrite Tuned Cavity for the TRIUMF KA ON Factory Booster Ring, European Particle Accelerator Confer-ence, Rome, Italy, p 1321. (1988). [15] J. E. Griffin and G. Nicholls, Review of Some Dynamic Loss Properties of Ni-Zn Accelerator RF System Ferrite, IEEE Particle Accelerator Conference, San Francisco, California, p 3965. (1979). [16] C. Haddock et al., Design of an A C Magnetic Biasing Circuit for the KA ON Factory Booster RF Cavity, Proc. 11th Magnetechnology Conf. Tsukuba (1989) 4-29 [17] C. Haddock, Evaluation of Eddy Current Induced Losses in the Cooling and Support Structure of the Booster RF Cavity Using the Code PE2D, TRIUMF TRI-DN-89-K100. [18] C. C. Friedrichs et al., Test Results of the Los Alamos Ferrite- Tuned RF Cavity, IEEE Particle Accelerator Conference, Washington D.C. p 1896. (1987). [19] R. L. Poirier et al., Perpendicular Biased Ferrite Tuned Cavity for the TRIUMF KA ON Factory Booster Ring, IEEE Particle Accelerator Conference, Chicago, Illinois, (1989). [20] R. Baartman, Optical Effects of Tuner Solenoids, TRI-DN-89-K46. [21] L. W. Funk et al.,52 MHz RF System for HERA, European Particle Accelerator Conference, Rome, Italy p 1099. (1988). [22] G. R. Swain, Design of Main Ring Cavity, LA-11432-C, Proc. Advanced Hadron Facility Accelerator Design Workshop, Los Alamos, New Mexio, p 365. (1988). [23] S. Kwiatkowski, Design of the 150 k W, 46-62 MHz Power Amplifier for the KA ON Factory Booster Ring, European Particle Accelerator Conference, Rome, Italy, p 1220. (1988). [24] T. Enegren et al., Higher Order Mode Damping in KAON Factory RF Cavities, IEEE Particle Accelerator Conference, Chicago, Illinois, March 1989. [25] T. A. Enegren, Measurement of RF Cavity Shunt Impedance, TRI-DN-90-KI04. [26] T. A. Enegren, Coupled Transmission Line Higher Order Mode Damper, TRI-DN-90-KID5. 4-30 BEAM STABILITY AND INSTRUMENTATION Chapter 5 5 BEAM STABILITY AND INSTRUMENTATION 5-1 5.1 Collective Instabilities ........... 5-1 5.1.1 Longitudinal Microwave Instability 5-1 5.1.2 Longitudinal Coupled-Bunch Instabilities 5-2 5.1.3 Fast Transverse Instability ........ 5-4 5.1.4 Transverse Coupled-Bunch Instabilities 5-4 5.1.5 Electron-Proton Instabilities 5-6 5.2 Stabilizing Systems . . . . . . . . . . 5-8 5.2.1 Emittance Dilution in the Collector 5-8 5.2.2 Longitudinal Coupled-Bunch Damper 5-10 5.2.3 Transverse Active Dampers 5-12 5.3 Beam Instrumentation 5-17 5.3.1 Introduction 5-17 5.3.2 Modes of Operation 5-17 5.3.3 Instrumentation for Synchrotrons and Storage Rings 5-18 5.3.4 Instrumentation for Transfer Lines 5-22 5.3.5 Procedures and Measurements 5-22 5.3.6 Potential Problem Areas . . . . 5-23 5 BEAM STABILITY AND INSTRUMENTATION 5.1 Collective Instabilities Collective instabilities are induced by the beam through the charges and currents set up in the equipment it passes through. Such instabilities at frequencies not too far above the rf frequency can generally be controlled by feedback systems. Higher frequency instabilities can be suppressed by furnishing a sufficiently large momentum spread. 5.1.1 Longitudinal Microwave Instability A fast, short wavelength instability can occur if the longitudinal emittance of a beam of given intensity is too small. This is called the microwave instability. By 'fast', we mean faster than the synchrotron frequency, and by 'short wavelength', we mean short compared with the bunch length. The accumulated longitudinal emittance in the A ring was chosen to be very large compared with the TRIUMF emittance but not so large that extra cavities would be needed in the Booster (where space is at a premium). In this way, a longitudinal emittance of 0.048 eV-s was arrived at. To test for microwave stability, we use the Keil-Schnell criterion with local values of current and momentum spread as first suggested by Boussard[l]: 1 IZII < ,82117IEBc (8P)2 n e I P Cwhh (1) Here I = circulating current, Bc = average/peak beam current, E = total energy of beam, and 17 = ,;2 - ,-2. The longitudinal impedance divided by mode number ZII/n has a negative imaginary space charge term and a positive imaginary to positive real (depending upon frequency) term arising from the short range part of the vacuum chamber wakefield (the broadband resonator) . Calculations in which contributions from vacuum chamber irregularities, pick-up monitors, collimators, etc. are included[3]' show that the magnitude of the latter term is easily restricted to around 7 n in all rings and with effort can be made as small as 2 n. In rings A and B, the space charge term dominates and is a maximum at injection. How-ever, condition (1) is worst at extraction in B where it gives an upper limit on the circulating current of 7.2 A. We are safe even if the broadband term is 100 n. In the Driver, the broadband resonator dominates and condition (1) is again worst at 1 It can be shown[2] that this criterion is equivalent to constraining the fractional change in synchrotron tune to an upper limit of 0.28 . This makes clear the connection between avoidance of a 'fast' instability and the Keil-Schnell-Boussard criterion for a bunched beam. 5-1 extraction. If we assume the same emittance 0.048 eV-s and It = 00, (1) is violated by a factor 5, i.e. we require IZII/nl to be less than a difficult-to-attain 2 n. Two approaches are used together to alleviate the problem; 1] is made larger by making It imaginary (,t = 30)), and the longitudinal emittance is artificially increased in ring C by a factor of 4. With these measures, (1) gives an upper bound on the broadband impedance of IZII/nl = 17 n. In the Extender, as in the other rings, it is preferred to operate below transition since then chromaticity need not be corrected. In bunched mode, the Extender is stable with the same choice of It as in the Driver. On the other hand, in debunched mode, where the above consideration on the sign of 1] does not apply, local values of momentum spread become too small compared with the stability threshold for this choice of It. For this reason, the Extender is designed with It tunable from 30) down to 10. With the latter value, the debunched beam is stable for a broadband impedance, IZII/nl, of up to 7.4 n. 5.1.2 Longitudinal Coupled-Bunch Instabilities RF Cavities: In all the rings, the rf cavities dominate the longitudinal impedance. This is because the synchrotrons are fast cycling and therefore require high rf voltages. Not only are a relatively large number of cavities needed but also they must have high Q's, else electrical power consumption becomes too large. As a result, the rf cavities have many high-Q, high-impedance parasitics. In the dc rings, this does not necessarily cause problems because the parasitic resonant frequencies, W p , can in principle be made to lie between multiples of the beam's revolution frequency, woo In the synchrotrons, however, the revolution frequency changes as the beam gains energy and so coincidence between a coupled-bunch mode frequency and a given parasitic is unavoidable. Since the synchrotrons are fast cycling and the parasitics are narrow, a given resonance is crossed quickly. Then a possible strategy is to tune the parasitics not to overlap with each other and not to drive the same coupled bunch mode at the same time. Unfortunately, there are too many cavities and too many parasitic modes for this strategy to be useful. All the parasitic modes must be passively damped to a level where they can be controlled by active damping. At the frequency of a coupled bunch mode, nwo + mws , the impedance ZII gives rise to a fractional change in synchrotron frequency which for resonator impedances with Q ~ n/ h is given to within a factor ~ 1 by the ratio of )IZII to V cos ¢>s [4]. Resistance gives rise to exponential growth of the coupled bunch mode, and reactance gives rise to a longitudinal tune shift . The fractional change, 8, in synchrotron frequency must be kept small compared with 1 else different longitudinal modes will couple, making it difficult to damp them actively. A conservative upper limit for the magnitude of 8 is 0.1 [5]. Keeping 8 small also achieves three other goals. (1) The power needed in the damping system goes as P. Even for 5-2 b = 0.1, tens of kilowatts of power are required. (2) In the first half of the accelerating cycles in Band D, where the bunches fit most tightly in the buckets and little longitudinal motion can be tolerated, the fractional frequency spread due to nonlinearity of the rf waveform is about 0.1. Hence, with b < 0.1, Landau damping can be effective. (3) Besides causing instability, large impedances can reduce the effective stable phase space area when wp = nwo. This is because the beam contains Fourier components of '" 0.5 A at all n (due to the 5 empty bucket 'kicker gap'), and", 6 A at n =integerxh. Limiting b to 0.1 limits the induced voltage to 20% of the rf voltage, thus momentarily changing the bucket height by 10%. By design, the bunch occupies 80% of the bucket height, so b = 0.1 allows some margin for tails to occur. Using I = 3 A and the voltage programs of Section 4.1 in the inequality IRs V '" < 0.1 cos 'fJs (2) gives an upper limit for the parasitic shunt resistance, Rs, of 10 kn for the Booster and 28 kn for the Driver. In both cases, this gives a per cavity limit of close to 1 kn. Methods of achieving this limit are being developed and are reported in Section 4.3.5. Besides the rf cavities, there are two other impedances which should be considered in connection with longitudinal coupled-bunch instabilities; reactive impedance due to the vacuum chamber and space charge, and low frequency impedance due to the kicker mod-ules. Inductive Wall Effect and Space Charge: The broadband resonator (which gives rise to the inductive wall effect at low frequencies) and the direct space charge effect contribute only a real part to the frequency shifts. Hence, they do not contribute to growth rate but can cause susceptibility to instability through the loss of Landau damping. Landau damping is lost when the magnitude of the frequency shift becomes larger than the half spread in incoherent frequencies. In an effort to retain as much stability as possible, synchrotron frequency spreads have been kept large by lowering the rf voltage so that the bucket height is usually only 20% larger than the bunch. (This is also essential for reducing the Laslett tune shift at the beginning of the cycle by increasing the bunching factor.) Towards the end of the acceleration cycles in Band D, however, the bucket area tends to grow rapidly because It is being approached. Because of beam loading, V cannot be made small enough to maintain a constant bucket area. Landau damping of longitudinal modes is therefore only present in ring A and in ring C after longitudinal emittance blow-up and in the first half of the acceleration cycles in rings Band D. Kickers: Kicker modules have maximum real impedances equal to one half of the module's characteristic impedance [6]. Multiplying by the number of modules, this gives 94 n for 5-3 the Booster and 240 n for the Driver (see Section 3.3). These numbers are small compared with the rf cavity impedance. Even allowing for the fact that the Booster extraction kicker is short circuited at one end rather than matched, the active damping system required for the cavity parasitics will also easily damp longitudinal instabilities due to the kickers. 5.1.3 Fast Transverse Instability The transverse broadband (short wake) impedance in electron machines is often dominated by the large number of rf cavities needed to overcome energy loss due to synchrotron radiation. For example, in PETRA, where a fast instability caused by the short range wakefield was first observed, there were 300 rf cells, each contributing rv 4 kn/m to the transverse broadband impedance. The KAON Factory Driver ring, on the other hand, will contain only 20 cavities, each contributing much less than 4 kn/m. The impedance of the vacuum chamber can be approximated by Z , = 2RZII .... b2 n' (3) The result, using ZII/n = 7 n, is labelled ZBB in Table 5.1.1. Clearly, these are much larger than the contributions from the rf cavities. To check for a possible fast instability of the PETRA type, the above impedance, together with the space charge impedance (Zsc) are to be compared with the threshold impedance (ZTh) found by using the local (peak) value of the beam current in the coasting beam stability criterion. Two cases were considered: natural chromaticity (~ = (8v/v)/(8p/p) = -1.3) and corrected chromaticity. See Table 5.1.1. In every case, the threshold impedance is larger than the the sum of the space charge and the short-range wake impedances. However, these calculations are by no means exact and a safety factor of at least 2 is prudent. It is clear, therefore, that it would be unwise to correct the natural chromaticity in the D ring and in the E ring during bunched operation mode. 5.1.4 Transverse Coupled-Bunch Instabilities Longer range transverse wake fields, due to the resistive wall, the kicker magnets, and parasitic modes in the rf cavities will drive coupled-bunch modes. Resistive Wall Effect: In all rings, image currents will flow through a thick wall of 316LN stainless steel of resistivity 75 J.Ln-cm. In the fast-cycling synchrotrons Band D, the vacuum chamber in the magnets is ceramic but its internal rf shield is stainless steel. By 5-4 Table 5.1.1: Transverse Impedances (in Mn/m) and Thresholds for the Fast Instability Ring ZBB Zsc ZTh e=O e = -1.3 A 0.35 66. 240. 240. B (at 3 GeV) 0.35 17. 28. 33. C 1.7 41. 61. 65. D (at 30 GeV) 1.7 5.1 8.1 24. E (bunched"t = 30)) 5.4 5.1 14. 30. E (debunched"t = 10) 5.4 5.1 71. 55. 'thick', we mean that the wall is thicker than a skin depth at the lowest frequency mode. Modes occur at w = (n - lIy)wQ and the lowest unstable one is where the integer n is the nearest one above the tune lIy . The lowest frequency mode always dominates because the resistive wall impedance is proportional to W- 1/ 2. The impedances, and the resulting instability growth rates are shown in Table 5.1.2 for both the natural and the corrected chromaticity. Again, large negative chromaticities are preferred in the large rings. The resistive wall effect is significantly larger in A than in B because the tune in the A ring is below the integer rather than just above. This makes the lowest unstable mode frequency 3 times smaller in A than in B. Table 5.1.2: Impedances and Growth Rates for the Resistive Wall Effect Ring ZRW Freq. growth rate (8-1) (Mn/m) (MHz) e=O e = -1.3 A 0.22 0.30 1500. 1400. B (at 0.45 GeV) 0.13 0.80 710. 630. C 1.1 0.21 1400. 1000. D (at 3.0 GeV) 1.1 0.21 1400. 450. E (bunched) 5.6 0.21 1000. 38. E (debunched) 5.6 0.21 1000. 1000. Kickers: As in the longitudinal case, kicker impedances were estimated using formulas given by Nassibian[6]. See Table 5.1.3 for impedances and growth rates. The Driver ring stands out in this case both because of the large kick strength needed to extract at 30 Ge V 5-5 and because this ring needs two extraction kickers (see Section 3.3). It is clear that large negative chromaticity is desirable especially in the Dring. Table 5.1.3: Transverse Impedances and Growth Rates Due to the Kicker Mag-nets Ring ZKi Freq. growth rate (S-1) (Mn/m) (MHz) e=O e = -1.3 A 0.020 4.8 140. 120. B (at 0.45 GeV) 0.016 16. 72. 58. C 0.093 7.1 120. 85. D (at 3.0 GeV) 0.740 7.4 970. 270. E (bunched) 0.460 2.9 77. 3. E (debunched) 0.460 2.9 77. 77. RF Cavities: Parasitic deflecting modes in the rf cavities have frequencies which depend upon the transverse dimensions of the cavity: they therefore lie considerably above the fundamental frequency and do not couple well to the beam. Our lowest transverse mode will have a frequency ~ 300 MHz, an impedance f'V 1 Mn/m with Q ~ 30,000[7]. When a coupled-bunch mode coincides with such a cavity mode, the growth rate is only 400/s in B and 100/s in D. Because the Q is so high, the mode can be crossed quickly: if coincidence does not occur near top or bottom energies, where coupled-bunch mode frequencies are stationary, the crossing time will be a small fraction of a millisecond. Adding up the contributions from all cavities, we find that the number of e-folding times will not exceed 2. Hence, with a little care regarding the mode frequencies, transverse impedance from the rf cavities will not cause problems. 5.1.5 Electron-Proton Instabilities Electrons can accumulate in the potential well of an unbunched proton beam. If the beam is bunched, accumulation does not occur because the electrons can escape to the vacuum chamber walls between bunches. Hence, this effect is only important for the debunched mode in the E ring. Transverse unstable modes of plasma oscillation between electrons and the beam can lead to beam loss. In the ISR, with clearing electrodes not powered, the beam lifetime was around 1 hr at a pressure of 3 x 10-10 Torr, and powering the clearing electrodes reduced 5-6 the loss by 3 orders of magnitude[8]. Scaling, we find that requiring the loss to stay below one part per thousand in the E ring yields a vacuum requirement of 3 x 10-8 Torr without clearing electrodes. A more stringent requirement comes from the effect the electrons have on the protons' tune. Even a slight accumulation of electrons can destroy the delicate balance of electric and magnetic forces between protons in the beam. Specifically, to avoid augmenting the space-charge tune spread, and thereby interfering with the 1/3-integer slow extraction scheme, the ratio of electron density to proton density must be kept below f'.I II"'? = 0.001. This can be achieved either with clearing electrodes or with a maximum background gas pressure of 10-11 Torr. The former method is more economical if clearing electrodes are placed outside of the magnetic elements so as not to increase their aperture requirements. Since the electrons spiral tightly around magnetic field lines, they can only reach the clearing electrodes via the relatively slow E X B drift caused by the space-charge electric field[9]. This places an upper limit of 10-9 Torr on the background gas pressure. One additional requirement is that the vacuum chamber be smooth in the direction of the beam so that no potential wells can occur which are hidden from the clearing electrodes. Fortunately, this is consistent with the requirement for low impedance of the beam pipe. 5-7 5.2 Stabilizing Systems 5.2.1 Emittance Dilution in the Collector Emittance Dilution Procedure The longitudinal emittance is to be increased by a factor 4 in the Collector, from 0.048 eVs to 0.192 eVs. The technique is to excite (one or several) coherent multipolar perturbations of the ensemble in longitudinal phase-space, and rely upon filamentation to spread and smear the ensemble. Let </J be the individual particle phase, and n" be the synchrotron frequency. Including the effect of the high frequency cavity (HFC), the basic equation of motion is if> + n~(sin</J + psin[N</J - ~(t)]) = o. The free variables, used to optimise the growth rate, are: p, the perturbation amplitude; N, the multiple of the fundamental radio frequency; and ~(t), the phase modulation function. Firstly, N / h must be an integer plus either zero or !, in order to make the emittance blow-up of each bunch identical. Secondly, N x h must be an integer so that the perturbation does not phase-slip with respect to the rf-buckets. Since h = 225 is odd, the fractional part of N / h must be zero. This has the unfortunate consequence that there is some beam loading, since the cavity resonance lies on a harmonic of the RF (where there is a strong beam component). Ideally, N should be large (I"V 20) so that the macroscopic structure imposed on the bunch is of short wavelength. Contrarily, for a given value of p, the strength of the perturba-tion varies as l/VN; favouring lower N. Two cases have been studied in detail: N=20 and N=14. Though the diluted emittance quality is superior at N=20, there are strong practical/technical arguments in favour of N=14. A wide-bore 1.2 GHz (N=20) cavity would likely have low Q and shunt resistance, and high power requirement. A narrow bore cavity would imply a local constriction in the beam pipe which would present a problem for the vacuum pumping system. Further, I'V 850 MHz is in the TV-band, so that standard commercial technology could be adopted. Thus N = 14 has been chosen for the carrier frequency. Following Kats[10] the modulation function is : ~(t) = a sin(wdt) - (}o with free parameters a, the phase modulation amplitude; Wd, the phase modulation frequency ; and (}o equal to zero or 7r /2. Systematic numerical[ll] and theoretical studies[12] have shown that the growth rates are highly dependent on the choice of Wd and a (and less so ( 0 ) since the 5-8 ---.. u Q) (f) > Q) Q) u c a ~ ~ 'E w .2~ .20 .15 .10 .05 emittance growth is a resonant process. The bunch is resonant at integer multiples of the synchrotron frequency. Strong resonances are excited at Wd = 4w" and 8ws , with a ranging from 1.8 to 1r. The synchrotron frequency varies with oscillation amplitude, and so emittance growth pulls the bunch off resonance. This accounts for the experimentally observed Vi growth law. However, by sweeping the modulation frequency to follow the average (ws ) a driven linear growth can be maintained. The differential growth rate between the core and periphery of the bunch also accounts for the production of an halo. Figure 5.2.1 shows a typical phase-space plot and emittance growth law for the modulation-induced emittance blow-up. wd= O.2 1093 = 40, Dongiful:linnl Ql1nSf> §pnrf> 11.1 3.7 -3.7 -11. 1 .OO~--~----~----.----.----.----.-----r -18.6 -t-,........,.---.-,........,--.---r--lr-r--r-=::?-r......-;::::""--'-,........,.---.-.,..-.--.---r--l-+ o 2000 4000 6000 8000 10000 12000 14000 Time (,uusec) -200 -120 Time 14.20 (rna) -40 40 120 RF -phose (deg) Energy(r) 3.00 (GeV) RevlI Fig.5.2.1 : Collector Longitudinal Emittance Blow-Up The perturbation amplitude p should be small, to reduce the power requirements. Keeping p small also minimises the uncontrolled emittance growth which arises when the HFC is suddenly turned on. A simple empirical scaling law[13] applied to the Collector suggests p ~ 0.12 at N = 14. Tracking studies, however, indicate p = 0.21, unless the excitation frequency is swept, or dual-frequency excitation is used, or the main rf system voltage is reduced to increase the effect of the high harmonic perturbation. Cavity Specification Take N = 14, P = 0.15 and wide-band phase modulation to pass side-bands up to ±4 X Wd from the carrier frequency (850 MHz). Adopting Wd = 4i1s gives a bandwidth of 0.26 MHz. This implies a quality factor Q = 850/0.26 = 3300. Assuming R/Q ~ 100 i1 and two cells with V = 140 kV each one finds a total of 60 kW of rf power to be delivered by a single klystron feeding a Magic-Tee[14] which is quite manageable. The substantial beam induced signals are absorbed in the resistive load of the Magic-Tee and do not pass on to the klystron. Nevertheless, compensation for beam loading (up to 34 kV) has still to be provided by the klystron, and this will require a feed-forward system. 5-9 200 3858 5.2.2 Longitudinal Coupled-Bunch Damper Booster Active Damper The most troublesome Booster cavity parasitic occurs at I'V 70 MHz and has RIQ = 25ft This higher order mode (HOM) will drive the n = 22,23; m = 1,2 coupled-bunch modes immediately after injection, and likely others since the revolution frequency varies by 27% during acceleration. The amplification of any initial phase oscillation Xo would be of order 106 for the dipole mode. The quadrupole mode must be damped as the design intensity is about a factor 3 above the Landau damping threshold. The tolerable amplification depends on the seeding, due to injection errors and bunch-to-bunch variation induced by periodic transients, but experience at the CERN PS Booster indicates X I Xo ~ 5. It is therefore essential to include a coupled-bunch feedback system. When a higher order mode (HOM) is crossed by a revolution harmonic, the integrated growth for coupled-bunch instability depends only on the RIQ, which is a geometric factor. However, the peak growth rate during crossing is proportional to the shunt impedance Rs. Consequently, the Rs of the HOM's must be reduced, by a higher-order-mode damper, to the point that the growth rate is smaller than the synchrotron frequency; so the instability can be cured by active damping. (HOM damping techniques are discussed in section 4.3.5.) For example, the ring impedance at the f = 70 MHz mode has to be reduced to Rmax = 16 kn so that the rate does not excede wallO. The Feedback System The voltage requirements for the active damper can be estimated[5] assuming an initial phase error /:}.</> using: (4) A realistic figure is 6.</> I'V 10 which gives Vfb ~ 2 k V. It is appreciated that Vfb is a noise-like signal, and by no means sinusoidal. In order to damp all possible dipole coupled-bunch modes, a bunch-to-bunch damping scheme is adopted. Digital processing forms the time derivative of the phase error, and this signal modulates a multiple of the revolution frequency, so that the voltage VI is applied with a one-turn delay. A similar system would be used for the quadrupole modes, with peak detection replacing the phase detector. One may take advantage of the one-tum-delay to incorporate more sophisticated damping algorithms. Ordinarily, the longitudinal kicks are proportional to the phase error, which gives an exponential decay damping. If the errors are beyond a pre-defined threshold, as 5-10 may happen at injection, then the maximum possible kick is applied. This avoids excessive filamentation due to non-linearity. This technique known as bang-bang[22] damping gives a linear damping. When the error is sufficiently reduced, the proportional algorithm takes over. The operating frequency range of the feedback must be such that the bunches see a quasi-constant voltage. The suggested frequency band corresponds to p = 2 (n = 90 to 112), a compromise between proper damping of the dipole mode on long bunches at injection (low frequency), and an acceptable size (high frequency preferred). Because the revolution frequency varies, the bandwidth must span 92-150 MHz. A low power requirement favours the use of a directional coupler over an untuned cavity. At synchronism Vj = 2 X Viine and the power feeding a single 50 n line is 62 kW; but using a four gap structure this could be reduced to I'V 23 kW. The four-gap directional coupler would be 1.7 m long, excluding flanges. There is a substantial beam-harmonic at p = 2, and this will produce some beam-loading of the active-damper which must be compensated either directly by a local feed-forward loop, or through the main rf-system. Driver Longitudinal Damper The specification of the Driver active damper depends on the Rand Q of the HOMs, the degree to which HOMs track In, the number of revolution harmonics which lie on or passs through HOMs, and the part of the acceleration cycle during which this occurs. Consequently, the details are best left until the dangerous oscillation modes are diag-nosed. Nevertheless, assuming the beam interation with the HOMs is similar to that in the Booster, the likely requirements can be outlined. Firstly, the HOM shunt resistance (R) must be damped to ~ 1 kn per cavity, of which there are 18, to reduce the growth rate to 110ws. With a phase-detector error of I'V 10 the feedback voltage is in the range Vfb = 5 - 7 k V per turn. Since the bunches are 2000 long at injection, the feedback central frequency is 2~ X In, and the bandwidth 35 MHz if all modes (n = 1, ... 224) must be damped. To reduce the power requirements to acceptable levels, the damper is divided into several units each developing up to 1.5 kV. This is at the expense of increasing the ring impedance. A satisfactory solution uses five ),,/2 drift-tube resonators of length 1.25 m. The twin-gap resonators have 50 n characteristic impedance, and each unit is driven by a distributed amplifier capable of delivering up to 12 kW. Beam induced signals will have to be compensated, and this will add to the power requirement. Extender Active Damper If a longitudinal damper is required in the Extender, its implementation would be easier than in the preceding ring. The short bunches, only 540 long, allow a high frequency band Pin + nlo with p = 3, and n = 0, ... 224. This gives a central frequency of 203 MHz and 5-11 bandwidth 31 MHz. The feedback voltage is Vfb ~ 2 kV per turn if the HOMs are damped to 1 kf! per cavity. 5.2.3 Transverse Active Dampers Introduction Nearly all recent storage rings and accelerators have been equipped with resistive transverse feedback systems[15,16,17,18,19]. These systems are designed to damp the lowest mode of the head-tail instability[20]. In some cases these systems have become indispensable to the proper operation of the machine, as well as being an invaluable diagnostic tool. The essential components of a feedback loop consists of a pick-up electrode, a beam position detector, a delay unit, a power amplifier and a deflection electrode (or kicker). Normally, for convenience, the pick-up and kicker are located in the same region of the machine with an odd multiple of 7r /2 betatron phase advance between them. The main parameters for consideration in the feedback system are the following: Gain The gain is directly related to the required damping rate (or coefficient aFB) for a resistive feedback system. The required rate depends on the intended beam-current and the wall impedance; and a margin must be allowed to avoid emittance increase from filamentation. The maximum gain is limited by loop stability, and the noise present in the detection of the beam position which may 'heat' the beam. Bandwidth The bandwidth must be sufficient to treat each of the bunches separately, so the bandwidth approaches h x 10/2, where 10 is the revolution frequency and h the harmonic number. Feedback Damping Rate It is widely recognised that it is inadequate to provide damping merely to combat coherent instability[21][22]. The damping rates must be sufficiently high to damp an oscillation in a time that is short compared with the decoherence time of the beam, else the coherent signal from the beam is obscured by filamentation and the active damper becomes 'blind'. Consequently, the feedback damping rate is the sum of two components: aFB 2: az + ac. Here az is the damping rate to cancel instabilities, and ac is the damping rate to avoid fila-mentation. The tables of section 5.1, for resistive wall effect and kicker magnet impedance, 5-12 give the instability growth rates (az). The appropriate value[21] of ac depends on the allowable emittance increase percentage (.,\ = 2D..u / u), the ratio of the initial coherent os-cillation amplitude to the half-beam size (ao/u), and the local coherent frequency spread D..w = 7r/oD..Q[coh] (which is half the shift). (ao) vA ac=D..w - ~u 4 In 2 (5) The expected injection errors, if undamped, will give rise to about 10% emittance increase. The damping system will reduce this to .,\ = 2%, so that ac = 2.12 x D..w x (ao/u). Deflecting Field The maximum deflecting field is proportional to the product of absolute gain (a/ 10) and the maximum detected beam deviation (Upu) and inversely proportional to the beta functions at the pick-up and kicker. During injection, large values of Upu are unavoidable and should not saturate the feedback system. For a resistive feedback system to give a damping time constant of TFB (= 1/ aFB) for coherent motion, the kicker must produce a deflecting electric field (or equivalent magnetic field): (6) Using equation (2) and the required damping rate and the expected injection errors Upu gives the approximate integrated electric field required (Eu . 10)' Tables 5.2.1 and 5.2.2 summarise the relevant parameters for this calculation and its results. Injection errors In all rings except D and E, injection and/or extraction take place vertically. In the Driver ejection is horizontal, as is injection into the Extender. There are two sorts of injection error: (i) static errors due to alignment and field errors (ii) dynamic errors due to firing of the injection kickers. The kicker errors dominate over the static errors, and determine the required peak damping rates and peak power. The direction orthogonal to the injection plane has only small static errors, and thus low power requirements. Static Errors Since the resolution of the position detectors is 0.1 mm, in the transfer line and injection straights, it is reasonable to assume that the static errors can be routinely tuned down to 0.2 mm or better. Dynamic Errors During a transfer the beam is kicked twice: out of one ring, and in to the next. Each kick contributes errors, which manifest themselves as position errors .,\/4 downstream. The kicker pulse is square in time with distortions at the head and tail. Those 5-13 bunches at the centre of the pulse may suffer kick errors of ±1 %, but for those close to the pulse ends the errors are greater. Including extraction 'precursors' and injection 'ringing' there is a total variability which can be 2-3% of the full kick height for the few bunches which encounter the short-lived kicker errors. Consequently, the expected dynamic errors are up to 3% of the beam stay-clear aperture, and if undamped, will give rise to roughly 10% emittance increase. The damping system will reduce this to 2% per ring. Table 5.2.1 Summary of Horizontal Dampers Ring A B C D E Inject error (mm) 0.2 0.2 0.2 0.2 0.8 ~QH[coh] (%) 1.21 0.77 0.32 0.15 < 0.01 (f3H) (m) 15 15 30 35 30 ac (8-1) 750 520 140 55 5.4 Peak Vc 7 5 7 3 6 Peak Vz 35 20 170 255 3300 Peak VTot 40 25 180 260 3305 CW VTot < 10 <5 < 40 < 50 < 200 Quantity 1 1 1 1 3 Length (m) 1.6 1.5 1.5 1.5 1.4 Impedance n 25 25 50 50 50 Peak Power 2 x (20 W) 2 x (10 W) 2 x (165 W) 2 x (340 W) 6 x (5 kW) CW Power 2 X (1 W) 2 x (1 W) 2 x (10 W) 2 x (15 W) 6 x (20 W) The direction orthogonal to the injection plane has only small static errors, and thus low power requirements. This distinction is clearly seen in the tables. In order to reduce the power levels to acceptable levels (less than 5 kW peak), it is decided to divide the vertical deflectors in C,D and the horizontal deflector in E into three units, each composed of a double strip-line. Similarly we are motivated to split the vertical B ring damper into 2 units. It is important to note that the voltages and power levels quoted in Tables 5.2.1 and 5.2.2 are based on the maximum growth rates, which arise when the chromaticity is corrected. If the Driver lattice is run with the natural chromaticity, the integrated vertical kick becomes 1.13 kV; in this case only 2 damper units would be required. If the Extender lattice is run with natural chromaticity, appropriate to the bunched mode, the integrated horizontal kick becomes only 140 Volts; in which case only 1 damper unit is required. 5-14 Table 5.2.2 Summary of Vertical Dampers Ring A B C D E Inject error (mm) 0.2 2.46 1.9 2.3 0.2 ~Qv[cohl (%) 2.03 1.54 1.57 0.58 0.02 (!3v) (m) 15 15 25 30 30 et.a (8-1) 2000 12300 1780 1060 4 Peak Va 19 930 690 390 2 Peak Vz 35 180 1600 2300 830 Peak VTot 55 1110 2290 2690 830 CW VTot < 12 < 15 < 50 < 60 < 200 Quantity 1 2 3 3 1 Length (m) 1.6 1.5 1.5 1.5 1.4 Impedance n 25 50 50 50 50 Peak Power 2 X (30 W) 4 X (1.5 kW) 6 X (3 kW) 6 X (4 kW) 2 X (4 kW) CW Power 2 X (3 W) 4 X (1 W) 6 X (2 W) 6 X (2 W) 2 X (200 W) Hardware The transverse feedback system hardware can be divided into three subsytems: (i) the low-level elctronics, (ii) the power amplifiers and (iii) the sets of pick-up and feedback electrodes. Electronics The processing electronics consists of up to 45 (225 for C,D,E) parallel chan-nels clocked at the radio-frequency. Each channel filter works at h times lower sampling frequency and can therefore be easier to realize. Clocking from the rf ensures that the response of the filters follow the variations of the revolution frequency in Band D rings. Each bunch vertical and horizontal position is sampled once per turn; the corresponding level is held until the next turn and used to amplitude modulate an rf carrier (with some appropriate delay before application to the power amplifiers for feedback). The processing electronics has also to remove the dc component due to the non-zero closed orbit in A,C,E, and modulation components due to equilibrium orbit shifts in Band D. Further, power supply ripple and synchrotron motion has to be rejected[15]. All those signals whose frequency components are close to multiples of the revolution frequency will be rejected with notch filters. Good quality coaxial lines are used to carry the pick-up signals. Amplifiers The high power amplifiers in rings C,D and E are push-pull wide-band class-A, capable of delivering 5 kW peak into a 50 n load, and capable of withstanding a fair amount of beam induced rf. The requirements for low power amplifiers, in rings A and B, 5-15 are very modest. The very small cw power arises merely from errors in the position detection system, which have been estimated to be one half the resolution of the pick-ups. Though the instability growth rates scale as II'b, so does the electric rigidity and so the cw power does not drop-off with beam energy in the Booster and Driver. Pick-ups The horizontal pick-ups are placed at points of zero dispersion. This avoids coupling to a synchro-betatron resonance, and helps suppress common-mode signals from the synchrotron motion. Where possible the signals are taken from pick-ups placed in the vicinity of /3. Deflectors Each damper consists of a double stripline, that is two transmission lines oper-ated push-pull. One could use single strip-lines for the A,B,C,D horizontal deflectors, but it was decided to keep open the option for quadrupole mode damping; and the two lines, symmetrical inside the vacuum tank, will improve the geometry of the deflecting fields and reduce the cross-talk between damping of the vertical and horizontal directions. Since two strip-lines are used for damping horizontal motion in rings A-B, it was decided to reduce the characterist ic impedance to 25 n in order to reduce the beam-loading. Ideally each deflector horizontal or vertical should be placed at the local maximum of the respective beta function (f3H or f3v). However, if lack of space dictates otherwise it is conceivable to place the low-power deflectors (horizontal in B,C,D) in the same vacuum tank as one of the high-power units (vertical in B,C,D). The location at a non-optimal beta-function can be easily compensated by higher line voltage and higher power, smce these are already so modest; the voltage is multiplied by 2 and the power by 4. One should be cautious when estimating the cooling requirements, for the peak power is delivered only until the injection errors are eliminated; after that time the deflectors work at the noise-level which may be an order of magnitude smaller power consumption. Further there is the u.ncertainty over chromaticity correction. Consequently, it is not yet determined whether water cooling will be required. 5-16 5.3 Beam Instrumentation 5.3.1 Introduction Both peak beam intensity and rf structure are similar to those in existing machines; no fundamentally new beam behaviour is expected and therefore the diagnostic hardware should be similar to that used elsewhere. However, the higher average intensity means that there will be a greater concern with beam quality and loss monitoring than with earlier machines. The functions of the beam instrumentation system are to measure the position of a bunch and its distribution in both transverse and longitudinal planes and also the intensity of the retained and lost beam. For beam physics experiments and at higher intensity where instabilities may arise some of these parameters will be measured at GHz frequencies. 5.3.2 Modes of Operation The 100 /-LA reference design has 4 pC in 2.5 ns long bunches quasi cw transferred between injector cyclotron and accumulator ring. In the Accumulator ring the bunch population increases from 4 pC to 50 nC and the bunch lengthens from 5 to 12 ns; see section 2.2.1 and Fig. 2.2.6. The beam remains bunched throughout subsequent rings and transfer lines with the possible exception of the E ring and slow extraction line. A beam-free period ~ 1 ms will be maintained between ring cycles for equipment reset and calibration. The intensity of the polarized beam is expected to be an order of magnitude smaller than the unpolarized intensity. In addition, early operation and some beam physics studies will be at low cw current. It seems reasonable to require that each function be measurable over a dynamic range of at least 1-100 in the IA line and 1-1000 in the rings. Early commissioning at even lower currents may require, temporarily, more sensitive devices. The A ring is a special case where a dynamic range of 1-20,000 may be required over the long term. Simple modifications to the drive of the existing pulser unit for the cyclotron ion source should permit flexible macro-pulsing and give control over bunch populations by varying accumulation time; complete pulse trains could be eliminated. The existing 5: 1 selector will eliminate 4 out of 5 bunches in both injection lines and rings. By these means the KAON Factory may be commissioned with high peak intensity for low average current. The relatively small emittance of the H- beam together with injection by painting permit the controlled population of a variety of distributions, some bright with low emittance and high tune shift, others more diffuse. 5-17 5.3.3 Instrumentation for Synchrotrons and Storage Rings Hardware requirements have been reviewed[24] based on the instrumentation of similar machines[25]. We expect that a suitable instrumentation inventory would contain the following items: • Position Four beam-position sensors per betatron oscillation will be placed at envelope max-ima. Single-frequency AM/PM processing will be used with a data acquisition system gated to record trajectories for several consecutive turns in addition to perform-ing averaging for closed orbit determination. Heterodyning would be necessary for the Booster synchrotron. The trajectory information will be useful for setting up matched injection and, in conjunction with a diagnostic kicker, for checking lattice functions. Two sets of wide-band position monitors will be provided in each ring at locations with low horizontal dispersion (x), high momentum resolution, (rJi ..jfJ), and at a position sensitive to vertical oscillations (y). One set would be of the wall current type, flat between rv 1 MHz and a few GHz. The other would be of the split cylinder (or box) capacitive type with a flat response between tens of kHz to tens of MHz; two additional monitors (x', y') would be at locations with about 90 deg betatron phase advance from x and y. • Distribution Scintillator screens and a video system will be available for initial, low intensity, commissioning. For more precise work flying wire scanners or fast extraction on to a harp are being considered. Residual gas parallel-field ionization monitors are being considered for those situations where continuous observation may be desirable. Good emittance matching between transfer lines and rings is important for high intensity machines. Devices[26] to detect quadrupole or beam width oscillations will be included. Moveable beam scraping jaws will be used to measure the beam halo. • Intensity No storage longer than 0.1 s is envisaged for regular operation; storage of order 1 s may be used for beam physics experiments. Consequently standard (ac) toroids are acceptable for current monitoring and transmission checks. One wall current monitor per ring, triggered from the rf system, is envisaged for the study of longitudinal instabilities within a bunch. • Timing While beam phase signals may be derived from other monitors, e.g., position sensors, it may be more convenient to have dedicated annular pickups near rf cavities or other systems requiring feedback or feedforward. The site-clock and timing philosophy is described in Section 9 (Central Control System). 5-18 • Diagnostic Kickers and Dampers Fast kickers capable of providing a deflection in horizontal and vertical planes of up to 10% of the local beam divergence will be provided for diagnostic purposes in each ring. The pulse duration will be less than one turn. Coherent transverse (coupled-bunch) motion detected by the wide-band position monitors would be damped by impulses imparted with an approximately one turn delay and 90 deg phase shift by a stripline deflector with high-power driver. The Booster delay line must track the frequency swing. • Loss It is proposed to install segmented ion chambers in the tunnels; one above, one below and, maybe, one in the mid-plane of the rings to assign loss signal to the appropriate ring. These would display loss histograms by geographic region; and inbrmation, binned in time with a resolution of milliseconds, could be retrieved. They would be supplemented by fast loss monitors near injection, extraction and collimation regions and other regions to be determined experimentally in order to measure loss as a function of bunch number or rf phase. These latter would use a time-to-digital converter triggered by beam pickup or rf system. • Polarization Space is reserved within the synchrotrons for polarimeters of the type presently used at TRIUMF scaled up to the appropriate momentum. These measure asymmetries in scattering from hydrogen using nuclear detectors in coincidence. Polarimeters associated with the experimental areas would service the E ring. Some beam instrumentation parameters are summarized in Table 5.3.1. 5-19 Table 5.3.1A Summary of Beam Instrumentation Device type Accumulator Booster Collector Driver Extender COD Beam Position 48 48 134 134 132 Length (em) 20 20 20 20 20 Sense at RF 46.1 46-61 60.7 60-62 62.5 LF cut-off (MHz) /0 = 1.0 /0 = 1.0 5/0 = 1.35 5/0 = 1.35 5/0 = 1.36 Diagnostic Kicker 2 2 2 2 2 Length (em) 50 100 85 180 80 rise/fall (ns) 217 165 741 720 720 Duration (J.Ls) ~ 0.976 0.74 - 0.98 ~ 3.7 ~ 3.7 ~ 3.68 LF Wide band Posn. 5 5 5 5 5 Length (em) 20 20 20 20 20 LF cut-off (kHz) q/o = 200 q/o ~ 200 q/o = 54 q/o ~ 54 q/o = 54 HF cut-off (MHz) 2: 1.02 2: 1.35 2: 1.39 2: 1.39 2: 1.39 HF Wide band Posn. 2 2 2 2 2 Length (em) 20 20 20 20 20 LF cut-off (MHz) ~ 1.02 ~ 1.02 ~ 0.270 ~ 0.270 ~ 0.272 HF cut-off (GHz) 2: 2.5 2: 2.5 2: 2.7 2: 3.0 2: 3.9 Quadrupole Mon. 2 2 2 2 2 Length (em) 20 20 20 20 20 Profile Monitor 3 3 3 3 3 Length (em) < 40 < 40 < 40 < 40 < 40 Halo X&Y Scraper 1 1 1 1 1 Length (em) 10 10 10 10 10 LF Current Monitor 1 1 1 1 1 Length (em) 15 15 15 15 15 LF cut-off (Hz) 100 100 20 20 10 HF cut-off (kHz) 2: 10 2: 10 2: 10 2: 10 2: 10 HF Current Monitor 1 1 1 1 1 Length (em) 30 30 30 35 35 LF cut-off (MHz) /0 = 1.0 /0 = 1.0 /0 = 0.270 /0 = 0.270 /0 = 0.272 HF cut-off (GHz) 3-4 3-4 3-4 4-5 4-5 Phase probe 3 12 8 12 4 Length (em) 10 10 10 10 10 Polarimeter - 1 - 1 -Length (m) - 1.2 - 1.5 -5-20 Table 5.3.1B Summary of Beam Instrumentation Device type Slow loss Monitor Length Time-slice (ms) Fast loss Monitor Length Time-slice (ns) Longitudinal Damper Length (m) Number of gaps kV per gap H Transverse Damper V Transverse Damper Mechanical Length (m) Definitions Accumulator Booster Collector Driver Extender 48 48 128 128 128 1 24 ",1 1 1 1.6 1 24 ",1 2 1.5 4 ",1 1 2 1.5 1 64 ",1 1 3 1.5 1 64 ",1 5 1.5 10 ,....,1 1 3 1.5 1 64 ",1 3 1 1.4 The number of COD monitors is the total for both planes Hand V. Each monitor measures displacement in only one plane, either H or V. The number of LF Position monitors is the total for both planes Hand V. Each monitor measures displacement in only one plane, either H or V. The LF cut-off is the fractional part of the betatron tune multiplied by the revolution frequency. The Profile Monitors are either of the fast-wire or residual gas ionization type. Space has been allowed for photo-multiplier tubes downstream of the flying wire, if this is used. The single Halo Scraper works in both planes Hand V. There are 4 motorised jaws, and the step-size is ~ 0.1 mm .. Each quadrupole (or non-intercepting width) monitor measures in only one plane, either H or V. The LF Current Monitor has a resolution of < 1%, and range> 4 decades. The Loss (or spill) monitors are placed away from the beam pipe and so are accorded zero-length in the table. The Fast Loss monitor are place at inject/eject etc .. The purpose of Phase Probe is two-fold: for RF cavity/beam synchronisation, and for ease in implementing the feed-forward loops for beam-loading compensation. The probes are placed close to the RF cavities. Each transverse damper consists of a double striplinej each line is placed diametrically opposite its partner in the vacuum tank. It is not essential to place horizontal and vertical dampers in separate vacuum tanks, so that a total of 3 damper units utilize only 3 m of floor-space. 5-21 5.3.4 Instrumentation for Transfer Lines The profile monitors in the transfer lines between rings would be secondary emission grids, supplemented by scintillator screens. Beam position monitors, halo detecting jaws and loss monitors would be similar to those in the rings. The injection line, lA, with its almost continuous train of bunches containing 4 pC H- ions will use equipment similar to that developed for the TRIUMF 100 J.lA proton lines. This consists of 0.45 m stripline position sensors, fast (capacitive) and slow (toroid) intensity monitors (the latter for calibration) and 1 Hz scanning-wire profile monitors. Arrange-ments will be made for tomographic emittance measurements in an achromatic section. The H- momentum distribution will be measured at an intermediate dispersed focus. A capacitive pickup is required to synchronise the A-ring rf. The longitudinal distribution may be obtained by integrating the capacitive pick-up signal or from a TDC started from a scattered proton and stopped from the capacitive pick-up pulse. The detectors associated with the polarimeter would provide sufficient resolution for the former. Beam intercepting jaws in the achromatic and dispersed sections will be used, not to trim the emittan Je, but to initiate a trip in the event of an unacceptable excursion in H- beam parameters. For this the jaws need not be solid but could be a wire grid. Collection of electrons stripped by intercepting monitors gives a much stronger signal than secondary emission. A special effort will be made to collect those from the stripping foil. It may prove necessary to bias the latter to prevent (e-p) instabilities. Similar profile monitors and halo jaws, but exploiting secondary emission, would be used in the 30 Ge V slow extraction line. Inductive pick-ups would use the > 1 ms beam gap for reset. 5.3.5 Procedures and Measurements • Linear Optics The diagnostic kickers and wide-band position sensors will be used to measure the betatron tune. RF knockout techniques may be possible in the storage rings espe-cially in the case of an extended store. The transverse lattice functions, f3( s), will be measured using a fast kick followed by trajectory acquisition using the BPMs. Dispersion functions 1]( s) will be obtained from a measurement of closed orbit as a function of beam momentum. The latter is altered by shifting the accelerating rf frequency. Chromaticity will be obtained from a measurement of tune as a function of central momentum; a beam with a special low momentum width (8p/p) beam would improve the resolution . • Non-Linear Optics Schottky noise techniques will probably not be appropriate for the intense bunched beams and fast-cycling synchrotrons of the KAON Factory. Mapping of phase space to reveal magnetic non-linearities and imperfections would require a large diagnostic kick. It may be possible to use the extraction kicker with a slow fall. Transverse phase 5-22 space in the storage rings and in the synchrotrons at injection may be mapped by off-centre injection of a low emittance beam. Fast loss monitors, I"'V 1 turn response time, would assist in identifying troublesome resonances. The signature may be enhanced by moving the tune closer to a suspected resonance. • Instabilities It is required to measure the head-tail mode number m of the transverse coherent instability, the coupled bunch mode number n and also the inter and intra-bunch longitudinal mode numbers. Since there will be a coherent signal time-and frequency-domain analysis is planned. Bunch lengths in the Driver approach 2 ns and the likely resonance frequencies require position-monitor bandwidth from 0.2 MHz to 5 GHz. The self-bunching of intense beams requires a longitudinal intensity monitor with response to 5 G Hz. 5.3.6 Potential Problem Areas At the present time, one major uncertainty is the effect of the fringe fields of the rapid cycling magnets on beam pick-up electrodes placed nearby. An experimental program to investigate this is planned. The rf frequency is higher than in many machines and the bunch length can shrink to 2 ns and the intrabunch structure will be smaller still. This demands that the sensitivity of some position and intensity monitors extend well into the GHz region. The polarimeters may require either novel targets for rings or 'ballistic' data acquisition systems for transfer lines to be developed. 5-23 References [1] D. Boussard, CERN LAB II/RF /Int./75-2. [2] A. Hofmann and F. Pedersen, IEEE Trans. Nucl. Sci. NS-26, 3526 (1979). [3] R. Baartman and C. Oram, Proc. 14th ICHEA, to be published (1989). [4] R. Baartman, Proc. Hadron Facility Tech. Workshop, p.210 (1987). [5] D.Boussard, Study of feedback system against coupled bunch instabilities zn the Booster, TRI-DN-K83. (Note in preparation.) [6] G. Nassibian, CERN/PS-BR/84-25. [7] The Physics and a Plan for a 45 Ge V Facility that Extends the High-Intensity Capa-bility in Nuclear and Particle Physics, LA-10720-MS (1986). [8] E. Keil, CERN 72-14. [9] B. Angerth, E. Fischer, and O. Grabner, Proc. 8th ICHEA, p.298 (1971). [10] J.Kats, Particle diffusion produced by HFC, p. 1281 IEEE PAC 1987. [11] S.R. Koscielniak, Longitudinal Emittance Dilution Studies for the Collector Ring, Proc. of the Summer Study on High Energy Physics in the 1990s, p.456. [12] V.Balandin, M.Dyachkov, E.Shaposhnikova, The Resonant Theory of Longitudinal Emittance Blow-up by Phase-modulated High Harmonic Cavities, to be published in Particle Accelerators. [13] D.Boussard, Scaling up of CERN PS controlled blow up to the Collector ring, TRI-DN-K81. [14] R.E. Collins, Foundations for Microwave Engineering, McGraw-Hill Book Co. p.283 (1966). [15] J .L. Pellegrin, Transverse Oscillations Damping with Wide-band Feedback on SPEAR, IEEE Trans. Nucl. Sci. Vol.22 No.3 p.1500 (1975). [16] E.Higgins et aI, The Fermilab Transverse Instability Active Damping System, IEEE Trans. Nucl. Sci. Vol.22 No.3 p.1473 (1975). [17] Carron, Myers, Thorndahl, The 50 MHz Transverse Feedback System in the CERN ISR, IEEE Trans. Nucl. Sci. Vol.24 No.3 p.1833 (1977). [18] R.Bossart et al, The Damper for the Transverse Instabilities of the SPS, IEEE Trans. Nucl. Sci. Vol.26 No.3 p.3284 (1979). [19] K.Willie, Damping of Coherent Transverse Oscillations in PETRA, IEEE Trans. Nucl. Sci. Vol-26 No.3 (1979). 5-24 [20] F. Sacherer, Transverse Bunch Beam Instabilities, IXth Internat. Conf. on High En-ergy Accelerators, Stanford 1974. [21] G.R. Lambertson: Feedback to suppress beam instabilities in future proton rings, IEEE Trans. Nuc. Sci. Vo1.3 No.5, p.1857 (1985). [22] Bossart, Louwerse, Mourier, Vos: Operation of the transverse feedback system at the CERN SPS, IEEE Trans. Nuc. Sci. p.763 (1987). [23] S.Myers, A first look at the requirements for transverse feedback for the LEP Main Ring, LEP Note 436 (1983). [24] S. Koscielniak, Diagnostic Activities and Requirements for the KA ON Factory, KPDS Design Note (to be published). [25] H. Koziol, Private Communication, 1989. [26] F. Pedersen and G. Rees, Private Communication, 1989. 5-25 BEAM PIPE AND VACUUM SYSTEM Chapter 6 6 BEAM PIPE AND VACUUM 6-1 6.1 Design Requirements . . . . . 6-1 6.2 Stainless Steel Vacuum Pipe Design 6-3 6.3 Ceramic Chamber Designs 6-6 6.3.1 Reference Design . 6-6 6.3.2 Integrated Design . 6-8 6.4 Pumps ........... 6-11 6 BEAM PIPE AND VACUUM 6.1 Design Requirements The design criteria for the vacuum and rf shield are conventional for large modern accel-erators, namely: the vacuum system emphasis is on reducing outgassing to the absolute minimum, and the rf shield is as smooth as reasonably possible to limit the longitudi-nal and transverse impedance seen by the circulating beam. The vacuum system must provide an enviroment in which the beam can be stably transported. Three effects have been considered that can cause beam loss, namely: ion desorption causing local pressure run-away, electrons captured by the beam causing the e - p instability, and beam-induced multipactoring causing excessive production of electrons. In addition some provisions must be made for avoiding the yet-to-be-understood instability in the Los Alamos Proton Stor-age Ring (PSR). The following four paragraphs review each of these effects and state the requirements they place on the beam pipe and vacuum system. As the beam circulates in the accelerator, interactions between the beam and residual gas create electron-ion pairs. The electrons can be captured by the beam if it is not bunched (as in the E ring) and when a threshold level of neutralisation is reached the beam is lost in a few turns due to the e - p instability [1]. These considerations give rise to a requirement for the E ring of a clearing electrode system and a vacuum of 10-9 Torr. The electrons produced by beam gas interactions can also cause problems due to beam-induced multipactoring. The electrons oscillate in the beam bunch and between bunches maintain their kinetic energy (about 200 e V) and strike the walls, creating secondaries. If more than one of these secondary electrons on average gets captured by the next bunch then a situation exists in which exponential growth of electrons in the beam path can potentially take place. This does not appear likely to occur in any of the rings, for two reasons: Firstly the time between beam buckets is long compared with the average time for an electron to strike the wall and return to the centre of the pipe. Secondly the electrons resident close to the centre of the beam with a low kinetic energy which are stored by the normal beam structure are mostly lost in the kicker gap. The ions formed by beam-gas interactions are repelled by the proton beam and strike the vacuum vessel walls with an energy of the same order as the beam potential (200 V), and cause gas desorption. If on average more than one gas molecule comes off the wall per incident ion, this produces a pressure rise that in the absence of pumps would be exponential. If the pumps either do not have adequate pumping speed or are so far apart that the local pumping speed is excessively limited by conductance, then the pressure rise will be exponential even with pumps. These considerations imply that we require 100 fls pumps at least every 5 m along the beam pipe. This is shown graphically in Fig. 6.1.1 6-1 Effect f pump spacIng - 3 ~ __________ ~ __ ~ ______ L-__________ L-________ -;-3 amps beam on -4 -8 at 60007CS onlN' /so 5m e t res 5metres rd -9 ~----------.----------.----------.-----------r o 5000 10000 15000 20000 tIme (sees) Figure 6.1.1: Effect of pump spacing on vacuum behavior at beam turn-on which shows the pressure rise between two 100 fls pumps at 5 m and 5.5 m assuming a desorption coefficient of 5.6, and a beam pipe of typical dimensions (100 mm x 150 mm). Clearly the exponential pressure rise with 5.5 m spacing cannot be tolerated. Fortu-nately the length and aperture of the various dipoles allows use of a conventional discrete pumping system of ion pumps sufficiently closely spaced. However to limit the secondary coefficient to below 5.6 in several kilometres of pipe, all vessels will have to be baked prior to insertion. [2] The PSR instability[3], while not fully understood, is probably due to electrons oscillating in the beam. It is strongly affected by the amount of beam between beam buckets. The most likely source of these electrons is from the stripper. Hence in the A ring provision is made for clearing electrodes in the stripper straight section, and care must be taken to maintain the bunched structure. This effect may be due to some combination of the above three effects but is included here as a seperate effect as it is unique to rings with stripper foils. Ion pump lifetime and ion desorption coefficient considerations yield a vacuum of 3 X 10-9 Torr, which meets all requirements except in the E ring, where as stated above 6-2 1 X 10-9 Torr is required for debunched beam operation. Beam pipes in the A,e, and E rings can be of conventional stainless steel (316LN) design. While ideally these would be internally coated with a low conductivity coating to reduce the resistive wall effect, it is found to be more cost effective to dynamically damp the instabilities caused by this increased resistivity. In the magnet elements in the B and D rings one would ideally want a beam pipe which is transparent to the cycling magnetic field (i.e., less than 100 Hz) but, on the other hand, opaque to frequencies of and above the particle revolution frequency (i.e., greater than 100 kHz), while at the same time not presenting a structure which provides a large impedance to the beam. Two solutions to this requirement are considered for the KAON Factory, firstly a design based on the separate pipe and shield used successfully at the Rutherford Appleton Laboratory ISIS accelerator, secondly one in which the ceramic pipe has the rf shield "painted" on to its inside surface. This second design, referred to as the integrated design, probably has slight vacuum and impedance advantages, and is a simpler engineering design, but a full length prototype has yet to be built and tested. Until this is achieved the Rutherford design will be treated as a reference design. Both designs use metal strips with discrete capacitors at one end to provide the rf shield. This rf shield provides a suitable impedance to the beam, as has been demonstrated by measurement at LAMPF and ISIS, and by calculation[4]. These capacitors also serve the purpose of "rf-bypass capacitors" to inhibit current flowing around the loop formed by the beam pipe, the ion pumps and the ion pump ground[5]. As in other accelerators in which beam loss is critical the beam pipe aperture will be rectangular, to allow full longitudinal beam loss collection. Transport of beam which is lost out of the stable accelerating bucket, is provided to the scraI?'ers. This beam is assumed to have betatron oscillations in both the horizontal and vertical planes with a random phase angle between these two oscillations. However in order to save aperture in the quadrupoles the chamber corners will be slightly rounded as the closed orbit distortions will not be a maximum in both planes simultaneously. The vacuum system will extend from the TRIUMF cyclotron, from which it will be pro-tected by a fast valve, to just prior to the production targets, which will be separated by vacuum windows and protected against by fast valves. In addition there will be a fast valve between the Booster and Driver complex. 6.2 Stainless Steel Vacuum Pipe Design The stainless steel beam pipe will be of rectangular cross-section, with slightly rounded corners. The beam pipe will "partially-follow" the beam envelope, that is the individual sections of pipe will be of a constant but minimum cross-section (so as to keep the magnet apertures to a minimum), and a smooth change in aperture will be provided by internal 6-3 sleeves in the bellows that connect the rigid sections (see Fig 6.2.1). This arrangement avoids the cost of building a truly beam-following pipe, while keeping the impedance seen by the beam down to an acceptable level[6]. As the cost of dipole magnet gap is about ;--------- 410.00mm ----------t 5" CONf"LA T f"LANCES \ I~ !--i---~i I . ·~i ----~-----I~ BERYLLIUM COPPER CONTACT rINCERS. (CUSTOM DESIGNED) INTERNAL SUPPORTS r'ON PU ... P I CONNECTION rLANCE r-'---,.....-........, 266mm ~ CERN STYLE rLANCES BEAMPIPE SUPPORT .... C. OR E RINC STAINLESS STEEL VACUUM PIPE 47S.00mm Figure 6.2.1: Pipe section with ion pump port, bellows and beam position monitor 6-4 $280 per mm per m, the thickness of the pipe walls has a significant impact on the to-tal cost of the accelerator. For this reason we have limited our design to high strength stainless steel. In order to limit the outgassing of the pipe to a minimum, and provide a well understood surface physics situation for secondary emission, we have chosen to follow the conventional route of pre-baking all components in a vacuum oven[2]. This baking procedure further limits our choice of stainless steel, and leaves us with only one practical steel 316LN. Recent developments in high vacuum technology suggest that gas discharge cleaning may be be a feasible alternative; this will require further study and potentially provides an avenue for cost savings. Figure 6.2.1 shows our design of a section with an in-ternal sleeve that incorporates a bellows, pumping port and beam position monitor(BPM). This standard design comes in 4 versions: • Bellows and Sleeve • Bellows, BPM, and Sleeve • Bellows, Horizontal BPM, pumping port, and Sleeve • Bellows, Vertical BPM, pumping port, and Sleeve Detailed designs of each version have been undertaken, and these versions have been used in all rings and transfer lines. In only a handful of cases are special designs required where the beam is too large for this device. Vacuum seal clamps between stainless steel pipes will be of the CERN clamp design, although in some instances it will be modified to make it easier to handle remotely. Seals will be aluminium with knife edge, as baking above 140°C is not anticipated. All ports will have bolt flanges of the conventional ISO conflat design, with knife edges and flat copper washer seals. This allows compatability to all conventional vacuum equipment such as pumps, valves and gauges, so any supplier can be used and replacement components rapidly procured: Table 6.2.1 lists the major equipment required for each of the rings, and transfer lines. 6-5 Table 6.2.1: Beam Pipe and Vacuum Equipment List ITEM IA A AB B BC C CD D DE E Extr. Ceramic Beam Pipes (m) 0 0 0 128 0 0 0 651 0 0 0 SS Beam Pipes (m) 188 216 32 88 202 1078 42 427 51 1102 908 No. Ion Pumps (1201/s) 53 57 4 3 50 244 10 244 10 244 181 No. Ion Pumps (2401/s) 0 12 0 93 0 30 0 72 0 24 18 No. Sublimation Pumps 1 4 0 15 0 10 0 24 0 8 6 (10001/s) No. Beam Pipe Gate Valves (rf) 0 4 0 4 0 8 0 8 0 8 0 No. Beam Pipe Gate Valves 3 2 2 1 3 0 3 0 2 1 6 (non-rf) No. Pumping Ports 3 6 1 5 3 8 2 8 1 9 6 No. High Vacuum Gauges 3 15 1 13 3 24 2 24 2 24 7 (Penning) No. Beam Position Monitors 53 51 4 48 50 136 10 136 10 128 181 6.3 Ceramic Chamber Designs The various reasons for choosing ceramic chambers over other thin metal designs is dis-cussed in a design note[7], along with a review of the options for ceramic pipe. In that design note various thin metal pipe designs are considered, but it is found that none give a sufficiently low eddy current heating and longitudinal impedance (Z/n). This is be-cause when the wall is thin enough to limit eddy current heating sufficiently, it is not thick enough for the beam not to "see" the surrounding structures of magnets and pipe supports. Hence, we limit our discussion here to two designs. 6.3.1 Reference Design This design is a minor variation on the Rutherford Appleton Laboratory design as used in ISIS. This arrangement is shown in Fig. 6.3.1. The only significant change is that the wires will be either Ti or 316LN stainless steel, as the stainless steel wires used in ISIS 6-6 are very slightly magnetic. In this design the rf cage in the dipoles is made of top and bottom walls of wires 1.5 mm in diameter spaced centre to centre by 3 mm, and vertical walls of thin metal sheets, all held in place by machinable ceramic (MACOR) holders. In quadrupoles Figure 6.3.1: Ceramic pipe and rf shield (Rutherford Appleton Laboratory ISIS design) the cage design is the same except all sides of the cage are made of wires. The pipe is made of 30 ern lengths of straight ceramic (97.6% alumina) pipe, machined at the ends and sides, and glazed together to give the required curvature. Each ceramic piece is isostatically pressed, machined in the green state, then fired vertically. The vertical firing of the ceramic pieces yields excellent tolerances, but limits the length of the ceramic piece, as otherwise gravity excessively elongates or compresses the piece during the plastic stage of firing, depending on whether the piece is hung or sits in the furnace. However, firing in this vertical orientation allows quite thin walls, and hence requires only minor subsequent milling. The approximately 30 ern-long machined ceramic sections are dowelled piece-to-piece alignment; about 15 sections are glazed together along with ceramic flanges on each end to form the beam pipe. This 1200°C firing in air to make the glazed joints, is done with the pipe vertical so that gravity provides the necessary compressive force. When stacking the components for the curved dipole chamber, care must be taken to find 6-7 stable arrangement in the furnace. To reduce the danger of the tower going unstable during stacking it is fired in two stages. First the lower half of the stack is fired and then the remaining components are added and the complete assembly fired. A vitreous glass is used in the joints which is radiation resistant, has good vacuum properties and matches the thermal expansion coefficient of the alumina. Connection to the next stainless steel vessel is made using are-usable Helioflex seal, the ceramic being glazed at the vacuum seal surface. This seal arrangement will be able to withstand direct a direct hit by the beam without melting the seal. A metal clamp like that used at ISIS will be used. The metal wires follow the beam profile since this allows room for the MAC OR holders; at the end of the pipe where the beam is biggest the wires just touch the walls. Hence the extra space required by the shield is just the wire thickness. At one end of the pipe the wires are connected electrically to the next stainless steel vessel, while at the other end they are connected through capacitors held in the adjacent bellows section in a picture-frame-shaped MACOR holder. The outgassing rate of the ceramic pipe made by this method is about ten times that of stainless steel. The MACOR holders are vacuum baked after machining; however the rf shield has an outgassing rate equal to the bare ceramic pipe. Hence the complete structure has double the bare ceramic pipe outgassing rate. To provide adequate pumping capacity, and a pump down time of less than 24 hours, 240 l/s ion pumps will be installed in the B and D rings. At present this design has been fully developed for the B ring only, but only minor thick-ening of the wires is required for the Dring. 6.3.2 Integrated Design In this design the interior of the ceramic pipe follows the beam, and has strips "painted" on it with a ceramic/silver mixture developed from thick film technology. This arrangement is shown diagrammatically in Fig. 6.3.2. The 1.3 m long ceramic pipes are fired horizontally, and for dipole segments the required curvature is obtained by deliberately allowing the piece to slump along its full length during firing. The isostatically pressed alumina (99%) pipe, follows the beam so as to allow maximum wall thickness. Machining fired ceramic is expensive, and thick side walls are required during manufacture to avoid excessive slumping during the plastic stage in firing. There is a rule of thumb that the ratio of the wall thickness to transverse span, shall not be less than 1 to 4 during firing otherwise excessive slumping of the walls will occur, producing a pipe with a "slumped rectangular" cross-section rather than the rectangular (within tolerance) cross-section desired. The pipe is made by glazing together a few 1.3 m lengths. To demonstrate this technology 7 pipes were made in 3 firings from random samples of ceramic powder, and tolerances obtained indicating that less than 0.7 mm steps will be present between sections at the glazed joints. After glazing the pipes together the 6-8 END SECTION Figure 6.3.2: Schematic of Ceramic Pipe Integrated Design 6-9 required strips are fired on the inside of the pipe. These strips are drawn on the inside of the pipe using a variant of a drafting pen that dispenses the required paste. This has been developed for our study so that it dispenses the full strip width and thickness in a single pass. The strips go from one end to the other and around the end on to the end face of the ceramic pipe. It has been demonstrated on a short (30 cm pipe) that this technology is capable of drawing over steps at the glazed joints of up to 1.5 mm however further development is required to stop subsequent thinning of the strips at the joints during the firing process. This thinning is probably due to the glass coming close to its melting point during this process. Connections between these strips and the adjacent stainless steel pipe are made using rf spring contacts held in a MACOR window-frame-shaped holder, as in the ISIS design. These holders are an part of the adjacent bellows/pump section and as in the ISIS design are removable with that section. The seals, flanges and clamps would be the same in this design as the reference design. Further study of the vacuum properties of this design are required, as the outgassing rate of this design is not known, but it is anticipated that the ceramic/metal paste will be a good high vacuum material. It assumed that this design like the reference design will require 240 i/ s pumps. The ceramic pipe will be monitored for temperature rises due to beam spill to protect it against excessive temperature gradients. On the sides of the outside of the pipe ce-ramic/metal resistance thermometers (RTD's) are fired (in the same firing as the internal strips) on the surface and silver/ceramic strips bring the connections to the end of the pipes. These thermometers primary purpose is to monitor the -beam pipe temperature and hence avoid excessive heating of strips, glazed joints, and metal seals. These temperature rises are a serious concern as the ISIS machine has much wider beams than the KAON Factory especially at injection where the stopping power is highest (x3 minimum ionising), and hence ISIS has not demonstrated that a jointed ceramic pipe can withstand the tem-perature gradients that may arise in our case, when the stored beam passes through the wall. At 30 GeV while the primary proton passing through the wall is minimum ionising, secondaries mainly from 7r0 decays increase the heat deposition to 3 times that calculated from simply assuming minimum ionising. Also due to the potentially small beam profile temperature gradients become large when the beam hits the wall, and very high stress levels are quickly reached if the beam is allowed to repeatedly strike the wall at a single location. Calculations indicate that only two consecutive stored beams can be dumped at one location, without causing stress levels to rise unacceptably[8]. However the RAL/ISIS group tested their pipes and joints with "bright" cyclotron beams[9] and these studies probably indicate that there is no significant risk, if sensible precautions are taken. As a secondary aspect, monitoring these temperatures will provide a sensitive indication of very small long term beam loss. 6-10 6.4 Pumps As stated in the introduction, the physical length of the dipoles and the size of the beam stay clear area are such as to allow installation of a distributed pumping system. In view of the requirement of limiting personnel radiation dose during maintenance of the pumping system, the ease of replacement of pumps in a discrete system is operationally significantly superior. The cost of incorporating a distributed pumping system into a ceramic pipe would be high. For these two reasons a discrete pumping system was chosen. Apart from close to the TRIUMF cyclotron an ultimate vacuum of 3 x 10-9 Torr is feasible, hence the life time of ion pumps would be in excess of 20 years. So an ion pump system was chosen as it gives the combination of high reliability, and low capital and operational costs. The vacuum system in each ring is divided into sectors using rf gate valves. The A and B rings are divided into 4 sectors, while the C, D, and E rings are divided into eight sectors. Portable pump down stations will be used to initially evacuate sectors. Figure 6.4.1 shows the design of the pumping station. Each pump-down port will have a residual gas analyser (RGA), used only during initial pump-down and leak checking. As at SLAC each ring will have sensitive RGA heads to monitor the vacuum on a continuous basis. Two such heads will be installed per ring, and one per transfer line. Each transfer line will have its own vacuum system separated from the ring vacuum system by gate valves, and have a pump-down port that uses the same portable pumping station as the rings. Portable helium leak detectors will be used, and helium will be transported in gas bottles with the leak detector. No helium gas distribution system will be installed. All sector valves will be pneumatically operated and controlled by the central vacuum control system. All pumping port valves will be manually operated with removable torque handles, with their status monitored by the central vacuum control system. This will be a programmable logic control (PLC) system, and will communicate to the central control system through a suitable port resident in a VME crate. Terminal ports to this system will be available at each pumping port and in each ion pump power supply gallery. Cryo-pumps will be installed in the cyclotron to A ring transfer line, between the cyclotron and the fast valve. Sublimation pumps with about 1000 lIs pumping speed will be installed in all kicker and RF cavity tanks, along with 240 lIs ion pumps. 6-11 UAIIUAl vALVE -NW 2~ (LEAK OH) RIlles 8/0 (E) tlOmm 1I0\(lJEIIT rOR SEAL REPLACEMElIT ... .lfiUAL "lL U[TAl VAlV[o ON IOD -- CONtROl CHASSIS Figure 6.4.1: Mobile Pumping Station 6-12 References [1] Chris Oram and Rick Baartman, Vacuum Requirement" for the KAON Factory TRI-DN-88-K6. [2] E. Jones and C. Oram, Design of the Vacuum System - Cost Estimating and Major Design Considerations TRI-DN-89-K35. [3] D. Neuffer et ai, Proceedings of the International Workshop on Hadron Facility Tech-nology, February 2-5 1987, Transverse Collective Instability in the PSR, p.237. [4] G. H. Rees, Longitudinal Impedance Estimate for the Distributed Capacitors of the SAIC rf Shield TRI-DN-89-K10. and C. Oram, Revised Estimate of the Longitudinal Impedance for the Distributed Capacitors for the SAIC rf Shield TRI-DN-89-17. [5] R. Cappi, RF Bypass on the Proton Synchrotron Vacuum Flanges to be published in 1989 PAC, Chicago, March 1989. [6] R. Baartman and C. Oram, Beam Pipe Impedances for the KA ON Factory TRI-DN-88-K13 [7] C. Oram, Beam Pipe for the A.C. Magnets at the KAON Factory TRI-DN-89-K56. [8] T. Hodges, Beam Heating of Vacuum Components TRI-DN-89-K92. [9] G. Rees, Private Communication, 1989. 6-13 ACTIVATION AND SHIELDING Chapter 7 7 ACTIVATION AND SHIELDING 7-1 7.1 Introduction ..... 7-1 7.2 Shield Requirements 7-1 7.3 Operational Constraints 7-4 7.4 Beam Collimators 7-5 7.5 Beam Dumps ... 7-8 7 ACTIVATION AND SHIELDING 7.1 Introduction The proton spill that can be tolerated from the beam transport lines, storage rings and accelerators is constrained by the unavoidable radiation dose to maintenance workers from the resultant residual radioactivation of the beam line and ring components. For a uni-form proton beam power loss of 1 W m-1 the saturated residual radiation field from the unshielded source produced in iron would be approximately 3 mSv h-1 at 1 m from such a line source. The residual radioactivity source produced in iron and other components is often self-shielded by a factor of order 10. Experience at other accelerators [1] and rough geometric and self-shielding considerations indicate actual residual radiation fields in the 0.3 mSv h-1 range for maintenance work on accelerator/storage ring components 1 day after the beam is turned off. The beam spill is not expected to be uniform; variations in spill distribution by a factor 10, both higher and lower, are expected but on the scale of a meter or less will be washed-out by the hadronic cascade spatial distribution, especially where gaps between beam line components increase the effective relaxation lengths. 7.2 Shield Requirements To define the operating· shield requirements the maximum beam spill at a 'bare' point was assumed to be 30 W (1 nA at 30 GeV); the operating radiation field at 10 m distance from such a point through 0.2 miron, 0.3 m concrete and 8 m backfill lateral shield is shown in Fig. 7.2.1 as a function of longitudinal position directly over and parallel to the beam direction. The estimate is based on the empirical M oyer Model expression for the total dose-equivalent rate, DE per proton, through lateral shields with the parameters listed below from previous experiments and experience [2]. The prescription has been extended to lower proton energies by discounting the dose-equivalent rate by the fraction of protons stopping by ionization before suffering a nuclear collision. DE Hr-2e-({311+~) H Ho· f· EO.s where: DE = Total dose-equivalent in Sv p-l Ho = 2.8 X 10-13 Sv p-l GeV-o.s f = non-elastic nuclear collision fraction before stopping by ionization E = proton energy, Ge V r = bombardment point to field point distance, metres (3 = empirical angular distribution parameter, = 2.3 radian-1 7-1 (J = angle between incident beam and field point direct ion, radians d = lateral shield thickness in the same units at ,\ below, ,\ = effective hadron dose-equivalent relaxation lengthl T ~ > (/) c:: Q) -0 c:: -c:: Q) 0 > ::s O'" lJJ I Q) II) 0 0 0 -~ = 147 gcm-2 or 0.20 m for iron = 117 gcm-2 or 0.50 m for concrete = 117 gcm-2 or 0.70 m for backfill. I I I I 60 - / / / 50 -, / / 40 c- / / / 30 c- / f / 20 - / / / 10 - / , / 0 1-/ I I I -8 -6 -4 -2 0 I / ...... \ I 2 Z,m I \ \ \ \ ~ \ \ \ \ \ \ \ ~ \ \ I 4 I I \ \ 1"_ .J 6 8 ------Figure 7.2.1: The total dose-equivalent rate through a lateral shield of 0.2 miron, 0.3 m concrete and 8 m of backfill above a point spilling 30 W (1 nA at 30 GeV) of beam power as a function of longitudinal position along the incident beam direction. By integrating this distribution over longitudinal position an operating dose-equivalent rate of 0.4 p,Sv h- l is obtained above a continuous line source losing beam at 30 W m- l . Based on a macro duty factor of 1/2 for the operation of the KAON facility, indicated by the present TRIUMF operating experience, the annual dose-equivalent for continuous occupation at the field point would be 2.2 mSv and for the normal occupational factor of 1500 hours per year, 0.6 mSv. The 30 mSv h- l residual radiation fields adjacent to the accelerator or beam line components suffering such 30 W m- l beam losses will constrain their acceptance to a few points around the facility at most. Such spills are the maxima that are likely to be feasible for even limited on-going maintenance requirements in the enclosed accelerator and transfer line tunnels. Maintenance around points of higher spill rate are considered feasible only if they are accessible from above with cranes to allow 1 half and tenth value attenuation thicknesses are 0.7 and 2.3 times relaxation length, respectively. 7-2 the installation of temporary shields for personnel protection. This operation is, however, expensive in job preparation time. The radioactivation of accelerator facilities has two components. The first and usually dom-inant one is from high energy hadron reactions that knock particles or nuclear fragments out of the nuclei they collide with in the "cooling" process. This mechanism dominates for neutrons until they fall below a few MeV kinetic energy at which point they lose the rest of their energy by nuclear elastic (billiard ball) scattering until they reach thermal or near thermal energies « 0.1 eV) where the neutron absorption cross sections for all nuclei (ex-cept 4He) become appreciable. Such captures constitute the other major radioactivation mechanism in accelerators because the high energy hadron reactions produce significant numbers of neutrons, most of which start with only a few MeV kinetic energy. This latter radioactivation component is amenable to reduction independent of source strength con-trol by capturing such neutrons in nuclei that lead to short lived or stable species or by elimination of elements in structural components that are particularly vulnerable to such radioactivation. The two particular problems that warrant some attention for accelerator facilities are sodium, leading to l5-hr half-life 24Na, in concrete enclosures and cobalt, leading to 5.3 year half-life 6OCo, as an uncontrolled trace element at the 100-200 ppm level in steel components. Sodium at the 2% concentration (by weight) level often found in concrete, usually controls the residual radioactivation field for the first few tens of hours after turning off the accelerators inside concrete enclosures. Such radioactivation can be reduced by a factor 10 by reducing the sodium fraction to the 0.2% level, which is usually feasible at only a small premium in the concrete cost, or by the addition of 0.1 % (by weight) boron to the concrete, which may have a significant cost premium. Because it involves suppression of the thermal neutron flux, the latter strategy does, however, have the advantage of suppressing other thermal neutron capture activation in the accelerator structures by the same factor. This latter effect can be obtained more efficiently by the addition of a coating containing 0.5 kg m-2 of boron, but this strategy has a much smaller effect on the neutron capture in the concrete where most of the neutrons are thermalized. The lowest sodium concretes available are made from limestone based aggregates rather than the more usual silica based aggregates. The limestone aggregates have the additional advantage that they suffer the least radioactivation by the high energy hadron reactions as can be seen in Fig. 7.2.2 which shows the decay curves for the radiation field from saturated residual (3 decay. This radioactivity is induced by a cascade hadron flux of 1.4xl06 cm-2s-I with an E-I spectrum extending to 30 GeV inside a uniformly bombarded, infinite medium of aluminum, iron, silica and limestone. The two dotted curves in Fig. 7.2.2 correspond to radioactivation from 0.2%(by weight) sodium in concrete and 0.02% (by weight) cobalt in steel from neutron capture in a thermal neutron flux of 0.9x 106 cm-2s-l • For a uniform beam spill of 1 W m- I in the main rings the average effective cascade hadron flux in the accelerator components is estimated to be lAx 106 cm-2s-I and in the tunnel walls a factor 7-3 100 less. The (uniform) thermal neutron flux in the main ring tunnel from evaporation neutrons produced by the hadron cascade is estimated to be 0.9x 106 cm-2 S-1 for the conditions expected for the tunnel walls, i.e., limestone aggregate but no boron additives in the structural concrete. Thus the field inside the tunnel from the neutron capture 24N a in the concrete can be read directly from Fig. 7.2.2. The estimated field from the hadron cascade induced radioactivity in the tunnel walls is lower by a factor of 100. The field from the accelerator structure, including the 6OCO neutron capture component, will be discounted by a geometry factor equal to the solid angle fraction of 47T" steradians sub tended by the radioactivated structure at the field point, always <1/2 and usually <10-1 • , ..c. 1000 r-----~~ __ > (f) ::i.. ~ 100 'U Q) LL c o -o 'U o 0:: 10 in steel \ 0.2 % No in concrete 0.1 1.0 10 lOa td , days Figure 7.2.2: The solid lines show the saturated, infinite medium, dose-equivalent rates from a cascade hadron flux of 1.4x106 cm-2s-1. The dotted lines show the saturated, infinite medium, rates from a thermal neutron flux of 0.9x106 cm-2s- 1• 7.3 Operational Constraints In view of the residual radioactivation problems that accrue from relatively low fractional beam loss rates, good beam diagnostics and spill control will be a requirement for the KAON facility. As well as the obvious requirement for a surfeit of phase space distribution measurement facilities at as many longitudinal positions as can be afforded the control 7-4 and parameter monitoring system must maintain complete archival records, at least for the short term, of all parameters that could affect the machine operation. To avoid the need for unnecessary repetition of defective operations the control parameter data analysis system should be closely coupled to the beam dynamics analysis facilities but more partic-ularly capable of identifying all significant changes in operating parameters from previous satisfactory operation. The loss of a single macropulse in any of the 50 Hz Accumulator/Booster stages or the 10 Hz Collector/Driver/Extender stages should not normally have traumatic consequences unless the beam spot is very small. This is not different from the case for our present 500 MeV facility for a 20 or 100 ms duration fault condition. For a uniform beam spot 1 cm2 the initial energy deposition rate for a 10 I-'C pulse of 30 GeV protons would be about 18 J g-l j the adiabatic temperature rise in iron would be only about 40°C. The hadronic cascade may increase the heat deposition rate but only by a factor of a few at most. Thus the loss of single full power pulses can be tolerated unless the beam spot sizes drop to say the 0.1 to 0.2 cm2 range. Repeats of such failures could be tolerated only if they remain within the 10-5 loss fraction at any point constrained by radioactivation considerations or the available cooling to the bombardment point. At the 3 GeV stage the initial heat deposition rate is marginally smaller, 15 J g-t, and much less amplified by hadronic cascade effects. At 450 MeV the 2 I-'C pulses deposit only 4 J g-l with no significant hadronic cascade contributions so that beam spots as small as 0.1 cm2 are not expected to cause any serious problems from single full-power pulse losses at this stage. In regions where the spill rates are 30 W m-1 the annual integral dose to the outside of beam line components is esti~ated to be of the order of 105 Gy per year in a fully operational facility [3]. The dose rates in the interior of beam line components will be a factor of, say 4, higher and at cable tray locations a factor 4 lower than this value. This will restrict the use of materials to those that can tolerate at least 106 Gy radiation dose in beam line components for even a few years of maintenance-free service and cable insulations that can tolerate at least 105 Gy dose before failure. Materials with radiation damage thresholds an order magnitude greater are available for most applications and should be selected. 7.4 Beam Collimators To avoid complete dependence on low beam transport losses by an exquisite understanding of all beam loss modes for the facility, the use of beam scrapers/collimators at strategic points in the system is being examined. Such devices have been used previously on coasting beam facilities [4,5,6,7,8] during filling and to control emittance growth during operations. Their primary function was the removal of circulating particles in the phase space that would, if left in the beam, eventually be lost near the beam intersection points, where ex-perimental detectors were mounted. The KAON requirements are more demanding in that 7-5 it is necessary to reduce the uncontrolled losses generally, not just at a few points around the rings and the intercepted beam must be captured in heavily self-shielded absorbers to avoid the residual radiation fields in adjacent areas. In view of the cycling frequency of the systems (10 and 50 Hz) and the large mass of the thick collimator elements required for 30 Ge V, or even 3 Ge V protons, these collimators will likely be fixed over the cycling periods of the rings but would need to be adjustable in the long term, with position resolution of about 10 microns at two points for lateral and angular position control. Because the reflection albedo for charged particles at grazing inci-dence is always substantial, at least two beam edge defining apertures are usually required to obtain significant reductions in the particle populations outside the acceptable phase space of each orthogonal dimension. Although it may be possible to reduce the lateral density at the downstream collimator edge with thin aperture defining foils the conserva-tive approach presumes that all beam defining edges are thick and will require substantial shielding. Fig. 7.4.1 shows a conceptual design for such a collimator/catcher shield. To obtain significant relief from the radioacativation of downstream components the catcher -------- .... ......... " , , , , \ \ \ \ \ \ \ I ______ ...1_ / I I I I I I I I SIX CYUNOER OFTSET / / ~" . ,=~:~//// '--+- ----~--------------~·I 2 - J '" Figure 7.4.1: Conceptual design for a beam collimator and surrounding shield shields must be a very close fit to the beam edge. Analysis of a hadron cascade measure-ment for 28 GeV protons on iron made at the Brookhaven AGS[9] indicates that collimation to a diameter of about 6 cm is required to catch 90% of the hadron cascade in the shield, i.e., one order of magnitude reduction in the radioactivation of downstream components. The rather close position tolerance and high shield density requirement indicates the need for the off-set alignment device shown in Fig. 7.4.1. Six nested cyclinders, two with axis parallel to, but off-set from the nominal beam line axis and two with offset axis at a small angle to the beam line axis allows repositioning of the scraper/catcher assembly in lateral 7-6 position and angle at any time without moving the rather massive iron/concrete shields. The dimensions of these shields around the possible collimators for the various storage rings and accelerators are shown in Table 7.4.1. In order to collect beam loss in all three Table 7.4.1: Collimator Shield Dimensions Storage Ring Iron Dimensions Iron Total Dimensions Concrete /Synchrotron Power diam.xlength Weight diam. xlength Volume kW m tonne m m3 Accumulator 5 1.5x2.0 26 2.0x2.5 4.3 Booster 10 2.0x2.5 58 2.5x3.0 6.9 Collector 10 2.0x2.5 58 2.5x3.0 6.9 .5 0.7x1.85 5 1.7x2.35 3.2 Driver 15 2.2x2.7 75 2.7x3.2 8.1 2 1.3x2.25 22 1.8x2.75 4.0 Extender 30 2.5x3.0 108 3.0x3.5 10.0 2 1.3x2.25 22 1.8x2.75 4.0 planes, each synchrotron has three collimation systems, each of which in turn consists of a primary upstream collimator, either thick or thin, and a downstream collimator, always thick, to absorb particles scattered from the first collimator. For maximum efficiency, horizontal and vertical collimators are placed immediately downstream of a quadrupole focusing in their respective plane. The longitudinal collimators in the Booster and Driver are placed in regions of high momentum resolution (TJ / .JjJ;). Secondary collimators are placed 900 or less in phase advance downstream of the primary collimator they are asso-ciated with, except for the Driver, where the momentum collimator will be in the arcs, 1400 before its secondary collimator in the straight section. In some instances one device can fulfill two functions at the same time. The possibility [4],[5] of reducing the intensity locally at the second collimator edges by multiple coulomb scattering in relatively thin upstream collimators needs further investigation. This would be useful, in particular for the momentum collimator in the arcs of the Driver, where not enough space is available for a long collimator. Tracking studies are planned in order to assess the effectiveness of the collimation schemes. In addition to these systems two collimation systems dedicated to specific loss mechanisms are provided. In the Accumulator a collimator downstream of the stripper foil will remove particles scattered by the foil. In the Extender the collimation system will be replaced by one consisting of a primary collimator and two secondary collimators per plane in order to localize the losses generated by the resonant extraction process at the pre-septum in the 7-7 extraction straight. If the collimation strategy succeeds in burying the significant radioactivity production in the scraper/catcher collimators these will be the only devices that we would expect to require shielding for transport to maintenance facilities. This would be achieved by with-drawing the ~50 cm diameter cylindrical assembly into a horizontal shield flask after re-moval of the upstream accelerator or storage ring elements. The mass of a scraper/catcher assembly with its offset alignment devices is 3 tonne or less j a 10-cm thick lead flask enclosure would be approximately 7 tonne for a total weight well within the beam line element transporters. 7.5 Beam Dumps As well as the shielded collimators in each ring several beam dumps are required in the beam transfer lines. The IA transfer line from the existing TRIUMF cyclotron to the Accumulator ring will have a beam dump to accept the 5 rf micro pulses out of each 45 that must be removed to allow time to turn on beam kicker elements for injection and extraction in the various storage rings and accelerators without spilling beam. Each micropulse to be removed will be kicked onto a thin foil that will strip the H- ions to protons for separation at the next dipole bending magnet into a beam line leg going to a dump buried outside the transfer tunnel. The shielding and heat dissipation for this beam dump is specified to accept 10 kW of beam power. The next dump required is immediately downstream of the injection point in the Accumulator storage ring. This injection to the A-ring is accomplished by stripping the H- ions to protons. Because very thin foils are used for this stripping a significant fraction ('" 1%) are transmitted as neutral hydrogen atoms, HO, and a smaller fraction as H-, including those that might miss the stripping foil. The HO component, at least, will be stripped to H+ ions as soon as they clear the circulating beam trajectories and transported to an external beam dump capable of dissipating 1 kW of beam power. The H- beam passing the injection stripper will also be pass to the HO beam dump. To allow testing of the Accumulator ring without operation of the Booster synchrotron a beam dump capable of dissipating 5 k W of beam power will be installed under the tunnel floor in a continuation of the first leg of the A-B transfer line through a port in the first dipole magnet when the magnet is turned off. Similarly the beam from the Booster synchrotron can be directed into an external beam dump capable of dissipating 10 kW of beam power placed under the floor of the tunnel at the beginning of the inclined BC beam transfer line or above the roof of the tunnel at the end of the inclined section. A similar arrangement for the C-D transfer line as for the A -B transfer line will be capable of dissipating 10 kW of beam power from the Collector storage ring. 7-8 A dump, to dissipate 30 kW of beam power, is planned for a switchable leg of the fast experimental extraction beam line for testing and commissioning operation of the Driver synchrotron before operation of the Extender storage ring. A 30 kW beam dump is also planned downstream of the beam clearing system from the E ring. This facility is required for controlled dumping of the residual beam left in the ring at the end of each slow extrac-tion cycle as well as allowing commissioning operations of the E-ring at 1% time averaged power without depending on the experimental area facilities. The external beam dumps with their approximated design parameters are shown in Table 7.5.1. Table 7.5.1: Operational/Commissioning Beam Dumps Transfer Beam Line Power Iron Core or Spill Point Dissipation Weight Purpose kW tonne I-A Transfer Line 10 60 Chopper micropulse disposal Accumulator Injector 5 30 HO, H- beam spill dumps at injection A-B Transfer Line 10 60 Accumulator testing and disposal B-C Transfer Line 10 60 Booster testing and disposal C-D Transfer Line 10 60 Collector testing and disposal D-Extraction (Fast) 30 100 Driver testing and disposal E (Clearing) 30 100 Extender clearing, testing and disposal 7-9 References [1] A. H. Sullivan, CERN Radiation Protection Group (private communication, 1989) [2] Ralph H. Thomas and Graham R. Stevenson, Radiological Safety Aspects of the Oper-ation of Proton Accelerators, Technical Reports Series No. 283, International Atomic Energy Agency, Vienna, 1988. [3] E.W. Blackmore, Radiation Levels in the KAON Factory Accelerator Tunnels, TRI-DN-89-K36, May 1989. [4] T. Risselada, R. Jung, D. Neet, H. O'Hanlon and L. Vos, The CERN ISR Collimator System, IEEE Transactions in Nuclear Sciences, NS-26 No.3, p4131, June 1979. [5] B. W. Montague, Beam Scraping Targets in the ISR CERN-ISR-DI/71-51, October 1971. [6] A. Ijspeert and L. Vos, Control of Background Rates in the Underground Physics Experiments of the CERN SPS Proton-Antiproton Collider, IEEE Transactions on Nuclear Science, NS-32 No.5, p.1635, October 1985. [7] L. Vos, Background Control in the SPS Collider, CERN SPS/DI-MST /ME-83-2, July 1983. [8] W. C. Middelkoop, B. de Road, P. Sievers, Beam Scraping in the SPS, CERN LABII/BT /73-1, January 1973. [9] G. W. Bennett, H. N. Brown, H.W.J. Foelsche, J.D. Fox, D.M. Lazarus, G.S. Levine, T. E. Toohig, R.H. Thomas and J. Kostoulas, Particle Distribution in a Steel beam Stop for £8 Ge V Protons, Particle Accelerators Vol. 4, pp 229-238, 1973. 7-10 H- EXTRACTION FROM THE CYCLOTRON Chapter 8 8 H- EXTRACTION FROM THE TRIUMF CYCLOTRON 8-1 8.1 Principle .... . .... 8-1 8.2 Layout of Components. 8-2 8.3 Beam Quality 8-7 8.4 RF Deflector 8-9 8.5 Electrostatic Deflectors . 8-9 8.6 Magnetic Channels 8-10 8.7 RF Booster Cavity 8-17 8.8 Engineering of Equipment for Cyclotron Beam Extraction 8-17 8.9 Test Results and Future Beam Tests ........... . 8-20 J 8 H- EXTRACTION FROM THE TRIUMF CYCLOTRON The successful accumulation of cw beam in the A Ring requires the extraction of an H- beam from the TRIUMF cyclotron [1]. The scheme requires the addition of eight new extraction devices in the cyclotron chamber. The field parameters of the extraction devices have been specified, and they have either been built or prototyped. Several of the elements have been successfully operated in the cyclotron with beam. Devices yet to be tested are in general of conservative design and use conventional technology. 8.1 Principle Direct extraction of an H- beam requires the use of electrostatic deflectors and magnetic channels to peel off the outermost orbits. As outlined in Table 8.1.1 the cyclotron is required to accelerate 165 J.LA of H-. However, separated turns cannot be maintained beyond 200 MeV. Despite this, 452 MeV was chosen as the extraction energy [2] since higher energies simplify the booster synchrotron design and the vr=3/2 resonance at 428 MeV can be used to improve the extraction efficiency[3]. Also the unavoidable beam loss from electromagnetic stripping, which rises rapidly from 400 to 500 MeV to a total of 8%, is only 2% at 452 MeV. Extraction at 500 MeV, which would simplify the changeover between present running modes and KAON operation, is too high in energy above vr =3/2 to yield stable efficient operation. Table 8.1.1: Distribution of the Current from TRIUMF Cyclotron to Achieve 100 J.LA at 30 GeV Prestripper interception in cyclotron (available for research down beam line 1A Vacant buckets for kicker magnets 5/45 Total loss at injection Total loss at extraction Reset of painting parameters (rv 2 ms/20 ns) Incomplete conversion H- to H+ Total loss at collimators Beam obtained at 30 GeV Total beam required from cyclotron Routine H+ beam presently extracted (14 fLA) 5 fLA 5 fLA 15 fLA 1.5 fLA 5 fLA 100fLA 164.5 fLA 160 fLA At the required beam intensity the extraction process must be efficient to reduce loss and induced activation. H- extraction can be made practically loss free by positioning 8-1 a thin stripping foil upstream of the first electrostatic deflector (DCD) to shadow the septum, diverting the stripped proton beam down an existing beamline to a beam dump or experimental station. We then define extraction efficiency as the ratio of the H- extracted beam to the circulating beam. A septum of effective width < 0.5 mm can be safely shadowed with a 1 mm foil for normal beam divergences of 0.5 mrad. With the low radius gain per turn (1.5 mm) at the extraction energy up to half the beam would be intercepted by the stripping foil. However, the extraction efficiency can be significantly improved by exciting a coherent radial oscillation at the vr =3/2 resonance. The subsequent precession leads to large radial beam density modulations in the 440-470 MeV energy range. In the TRIUMF cyclotron five particle bunches per turn are accelerated at an rf frequency of 23 MHz. A local rf electric field at 11.5 MHz, 5/2 the particle rotation frequency, radially deflects each of the particle bunches in alternate directions on each turn. When the deflecting field is applied in the vr =3/2 resonance region a coherent amplitude develops similar to the coherent growth from static deflections at vr=1. As Vr increases beyond 3/2, the radius gain per turn from precession can be several times the separation due to the accelerating field. With this technique the extraction efficiency can be raised from 50% to 90% with no direct loss on components internal to the cyclotron. The effect of the rf deflector (RFD) at a relatively low strength is illustrated in the computer simulation in Fig. 8.1.1. The beam density, plotted as a function of foil position, is high where the precession has led to an accumulation of turns and it is low where turn spacing has increased. The septum protection foil is positioned in a broad density minimum. 8.2 Layout of Components The first step in the design process is to fix the position of the extraction elements in the cyclotron to determine the design parameters for each component. Several requirements must be considered in choosing an optimal layout. Firstly, the protons from the septum protection foil and the H- ions must be extracted from the cyclotron through existing ports. Secondly, the extraction devices should be placed to minimize modifications to existing hardware. Ideally the elements should be positioned and designed to minimize restrictions on the existing physics program during the early commissioning of the KAON Factory. Lastly, beam dynamics considerations restrict the positioning of extraction devices. The DCD must be positioned either immediately downstream from the protection foil or in one of the secondary shadows that recur N * 2 7r /vr downstream. Figure 8.2.1 shows the case of a foil at 2400 when Vr is slightly larger than 1.5. The width 2£ of the beam free region in the azimuthal direction is w / x' where w is the foil width and x' is given by 8-2 RFD @ 65V/mm·m 80 CJ) 60 CJ) BEAM DENSITY 0 40 -.J ~ 0 20 0 -en -4 Q) .c u c: -a..Cl:: -8 -12 298 300 302 304 306 R (inches) Figure 8.1.1: The bottom plot traces even and odd turns of a central particle in phase space during and subsequent to perturbation by the RFD (R1'V297 in.). The top trace shows the beam density that results, given as the percentage of the beam that would be lost on a 1 mm foil. TURNS I N -I I I e w~ I N+1-~ FOIL H+ N+2 -I N+3-~ N+4 -I N+5-~ I N+6 - : I 120 T I 240 I AZ IMUTH (degrees) I I 360 -I-I-I--Figure 8.2.1: The secondary shadows produced by a foil at 2400 are shown schematically for Vr '" 1.52. The shadows gradually fill in as the finite phase band generates an energy difference. 8-3 I Ad R for a matched beam with incoherent amplitude Ai and average radius R. (For a radial emittance of l7rmm-mrad, x' ",,0.5 mrad, so 2£ ",,2 m for w=lmm). The figure shows the shadows gradually filling in as the finite phase band generates an energy difference. This will occur in N turns where N is given by where: D..<jJ is the half width of the phase band and dr / dn is the radius gain per turn. For a 1 mm foil and a 40° phase band the shadows will disappear after ",,3 turns (dr / dn "" 5 mm). Thus the septum should be located within one or two turns downstream of the foil. In addition, after several turns, more of the circulating beam will be intercepted by the foil due to the large radial coherent oscillation, causing a large reduction in the efficiency. Since the beam is still well inside the isochronous field and v,. "" 1.5, the DCD deflected beam will oscillate around the equilibrium orbit reaching maximum separations at "" 60° + N * 240° downstream from the original deflection. The magnetic channels must be located in or near one of these maxima. A detailed analysis of the problem produced six basic schemes for extraction utilizing three different beamlines for proton extraction[4]. The scheme chosen as the optimum one is shown schematically in Fig. 8.2.2. The protection foil is positioned to extract protons down beamline 1 (BLl) which is capable of accepting 200 p,A. During KAON Factory Figure 8.2.2: Reference position of extraction elements in the cyclotron commissioning a wide foil can be used to extract 100 p,A or more of protons down BLl leaving a trickle of current extracted as H-. Two DCD's are positioned to deflect the beam at the position of the first secondary shadow. Four magnetic channels (MCl-MC4) are needed to steer the deflected beam out of the cyclotron. The strength and position of the 8-4 4~--------~------~~------~------~~ (A) NO QUADRUPOLE AND NO GRADIENT IN MC5 ENERGY SPREAD IN BEAM = +- .8 MEV. ~ -41----------+~~~~~~~~----4_~-L--~~ '-'" X (B) .3 M. QUADRUPOLE WITH 3.9 T./M. GRADIENT PLACED BETWEEN MC3 4£ MC4 .7 T.jM. GRADIENT IN MC5 ENERGY SPREAO IN BEAM = +- .8 MEV. 1oC' -4~--------+-~~~~~~~----~~~----+ 240 265 290 THETA(DEG.) 315 340 Figure 8.2.3: Shown are two plots of the radial beam envelope as the beam passes through the magnetic channels and into the exit port. The strong radial defocus sing forces from the cyclotron fringe field observed in (a) are compensated in (b) by a 4 Tim gradient in QMC1 and a 0.7 Tim gradient in MC5. Both plots include an energy spread of ± 0.8 Me V in the extracted beam channels is chosen to allow the beam passage through the exit horn where a fifth channel is located. A quadrupole magnet positioned between MC3 and MC4 will precompensate for the strong radial defocussing forces that occur as the beam exits the cyclotron by providing additional radial focussing (Fig. 8.2.3), and a weak gradient in the last extraction element MC5, is used to control the width of the beam through this magnet. In Table 8.2.1 the reference specifications of the extraction devices are presented including the clear separation of the deflected beam from the circulating beam at the channels. The angular positions are given with respect to the centre line of magnet valley 3 which is 5.50 upstream of the east dee gap. The positioning of the first DCD in the secondary shadow of the septum protection foil has two advantages: the azimuthal position can be chosen to coincide with the maximum width of the shadow whereas in the primary shadow some azimuthal clearance is needed to allow the protons to clear the DCD (Fig. 8.2.1), and the conflict of having the DCD in an extraction path used in normal TRIUMF operation is avoided, thus allowing simultaneous extraction of H- and proton beams. 8-5 Room has also been left to install an auxiliary accelerating cavity (RFB),[3] to further dilute the beam density by increasing the energy gain per turn at extraction. Device Foil RFD DCD1 DCD2 MC1 MC2 MC3 QMC1 MC4 MC5 Table 8.2.1: Specifications for Extraction Devices Angle Leading Effective Edge Strength Gradient Length ( elv3) (m) 239.1 136.75 0.2 MV/m 0.5 114.1 3.9 MV/m 1.0 123.6 3.9 MV/m 1.0 264.5 0.10 T 0.85 274.5 0.15 T 0.85 285.6 '" 0.56 T 0.90 293.0 0.0 4.0 T/m 0.30 296.0 '" 0.57 T 0.90 320.0 0.50 T 0.7 T/m 1.20 ~R (mm) 30 50 140 300 400 Electromagnetic stripping loss during normal operation is also reduced. When operating at full power the device would nearly double the existing energy gain per turn of 340 keY. However, the corresponding doubling of the energy spread and radial emittance of the extracted beam makes injection into the accumulator of the KAON Factory less efficient. 8.3 Beam Quality A Monte-Carlo code COMA[5] was used to simulate the characteristics of the extracted beam. The extracted beam quality is of importance for painting the beam into the accu-mulator ring. Acceleration of 165 IlA requires a phase band of at least 200 at 23 MHz. Fluctuations within the cyclotron cause this to vary by ± 50 at 2 Hz, and by an additional ± 50 over a 10 minute, or longer, interval. Thus the beam is accelerated in a 400 phase band centered around the ideal phase. For centred beams the transverse emittances required to accelerate 165 IlA are 1 7r mm mrad in each plane. Since the turns inside the cyclotron are not separated, the radial emittance of the extracted beam is not that of the circulating beam, but is also dependent on the radius gain per turn and the energy spread. Generally, in improving the extraction efficiency, by increasing the radius gain per turn, the radial emittance of the extracted beam becomes larger. Radial emittances of the extracted beam for various cases are given in Table 8.3.1[6]. 8-6 Table 8.3.1: Extracted Radial Emittances for Various Conditions RFB RFD E 6.E f3,€x Efficiency (kV) (V/mm·m) (MeV) (MeV) (7r mm mrad) ±10° ±20° ±10° ±20° 0 55 452 1.1 1.5 1.6 1.8 84% 0 82 452 1.1 1.5 2.0 2.5 87% 0 110 452 1.1 1.5 2.7 3.3 90% 150 110 463 2.5 3.5 4.5 6.5 91.5% 300 110 473 4.0 5.0 6.0 9.0 93% The extracted vertical emittance is dependent largely on the circulating emittance for a centred beam but is increased somewhat by two effects: the large radial off-centring produced by the RFD couples vertically due to the non-adiabatic passage through the variations in the magnetic midplane which are the cause of a further increase since extrac-tion takes place over several turns. These effects lead to a total increase of ",30% in the vertical emittance or an extracted emittance of 1.3 7r mm mrad. The various added devices can contribute to phase slip and/or additional increases in the extracted emittance[7]. The radial gradient in the RFD field, alternating in sign on neighbouring turns, can drive the lIz=0.25 resonance and increase the vertical amplitude. In the worst case growths of 50% have been simulated but this can be controlled by detuning the local liz value with the existing trim coils. Recent measurements show that the growth is no more than 20%. Beam dynamic restrictions place a number of constraints on the magnetic channel fields. The channels whether air core or iron compensated will have fringe fields that extend into the circulating beam. The radial gradients in the fringe fields can lower the already weak vertical focussing causing an increase in the vertical amplitude. A total tolerance integrated along the beam trajectory of f dBz / dr . dl < 60 mT /m·m is required to keep the change in liz to less than 0.05 which corresponds to a 25% increase in emittance. The third harmonic component of these radial gradients will tend to stretch the radial emittance at the lIr=3/2 resonance which lies'" 10-15 cm from the septum of MCI and MC2. To limit the increase in radial emittance by no more than 25% above that already occurring at this resonance, a limit of f 8Bz3 / 8R·dl of 10 mT /m·m is placed on the tails of the channel fringe fields. Finally, we require that the total phase slip produced by the leakage field be less than 100 • In the smooth tails far from the channel, trim coils can be used to control phase slip. However in the radial range 6.R close to the channel where f3 is changing more rapidly we would need to limit the total Bz change to < 45 mT·m·6.R. Within the aperture of each channel we have aimed to keep the integrated field experienced by any particle in the beam to be within ±0.2 mT·m of the field felt by the central particle, over ±l cm by ±1 cm, in order to avoid focussing effects that would increase the size of 8-7 the beam downstream. The total extracted H- beam parameters taking into account all of the above effects are given in Table 8.3.2. Table 8.3.2: Extracted Beam Parameters Extraction Efficiency Horizontal Emittance Vertical Emittance Extraction Energy Energy Spread Phase Width Phase Jitter - "V 2 Hz - "V 10 min Frequency 8.4 RF Deflector 88% 3.5 7r mm mrad 2.0 7r mm mrad 452 MeV 1.5 MeV 20 deg @ 23 MHz (2.4 ns) ± 5 deg (± 0.6 ns) a further ± 5 deg (± 0.6 ns) 23.055 MHz ± 0.002 MHz The prototype RFD, tested in the cyclotron, has performed to specifications, and has became a permanent installation. A voltage of 20 k V (one half of the achievable maximum) requires an electrode le~gth of 50 cm to generate the required amplitude. The radial oscillation is driven by the field between the tip electrode and the ground shield. To keep open the possibility to operate the cyclotron above 450 MeV, all hardware past the 517 MeV orbit is kept clear of the beam plane. An open C-type structure was selected (Fig. 8.4.1.). To suppress the leakage of the rf energy out of the structure, the electrodes are completely enclosed except for beam entrance and exit. The unit, very simple and lightweight, is connected to a cylindrical coaxial inductive stub designed to operate as a >../4 resonator. The power required is "V 4 kW to achieve 27 k V peak on the deflecting electrode. Coarse and fine tuning are obtained with variable capacitors. 8-8 GROUND SHIELD HOT ElECTRODE Figure 8.4.1: RF Deflector 8.5 Electrostatic Deflectors The electrostatic deflector consists of aim long septum at ground potential, and a positive antiseptum generating a radial field of 39 k V / cm, located in a magnetic field of up to 0.5 T. Both electrodes are curved to match the orbiting and extracted beams, and an opening for the pre-stripped beam of H+ is incorporated in the design of the support structure. The inter-electrode spacing of 13 mm requires an operating voltage of 51 kV. The deflectors must be positioned very accurately with respect to the shadow ,in the circulating beam. Both the entrance and exit of the septum are adjustable radially by ± 5cm and ± 10 cm, respectively, in 0.05 mm increments. The existing extraction probe for beamline 4 will be used to carry the protection foil. The septum is made of 140 vertical molybdenum foils 5 mm wide and 0.076 mm thick separated by 2 mm gaps. The curvature of the septum is achieved by two matching stainless steel templates 10 cm apart, contoured to the 452 MeV beam orbit, and centered on the beam plane. Each foil is preloaded to 34 N by retaining each end with BeCu leaf springs. In case of foil failure the leaf springs remove it from the beam plane, preventing a short circuit to the antiseptum. Seven foils at each end are insulated to allow beam current read back for proper positioning of the septum in the stripper foil shadow. These foils are grounded for normal operation. The hollow stainless steel antiseptum, supported by two alumina insulators, is cooled by nitrogen gas, while the support structure is water cooled. The high potential is brought to the antiseptum through one insulator by means of a standard high voltage cable enclosed in a stainless steel bellows of sufficient flexibility, 8-9 to permit the required motion of the unit. The septum of the prototype deflector has an effective thickness of '" 0.3 mm, and has held 50 kV with acceptable spark rates during the beam tests. At present, improvements to the insulators continue, and voltage holding at 100 I'A beam intensities will be tested in the spring of 1990. 8.6 Magnetic Channels The circulating beam passes close to the septum of the first two dipole channels, but the integrated field reduction required of these channels is small (85 to 130 mT·m). These channels are therefore of an iron-free design. For the later dipoles, where the separation has increased, iron core designs are proposed, to give almost total cancellation of the cyclotron magnetic field (500 to 570 mT). Additional windings on these dipoles are however necessary to compensate for the effect of the iron on the circulating beam. Iron-free channels are commonly used in extraction systems for proton cyclotrons,[8,9,10] and provided a basis for this design work. However, because of the relatively weak vertical focussing in the TRIUMF cyclotron, additional emphasis was placed on minimizing the gradients of the magnetic channel fringe fields at the circulating beam. Both two and three dimensional field calculations were used in the design of MC1.[ll] Figure 8.6.1 illustrates a cross section of the conductor arrangement for the MC1 prototype as determined using a 2D field computation.[12] A pair of opposed dipole coils is employed in this case, with a thin septum winding near the circulating beam. Conductor placement as well as the currents in the septum and cancellation windings were chosen to minimize the field gradients in the vicinity of the circulating beam, while at the same time maintaining a flat field in the channel. Three dimensional calculations of the field contributions from the ends of the windings, made using the group of programs CANAL_3D,[13] show that significant gradients in integrated field can result from the coil ends, but that these can be compensated by ±5% adjustments of the winding currents. The field integrated along the beam direction and its gradient are shown in Fig. 8.6.2. The integrated field within the channel is'" 85 mT·m and is uniform to within the beam dynamics tolerances over an aperture of ±1 cm vertically and ±1 cm radially. The integrated radial gradient at the position of the circulating beam (at R ~ -1.5 cm) is ~ 30 mTm-1'm, and the gradient averaged over the lIr = 3/2 resonance (12.5 cm inside the septum) is 1.0 mTm-1·m. Full (r,8) maps ofthe field from the channel, both in the median plane, and at z = 1.0 cm, were computed for beam tracking studies. In addition an (r, z) map of the field within the channel was generated to check on the beam properties over the aperture of the channel. 8-10 11=745 A -11 +12 +11 -12 12=665 A ® 0 0 ® 0 0 0 0 0 0 0 0 0 CIRCULATING 0 0 0 BEAM >® 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 EXTRACTED BEAM I , , I 0 cm 5 Figure 8.6.1: Cross section through MC1 showing the conductor locations and currents. The channel consists of a septum coil (11) and an opposing cancellation coil (12). Me 1 FIELD and GRADIENT 20 I I I 50 , r-.. ----- ... , ", " " ~, , --,' I , , --~ , , , , , , , , . \ , . - \ , I-" " " 0 .----. E -20 l-E E 0 r-.. E ......... -50 l-E ......... '-" - l f-- -- I --40 l? -60 N m '--, -80 -100 =a r-.. -150 L eo ~ -200 rn eo '-" SEPTUM - f- '--, -100 I I I 250 -20 -15 -10 -5 o 5 R (em) Figure 8.6.2: Results of the 3D model, in the median plane (z = 0). The vertical magnetic field (solid line) and the radial gradient (broken line), integrated along the beam direction. The water cooled conductors in the channel are 5 mm wide in the septum, and 10.4 mm elsewhere. Each turn of the septum coil consists of two small septum conductors (connected in parallel) joined to one larger return conductor. The current density in the conductors is 2.0 kA/cm2 in the septum and less than 1 kA/cm2 elsewhere. The total power for both coils is 10 kW. 8-11 " 1.0 0.8 E 0.6 u 0.4 Z 0.2 t:.o 0.0 r 0.2 0.4 0.6 0.8 1.0 em Figure 8.6.3: Contours of magnetic field inside one quadrant of MC3 Manufacturing and assembly tolerances for the channel were calculated using the 2D model. The tolerances on the position of conductors nearest to the beam are quite strict (±0.1 mm), and are determined by the perturbation to the field gradient (8:,), and the al-lowable radial component (Br) introduced by vertical asymmetries. For conductors further from the beam these tolerances are relaxed (±0.5 mm to ± 2.0 mm). The second magnetic cha:rmel (MC2) is similar in design to MCl. In this case the field must be scaled up by 50%, and the currents increased accordingly. Since the separation from the circulating beam is larger, doubling the size of the septum conductors is permitted, so both the required power and cooling water pressure remain reasonable. Since MC2 is located near the edge of a hill, studies were done of the effect of the pole steel on the channel field. Although the problem is strictly a three dimensional one, the effect was investigated using simpler two dimensional models. For the case where the pole steel extends fully over MC2, calculations including the effect of images in the steel, show that the perturbations caused by the steel are within the beam dynamics tolerances. Magnetic channel MC3, the first iron core dipole, is 0.9 m long, and is located in a region where the main-magnet field is 0.56 T. The centre of the channel is 14 em from the circulating beam. To circumvent sensitive gradient effects, saturation in the iron was avoided by limiting magnetic induction in it to < 1.7 T. Results of a POISSON calculation for MC3 are shown in Figures 8.6.3 through 8.6.5. Inside the channel the average field is ",3 mT, for a current of 17 kA·turns (which is optimal for compensation of the external field) and is uniform to within ±0.2 mT over the beam 8-12 S J2 .~ u :.::; <1> C c:r> CJ ~ E .......... l-E -----+--' c <1> '6 0 .... c:r> J2 ~ U :.::; OJ c c:r> 0 ~ 5.70 5. 68 5. 66 5 . 64 5. 62 5 . 60 5 . 58 5 . 56 5 . 54 5.52 5.50 50 40 30 20 10 a -10 -20 -30 -40 -50 20 2.0 ern a b ove medium plane ---~ 1.5 em above medium plane--~..-:;~ 18 1.0 ern above medium plane 0.5 em above medium plane at medium plane ---~-I' 16 14 12 1 0 8 6 Distance from centre of channel (cm .) > r Figure 8.6.4: Magnetic field outside MC3 20 2 .0 em above medium plane 1.0 em above medium plan o .~ em at medium 18 16 14 12 10 8 6 4 Distance from centre of channel (cm .) > r 4 Figure 8.6.5: Magnetic field gradient outside MC3 8-13 5B.1 9.3 Sq. 9.3mm Sq. Conductor x 5.2mm 1.0. Conductor Area = 62.Bmm Sq. Figure 8.6.6: Cross section of test channel. Dimensions in mm. cross section within an aperture of ±1 cm by ±1 cm. The external field gradient is -6 mT 1m at a distance of 14 cm. To evaluate the perturbation of the cyclotron field by the iron channel, a different type of calculation is necessary. This involves simulating the main magnet by a C magnet, and placing an iron block and a compensating coil of equivalent cross sections in the gap of the C magnet. The difference in the calculated magnetic field in the gap when the iron block is absent and present are the perturbations due to the iron block. It is found that the iron block causes a negligible difference in the magnetic field when properly compensated. At 14 cm from the centre of the channel the field is perturbed by 50 J-LT and 2 mT 1m. These reduce to less than 10 J-LT and 200 J-LT 1m at 20 cm. To check the accuracy of the above calculations, a test channel was built and measured in a magnet with a maximum field of 0.4 T. The channel walls were scaled down to achieve an induction of 1.7 T in the iron (Fig. 8.6.6). The measured results show that when the compensating current is increased from 0 to 2.3 kA·turns the field inside the channel drops from more than 90 mT to less than 2 mT. The requirements of magnetic channel MC4 are similar to those of MC3, except for a slight difference in the background field, which is 0.57 T for MC4. The cross section of the proposed MC4 is identical to that of MC3. The slight increase in the background field can 8-14 be compensated by increasing the current in the compensating coil to 18.0 kA·turns. A radial focussing gradient (see section 8.2) is generated by a separate iron-free quadrupole coil. A gradient of 4 Tim (see Fig. 8.6.7) can be produced by a current of 3360 A·turns in each of the coils. POISSON calculations show that the field deviates from the ideal gradient by < ±0.06 mT, over an aperture of ±1 cm by ±1 cm. Outside the coil the gradients are less than 4 mT 1m at a distance of 14 cm or greater at all heights. 4. to 4.08 4. 06 4.04 4. 02 E ~ 4. 00 I-3.98 3. 96 3. 94 3. 92 3 . 90~~~----r---~---.----.---~~~----r -2.0 - 1.5 -1.0 -0.50 . 0 0 . 5 1.0 1.5 2. 0 em Figure 8.6.7: Magnetic field gradient inside Quadrupole magnet. Magnetic channel MC5 is located in the exit horn of the cyclotron vacuum chamber, and is sufficiently far away from the circulating beam to neglect the perturbation due to a steel core. The channel may therefore be considered more as an active bending magnet than a true channel, and a conventional window-frame design has been chosen[14]. Nevertheless, the fact that the channel is located inside the exit horn means that the vertical space available is limited (10 cm), and this imposes constraints on the design of the channel ends. POISSON calculations were used to design a magnet with the required 500 mT strength and good field uniformity over the beam aperture. The result is a channel with 4 turns, carrying 4 kA through a conductor of 1.9 cm cross section. The current density is I'V 1.5 kA/cm2 , and the total power is 13 kW. A hollow conductor is used to provide direct water cooling. The gradient required in this channel will be provided by suitable shaping of the steel profile. 8-15 In summary the charmel fields are within the tolerances specified in section 8.3. The total integrated radial gradient from all charmels at the circulating beam is 60 mTm-1·m, while the radial gradient averaged over the Vr = 3/2 resonance is -3 mTm-1·m. The total phase slip produced is ",5°. Within each channel, the integrated field experienced by any particle in the beam is within ±0.2 mT of the field felt by the central particle. 8.7 RF Booster Cavity The cavity is >./4 wide in the radial direction and (3)./2 long in the azimuthal direction so that the ions receive two comparable energy kicks per passage. The radial gradient in the time-varying azimuthal electric field generates a radial magnetic component that gives a radial kick to off-phase particles. As they pass through the cavity the turn separation increases, and the phase spread decreases. The cavity is built as a capacitively loaded >./4 long section of a rectangular coaxial transmission line split along the beam plane, to allow for beam passage, operates at the fourth harmonic of the rf (92 MHz). The cavity, trape-zoidal in shape, with outside dimensions ",1.5 m by 0.9 m and 0.4 m high, and consists of two separate units, mounted on and serviced independently through the cyclotron chamber floor and lid. The cavity, rf power amplifier and the 60 m long transmission line assembly have been tested to full power in the cyclotron simulation chamber. A peak voltage of 180 kV with ",80 kW of rf input power has been achieved. The additional loading due to acceleration of a 200 p.A beam will increase the power requirement by ",15%, still com-fortably within the amplifier tested power rating of 120 kW. The cavity will be installed in the cyclotron in early '1990. 8.8 Engineering of Equipment for Cyclotron Beam Extraction The engineering specifications of all extraction elements are relatively modest, with the exception of the first and second magnetic channels where the tolerances on magnetic field gradient in the region of the orbiting beam are tight but achievable. It is envisaged that during the commissioning stages of the accelerator rings the ongoing experimental program using 68-520 MeV protons will continue. Placement of the extrac-tion elements for H- beam precludes proton extraction above 450 Me V unless the septa and channels are removed. As a consequence the extraction equipment will be installed and removed several times during commissioning. A study was aimed at circumventing the problem by retracting all hardware radially outside the 500 Me V orbit and parking it inside the cyclotron without breaking the vacuum. Such a scheme was not feasible for reasons of mechanical complexity and insufficient space at the cyclotron periphery[15]. 8-16 All equipment introduced in the vacuum chamber must be radiation resistant and be compatible with pressures of"" 3 x 10-8 Torr. This places some limitations on the materials used for construction of devices located in the beam plane of the cyclotron. The coils of the magnetic channels are a typical example. Previous TRIUMF experience with radiation hard, high vacuum compatible insulation of heavy copper conductors was rather limited. Following consultations with other accelerator laboratories and a search of manufacturing companies, a local supplier was found capable of producing the desired thin layers of alumina on copper with the required properties[16]. With temporary lead shielding in place the residual radiation fields inside the open cy-clotron tank are ",,8 mSv /hr at the centre, reaching 100-300 mSv /hr at the periphery. The radiation dose to personnel entering the cyclotron during shut-downs must be kept below the administrative limit of 5 mSv per day. This limits access to many locations to only minutes per day per person. This, and the fact that all extraction elements become radioactive after their first use, necessitates remote installation and removal of all equip-ment. On removal from the cyclotron shielded storage containers must be used before re-installation or disposal of the equipment. The intrinsically tight tolerances on the cyclotron magnetic field practically prohibit the use of any substantial amounts of ferromagnetic material within at least 1 m from the median plane. On the other hand, the magnet gap of 52.8 cm gives rise to large magnetic fringe fields (up to 0.1 T in the immediate vicinity of the cyclotron vacuum vessel) and auxiliary equipment (electric motors, mechanical drives, control elements etc.) in addition to satisfying the proxiI?ity restrictions must also be well shielded. It is not possible to mount any additional equipment from the vertical side wall of the cyclotron vessel. There is no access through the six magnet yokes which occupy most of the perimeter, and due to steel activation in the beam plane, no new penetrations in that region are possible. The leads and services for all new components therefore come through the floor or lid of the vacuum tank. All existing ports are presently utilized to capacity, and new holes have to be cut and flanges welded where space is available. The force of atmospheric pressure on the vessel is counteracted by over 300 pairs of tie-rods spaced 0.8 m apart. The 55 pairs of trim coils, 13 sets of harmonic coils and numerous cooling lines also restrict the space for new penetrations. All these conductors and pipes have to be spliced, temporarily moved out of the way, reshaped about the new ports, and reconnected without significantly affecting their respective functions. Three new ports for the RFB were added by remote cutting of 150 mm diameter holes in the floor and lid using electrical discharge machining (EDM), and remote TIG welding of a tube and a flange assembly into the holes. The operation was fully remote, and resulted in structurally sound and leak tight ports. The positioning of the septa and channels requires a total of eight mechanical drives for radial adjustment under beam operation. The travel for each drive is typically 5-10 cm with minimum incremental motion of ",,0.05 mm. In the case of the magnetic channels the magnetic forces are of the order of several kN in addition to friction and flexing of 8-17 the water cooled current leads. A generic drive and positioning system has been designed and successfully used on the first DCD during beam operation. A flexible water cooled conductor compatible with the cyclotron environment ( Fig. 8.8.1) was tested at 3000 A, and operated in a test configuration at 1000 A for days in order to assess its corrosion resistance. A high current, remotely operated and high vacuum compatible water cooled connector awaits testing. Stainless steel Bellows flexible Welding Coble Coolant Flow stainless steel Braid Figure 8.8.1: Transition from Rigid to Flexible Conductor All additional extraction devices require extensive protection against beam damage. Pre-vious experience shows that the most sensitive and fastest protection is accomplished by monitoring the local increase in radioactivity at locations where the mis-steered or stripped beam eventually impinges on the cyclotron vessel. However, at high beam intensities the background radiation level may mask the effect and render this method useless. Direct measurements of the beam currents, mainly the beam tails or halos intercepted on strate-gically located pick-up electrodes, are often unreliable due to rf field leakage from the dees. The background noise necessitates complicated signal processing, and spurious readings occur. The most reliable method for detecting small beam spills inside the cyclotron is based on measuring local temperature increases caused by the local beam heating. A system based on this principle, although relatively slow (the typical time constants are 1 sec.), has been utilized on several devices, and proved to be very safe. The tempera-ture monitoring system can be expanded within the present cyclotron control system to accommodate additional sensors, and will be used in combination with other methods as necessary. 8-18 8.9 Test Results and Future Beam Tests H- extraction tests are carried out in the cyclotron shutdown periods once or twice a year. The results from one test are shown in Fig. 8.9.1. A radial differential probe positioned just downstream of the DCD measured the transmission through the electrostatic deflector, protected by a 1 mm foil, to be 90% at a circulating current equivalent to 66 p.A at a 1 % duty factor (RFD strength = 100 V /mm·m, DCD field = 38 kV /cm). During the same test transmission measurements without the protection foil determined that the effective septum width was <0.5 mm. Later, a dc current of 10 p.A (20 p.A equivalent) was circulated to test the ability of the electrostatic deflector to hold voltages at these higher current levels. The beam was deflected by a stable field of 32 kV /cm for 20 minutes onto a wide foil 1.5 turns downstream and extracted as protons down a beam line to a dump. 800 Total 600 90% a: C400 H 200 o 292 296 300 304 Probe Radius ( inches) Figure 8.9.1: Experimental result from April 1987. A 1.25 mm differential density probe immediately downstream from the DCD records the modulations in beam density, pro-duced by the RFD (100 V /mm·m), and the separated beamlet, deflected by the DCD (38 kV /cm). A total beam scan indicates a 90% transmission past the 1 mm protection foil and DCD septum for a circulating current equivalent to 66 p.A dc (1% D.F.). The stability and width of the first secondary shadow has been confirmed by circulating a current of 20 p.A. A 3 mm foil was positioned in a beam density minimum produced by the RFD, and a 1.5 mm differential probe positioned 2400 downstream was scanned to record the foil shadows. The stability and width of the shadow matched that predicted by computer simulation and remained fixed for an eight-fold increase in the radial emittance and a phase excursion of ±10° (Fig. 8.9.2). When the circulating current was raised to 100 p.A equivalent, simulating actual KAON Factory operation, the hole did not measurably 8-19 narrow. In addition, the operation of the cyclotron as indicated by the positions of the beam density minima and the secondary shadow, was shown to be stable for a period of 5 hours. ..---.... :J 0 ---0::: -0 '-...... -0 (0) 2 1 0 (b) 2 1 o 7 . 67 7. 68 ----,,- - -." 7 . 69 Radius €= 1 TIm m . m r a d €= 8TIm m . m r a d 7.70 7.71 (m) 7.72 Figure 8.9.2: Experimental result showing the stability of the first secondary shadow. A 1.5 mm differential probe, 2400 downstream from a 3 mm foil (see dotted line in Fig. 8.2.1 records beam density for (a) a ±100 swing in the 200 phase band and (b) an eight- fold increase in the radial emittance. It is planned that H- extraction will occur within two years of funding. Presently a full current test for the DCD is planned for the Spring of 1990 to determine the voltage holding capability at KAON intensity levels. Prototype MCI and MC3 channels are being manufactured. MCI will be installed and tested in 1990. The second channel, MC2 is simply a scaled up version of MCl, and MC4 will be identical to MC3. It is envisaged that three cyclotron shutdowns, each 4 months long, would be sufficient to achieve extraction within 2 years after funding. TRIUMF will operate normally for the remainder of each year. In the first year the cyclotron would be readied for the addition of the new equipment, new ports would be cut, existing equipment moved or altered where needed and positioning tables and drives would be tested. In addition the two electrostatic deflectors and MCI could be installed and tested. The remaining magnetic channels and the extraction horn and beamline could be readied and tested in the second year. In this way experience in beam extraction can be gained, and beam quality measurements taken at least 6 months prior to the commissioning of the A-ring. 8-20 References [1] G. H. Mackenzie, et al., Proc. 10th Intern. Conf. on Cyclotrons and their Applications, MSU, p. 203. (1984) [2] R. E. Laxdal, H- Extraction Energy, TRI-DN-K-97. [3] R. E. Laxdal and G. H. Mackenzie, IEEE Trans. Nucl Sci., NS-32, No.5, p.2453. (1985) [4] R. E. Laxdal and E. Rossen, Placement of Device& for H- Extraction, TRI-DN-89-43 [5] C. J. Kost and G. H. Mackenzie, IEEE Trans. Nuc!. Sci., NS-22(3), p. 1922 (1975) [6] R. E. Laxdal, Emittance& of Extracted H- Beam&, TRI-DN-K (in progress). [7] R. E. Laxdal, et al., Tolerances on Fringe Fields from H- Extraction Devices, TRI-DN-K (in progress) [8] S. Lindback, A Study of the Regenerative Extraction System in the CERN Synchro-cyclotron, Proc. 5th Int. Cyclotron Conf., p. 235 (Butterworths, London, 1969). [9] E.J. Martin, IPN synchrocyclotron (Orsay), private communication. [10] S. Holm, Reconstructed Gustaf Werner Cyclotron, private communication. [11] J.B. Pearson, Field Design of Magnetic Channel No.1 for the H- Extraction Scheme, TRI-DN-89-K96. [12] J.B. Pearson, Guide to Using the CHANNEL Group of Programs, TRI-DN-89-K95. [13] S. Gustafsson, Manual for the Program CANAL_3D, IPNO-AGOR-0l-89, (I.P.N., Or-say, 1989). [14] E. De Vita, A Proposed Design for Extraction Channel No.5, TRI-DN-89-K63. [15] G. Spinney, G. Stanford and M. Zach, Engineering of H- Beam Extraction Equipment, TRI-DN-90-K106. [16] J. Lenz, Some Properties of a Plasma Sprayed Aluminum Oxide Film, TRI-DN-89-K66. 8-21 CONTROL SYSTEM Chapter 9 9 CONTROL SYSTEM 9-1 9.1 Introduction. . . . . 9-1 9.2 Functional Components. 9-1 9.2.1 Equipment Management System (EMS) 9-1 9.2.2 Development Support System (DSS) 9-2 9.2.3 Equipment Protection System (EPS) 9-2 9.2.4 Master Timing System (MTS) 9-2 9.2.5 Operator Interface (OPI) . . . 9-3 9.2.6 Beam Control System (BCS) . 9-3 9.2.7 Beam Instrumentation Support System (BISS) . 9-4 9.2.8 Analogue Display System (ADS) . 9-4 9.3 System Architecture ......... 9-4 9.4 Control System Design Methodology 9-7 9.5 Algorithms .............. 9-10 1 9 CONTROL SYSTEM 9.1 Introduction Traditionally, the specification of accelerator control systems has emphasized technological detail such as the number of input/output channels [1,2] rather than focus on the func-tionality to be provided by such a system. In the rapidly changing field of computer-based control systems the technologies available to the control system engineer at the beginning of a project will have changed significantly for the better before its completion. Conversely, the complexity of fourth generation [3] accelerators, such as the KAON Factory, forces the control system engineer to focus primarily on the fundamental operating requirements of the accelerator where significant issues need to be resolved. For the proposed KAON Fac-tory control system an engineering model-based approach was adopted from the outset to determine these requirements. The approach concentrated on the definition and selection of modelling domains and the application of object-oriented model-building techniques to these domains. 9.2 Functional Components In an earlier, conceptual, design for the KAON Factory control system [4] a thorough study of the interface requirements of the hardware components of the accelerator was reported. The report included an extensive review of contemporary software environments, computer interconnect mechanisms and operator display technologies. Its contents formed the basis of a starting point for the hardware design of the proposed control system. The more recent study [5] of the functional requirements of the proposed control system led to the identification of the following functional components or systems: 9.2.1 Equipment Management System (EMS) The EMS provides a uniform set of capabilities to a variety of equipment specialists. It provides unrestricted access to the low level capabilities of the equipment as determined by the specific needs of each specialist. General data logging, set point monitoring, trend analysis and graphical presentation facilities are provided for each device. The download-ing of new or updated device calibration data, the setting of trip points on monitored device parameters, alarm handling and the downloading of diagnostic software and soft-ware updates occur through the EMS. The EMS must be in place for the commissioning of 9-1 the accelerator equipment and is expected to be operational at all times, especially when other functional components are being maintained. 9.2.2 Development Support System (DSS) The DSS provides the facilities and environment needed by the control group to develop, test, simulate and build the software and hardware components of the control system. It includes test benches, performance monitoring equipment, commercial software, CASE tools etc. 9.2.3 Equipment Protection System (EPS) The EPS consists of hardware and software that function to protect accelerator equipment. The EPS is designed to disallow operation of subsystem equipment in any manner that is potentially harmful to itself or other subsystems. 9.2.4 Master Timing System (MTS) The MTS[6] provides timing pulses to all components of the accelerator including the programming of specific beam diagnostic measurements. Although the MTS requires the use of highly specialized rf hardware, it is considered part of the control system proper due to the global nature of its use. The MTS contains 2 components: • Dedicated key pulses directly linked to the magnetic field and produced shortly before the maximum (iJ) and minimum (iJ) fields in each of the Driver and Booster. • Two Standard clock-tick trains, emanating from the B and D rings, which are dis-tributed globally. Subsystems are triggered by counting clock-ticks from re-sets initiated by the key pulses. It is assumed that "beam-type" information is broadcast separately by the control system. The key pulses must be accurate to a fraction of a gauss and give beam-energy-related triggers for coordinating the 50 Hz and 10 Hz beam production cycles. The A,C and E rings do not generate key pulses, since these machines are linked to operations in either the Booster or Driver. 9-2 The problem of providing slow-timing (the Standard Clock) to drive auxiliary power sup-plies (correction magnets, bump magnets, etc.) and fast timing to drive longitudinal as-pects of beam control (transfer and synchronization) is solved by the generation of a single (fixed-frequency) clock at roughly the revolution frequency. Accuracy in the nanosecond range is obtained by incorporating additional programmed delays. The frequency of the clocks is chosen according to the criteria: (i) the magnetic field should not vary by more than 10-4 during one period, (ii) the number of clock-ticks to cover a whole machine cycle, should be less than 216, for 16 bit counters. The Booster clock frequency is 3.00 MHz, and the Driver clock 600 kHz. Transfer Timing and Synchronization. To facilitate inter-ring transfers the MTS is used to coordinate two further sets of hardware: (i) rf-synchronization circuitry, and (ii) kicker-gap trackers. The rf-synchronization is responsible for matching the frequency and phase of the rf wave-trains of consecutive rings at the time of beam-transferal. Pro-grammable gains and switches in the phase, radial and synchronization loops will allow the origin of the rf-wave train to be passed from one machine to another. It is intended to slave the Booster first to the Accumulator at iJ and then to the Collector at B; the Driver is slaved to the Collector at injection, and then becomes master to the Extender at extraction. The firing of inject/eject kicker magnets will be initiated by the key pulses; programmed nanosecond delays will provide fine tuning. Since the kickers have to be syn-chronized with the empty buckets of the kicker gaps, a kicker-gap tracker (sensing from the beam) will be provided to aid in adjusting the critical transfer-timing. 9.2.5 Operator Interface (OPI) The OPI system comprises all control room equipment and the operator interface in par-ticular. A single main control room will be provided for the combined operation of the TRIUMF cyclotron and all KAON Factory rings and beam transport lines. Workstations will be provided as operator control stations and for the graphical display of all operating parameters. Portable and remote consoles will also be supported. 9.2.6 Beam Control System (BCS) The BCS provides the functionality to operate the KAON Factory equipment in an in-tegrated manner for the production of 30 Ge V proton beams. The BCS supervises all systems needed for the overall production of the beam. Individual devices can always be monitored and manipulated using the EMS. The BCS is responsible for maintaining the subsystem settings required to produce known modes of beam operation. 9-3 9.2.7 Beam Instrumentation Support System (BISS) The BISS consists of algorithms, utilities and general software support for complex beam diagnostic equipment and experiments. 9.2.8 Analogue Display System (ADS) The ADS provides for the display of raw, high frequency analogue signals in the main control room. The signals originate from selected equipment throughout the site. 9.3 System Architecture A system architecture was derived for the control system functional components. The ar-chitecture was composed of logical entities, such as processors and communications chan-nels, that were able to provide the functionality required by the control system. The allocation of functional components to these entities lead to the multiple, interconnected processor architecture shown in Fig. 9.3.1. The allocation process took into account the constraints of typical real world computer hardware such as limited processing capacity, data transfer capability and response time to events. This system architecture contains the following logical entities: • a general purpose local area network (LAN), named the Factory LAN, interconnect-ing the operator consoles (workstations), the Ring/Factory Controllers, the EMS database management system and the control system software development facili-ties; • a Control Interconnect, tying together all general purpose and device control pro-cessors. This must provide both a fast communication link between every processor and deterministic response times for the transmission of control information. It is not expected to carry a substantial amount of data related to the control of devices; • multiple device control processors interfacing physical devices, including beam diag-nostic devices; • multiple general purpose processors managing the operation of each accelerator ring and the accelerator as a whole; • fast simulation computers for beam dynamics calculations; 9-4 • the Data IntercOIUlect, tying together every general purpose and device control pro-cessor. This interconnect must provide a high data-throughput communications link to the EMS data logging computer. The EMS also uses this interconnect to update device specific information stored in the device control processors and to down load all operation and diagnostic software; • the EMS data logging computer, incorporating a large volume database, connected to both the Factory LAN and the Data Interconnect; SITE LAN CONTROL INTERCONNECT O!VICt! ii LD Figure 9.3.1: Components of the System Architecture for the Control System The entities described above are functional in nature, and do not imply any specific hard-ware selection. An Ethernet or Token-rung LAN, running the TCP lIP communications protocol would be satisfactory for the Factory LAN. A popular hardware implementation is to house device control processors in VME crates. The VME backplane (when extended to other crates) could become the Control Interconnect with a dedicated token-ring or Ethernet forming the Data Interconnect. The emerging FDDI LAN standard may in fact provide both the required deterministic response time for the Control Interconnect and the high data throughput of the Data Interconnect. In a crateless implementation the device control processors, interfaced directly to the accelerator hardware, would be linked in an FDDI network. A breakdown of the components for each of the functional systems is given in Table 9.3.1. 9-5 Table 9.3.1: Controls System Component Breakdown Item Equipment Management SY8tem LAN cable (Data IntercoIUlect) (2 km) Tr&DllCeiver and cable Sub-eystem Specialist W /S VME-bued development system Logging/database engine Database IOftw&l'e StOl'Ble lerver Controls diagnostic equipnent Report printer Development Support System SoftW&l'e Development W /S VME-bued development system Tr&DllCeiver + cable Report printer Equipment Protection System Misc. cables (2 km) Master Timing System Backbone timing cables Intra-ring timing controllers Diagnostic timing channels KF-equipnent timing channels Timing channel cables Operator Interface Operator W /S 10 Pmable W /S Beam Control System KF equipnent FEPs Addititnal FEPs FEP resident IOfh,&I'e LAN cable (Control IntercoIUlect) (2km) 1 LAN cable (Fact. LAN) (2 km) 1 Tr&DllCeiver + cable (Control IntercoIUlect) 90 Modelling computer 1 VME crate + 2 LAN carda, etc 90 VME crate + I/O modules 108 VME crate + monitoring equip 20 LAN bridges 2 Beam In8trumentation Support SY8tem KF-DiagnOltic FEPs Tr&DllCeiver + cable (Control IntercoIUlect) VME crate + 2 LAN carda, etc Analogue Di8play SY8tem ADS FEP 1 Diagnostic Analogue Channels 855 KF Equip MUX channels 460 Backbone Analog Cables (1 km) 6 Analog input cables (1/12 km) 1315 50 MHz Digital Scopes 6 GHz cables (1/6 km) 8 GHz tr&Jlllient digitizer 9-6 Item Count 1 290 5 5 1 1 1 8 5 10 10 20 5 1 14 380 380 286 380 20 680 33 1 380 95 95 8 9.4 Control System Design Methodology The formidable task of defining the requirements of the control system entailed the analysis of the detailed behaviour of all individual accelerator subsystems and the consideration of how they interacted to affect the beam. Global design issues can only be finally determined when a more complete description of the commissioning and operation of the entire facility is available. Whereas the beam dynamics of the "core" of the beam is well understood, the intense beams envisaged in the KAON Factory contain significant "tails" with non-linear behaviour which will also affect operation of the accelerator and may have to be taken into account in the control algorithms. Distributed control is proposed wherever possible in the system architecture to provide local signal filtering and conditioning, and local alarm handling. Control of the intense beam, as opposed to control of the accelerator equipment, is much more centralized in nature [7]. Model driven and artificial intelligence techniques for tuning and fault diagnosis add significantly to the "network bandwidth" requirements. A Domain-Driven Specification Method [8,9] was used to develop and extract the require-ments of the KAON Factory control system. The control system was isolated from the details of each device's operation by assuming that each device sub-system could be mod-elled with a high level interface such as might be provided by an intelligent controller embedded within the device. To manage the complexity of specifying the proposed control system, a sequence of three abstract models of the control system and one real model were defined and studied. Earlier models are included in later models by construction. The 4 models and their properties, in the order of their creation are: • The Ideal Machine: in which all control system software and all KAON Factory hardware are assumed to be perfect. In addition all device behaviour and settings are known. From this model the fundamental operating modes of the facility and the overall structure of the control system database are established; • The Beam Physics Machine: in which the knowledge of machine behaviour is no longer assumed to be complete. All device settings and operational algorithms must, therefore, be determined by an experimental commissioning activity. It is in this domain of study that the requirements of the beam physicists significantly influence the design of the control system. The functionality to perform and analyze the beam dynamics experiments was added to the Ideal Machine model of the control system to derive the Beam Physics Machine model; • The Equipment Machine: in which errors in sub-system construction and possible device faults are allowed. The present charter's focus on integrated modes of oper-ation resulted in the exposure of classes of faults rather than individual faults, thus 9-7 requiring the subsystem intelligent controller to respond to individual. faults indepen-dent of the control system. FUnctionality was added to the Beam Physics Machine model, in the form of the EMS, to derive the Equipment Machine that was capable of handling the new fault conditions; • The Real Machine: in which equipment device interfaces are allowed to be faulty. The final system specification is represented by models of the Real Machine, which must contain the full requirements of the control system, including control logic fault detection and recovery. By constructing the above four models in sequence, complex practical issues such as those involved in handling equipment failures were delayed until the more important fundamental operating requirements of the control system were identified. ACCELERATOR OPERATOR I Flashl Start~ Attn : Stop \ I Beam \ I KAON Factory \ Equipment Subsystem Settings Spill Alert! / / Vlolation/ / RADIATION MONITORING KAON FACTORY CENTRAL CONTROL SYSTEM / / / Radl4ltion Levels KF Selecte Status Schedule BEAM Maintenance SCHEDULER Requests ---- --Safety Device Statu!; Info Selection CMD Experiment Figure 9.4.1: Context Diagram of the Control System The context diagram of the control system is shown in Fig. 9.4.1. The circle represents the control system to be constructed while the rectangles (termed terminator objects) represent external systems with which the control system communicates. The terminator objects are the sources and destinations of all control and information flows required by the control system. The solid arrows indicate flows of information and the dashed lines 9-8 indicate flows of control. Single arrowhead denote discrete transfers while double-headed arrows denote continuous flows. The circle representing the control system can be "expanded" to reveal progressively more detail of the entire control system. Relationships between the objects in the expanded context diagram are identified and documented in the Entity Relationship Diagram (ERD) shown in Fig. 9.4.2. The ERD is the primary mechanism used in designing the database for the entire KAON Factory. FOIL CAROUSEL IPO IPO = Is Part Of KAON PRODUCTION MODE SECTION MODE SUB-SYSTEM MODE SETTINGS Figure 9.4.2: Abridged Entity Relationship Diagram (ERD) for the KAON Factory. The diagram is a model of the information to be stored by the control system for its correct operation. The information describes the state and properties of both physical objects (magnets, rf, diagnostics, etc.), abstract entities (modes of beam operation, schedules, etc.) and the static and dynamic relationships between them. The Radiation Monitoring and the Access Control terminator objects in Fig. 9.4.1 represent the control system's interface to the KAON Factory safety system. 9-9 9.5 Algorithms The settings of the KAON Factory devices are predominantly static in nature, from the point of view of the BCS. Programmable devices (e.g. rf programs) are here considered static in the sense that these programs (value vs. time) are sent to the device infrequently, where they are repeatedly executed. These settings comprise the data content in the data-stores shown Fig. 9.4.2, and represent a major portion of the storage requirements for the control system. The distributed databases, Control and Data Interconnects, programmed run-up sequences (or scripts), and the device control processors in the hardware configu-ration (Fig. 9.3.1) provide the mechanisms by which these datastore values are sent to the subsystem equipment to put them into operation. To determine the complete list of device settings during the commissioning [10] of the rings, the following beam optics measurements are required: • phase advance per cell "p1:,11 • lattice functions /31:,11 • dispersion function D(s), D'(s) • emittance f( X, y) • coherent and incoherent tune (vertical and horizontal) • closed orbit distortion • chromaticity • x - y coupling These experiments are analyzed semi-ofF-line in a modelling processor, provided they have access to the results of primary measurements carried out in the front-end processors as follows: • n triggered consecutive snapshots of the beam position vs. s (distance around the ring). The snapshot is of the entire ring. Both n, the number of measurements, and the triggering times are programmable (via the MTS), and may occur in conjunction with a programmable transverse kick. • measurement of the beam position w.r.t. time (single shot or filtered). • n (variable) triggered beam width measurements for a (few) selected locations around each ring. 9-10 • measurement of the beam momentum for each accelerator. • measurement of the beam current or charge per pulse at each accelerator or ring. The organized collection of these primary algorithms form the core of the Beam Control System outlined above. They form the basis around which is built the set traditional application codes is built. These algorithms lead to specification of the beam diagnostics requirements and the beam instrumentation support system (BISS) described above. The identification and location of systematic errors in the placement or calibration of accel-erator equipment are fundamental requirements of the control system. These requirements are increasingly being dealt with using rule-based expert systems [11]. The correction of the errors described above require data and control of devices at geographically dispersed locations (inter- and intra-ring). These coordinated activities, which are effected through cooperating device control processors, and the implementation of simple (few kHz) site-wide closed loops between arbitrary devices in the facility accentuate the need for the fast Control Interconnect. The network bandwidth provided by the selected implementation must be sufficient to accommodate these and future similar control demands. References [1] Proposal for a European Hadron Facility, J. Crawford, ed. pp.341-346 [2] The Physics and a Plan for a 45 Ge V Facility that Extends the High-Intensity Capa-bility in Nuclear and Particle Physics, LAMPF report LA-10720-MS [3] M. Lee, private communication [4] Dawson et al., A Conceptual Design for the TRIUMF KA ON Factory Control System, TRIUMF report TRI-87-1. [5] Dohan et al., The TRIUMF KAON Factory Control System Project Definition Study, submitted for publication in the Proceedings of the International Conference on Ac-celerator and Large Accelerator Control Systems, Vancouver, 1989 [6] B.Frammery, Timing for the KA ON Factory, TRI-DN-89-K30. [7] Parker et al., A New Data Communication Strategy For Accelerator Control Systems, ibid. [8] Inwood et al., Domain Driven Specification Techniques Simplify the Analysis of Re-quirements for the KAON Factory Central Control System, ibid. 9-11 [9] Osberg et al., Dynamic Object Modelling as Applied to the KA ON Control System, ibid. [10] Fischer et al., Principles, Contents, and Conventions of the Analysis made by the LEP Applications Analysis Working Group, CERN/LEP-TH/GG-ek. [11] Brand et al., Coupling Models and Expert Systems On-Line to Accelerators, submitted for publication in the Proceedings of the International Conference on Accelerator and Large Accelerator Control Systems, Vancouver, 1989. 9-12 ALIGNMENT Chapter 10 10 ALIGNMENT 10-1 10.1 Preliminary Remarks. 10-1 10.2 Surface Geodetic Network 10-1 10.3 The In-Tunnel Network and Positioning of the Machine Elements. 10-4 10.3.1 Fiducialization of the Magnets .. . . . . . . 10-7 10.3.2 Data Collection and Instrument Calibration. 10-7 10.3.3 Survey Assistance to the Civil Engineering . 10-7 10 ALIGNMENT 10.1 Preliminary Remarks In order to position the more than 1800 magnets forming the rings and transfer lines of the KAON Factory, the x, y and z coordinates, and the roll, must be measured at the entrance and exit of each magnet. Each magnet will have a set of reference targets (fiducials) whose location will be known ±0.05 mm relative to the physical profile of the magnet poles. The magnet position measurements will be made to these fiducials by referencing to an in-tunnel first order geodetic network. The in-tunnel network is tied to a first order geodetic surface network through a number of surface-to-tunnel penetrations as illustrated in Fig. 10.2.1.[1] 10.2 Surface Geodetic Network The surface network will serve to position in plan all of the underground elements of the Booster ring, Main ring, transfer lines, experimental areas and the existing cyclotron with respect to one another. The accuracy to which these elements can be positioned is highly dependent upon the accuracy to which the surface network is determined. Thirteen monuments located around the KAON Factory site will form the surface network shown on Fig. 10.2.2. Seven of these monuments will be mounted on concrete filled steel piles driven 5 to 10 meters into the ground. The other six monuments will be erected directly above PEP-style tunnel floor monuments using the vertical tunnel penetrations. Two-PEP style monuments in the Booster tunnel and four in the Main tunnel will be shared between the surface and in-tunnel networks. Six of the concrete pillars will be located as far from the construction site as possible to maximize their stability. They will be built and their positions determined by trilateration as soon as possible after the construction site is cleared. These monuments will serve as reference points for the civil engineering works. In all measurements of the surface network all distances between stations will be measured in both directions, and the measurements adjusted using least squares. The network of the six pillars and the two Booster ring access point monuments will be measured by trilateration and adjusted to transfer the surface x, y coordinates to the Booster ring floor monuments. The entire thirteen station surface network will be measured and adjusted to transfer the surface coordinates to the main ring floor monuments. 10-1 8000 o lMO '000 t""I""I""I"'" mrn 500 1500 Figure 10.2.1: Main ring tunnel cross section showing surface penetrations and crowding 10-2 a 100 200 Figure 10.2.2: The surface geodetic network 10-3 All distances in the surface network will be measured with a Kern Mekometer ME5000 which has an accuracy of ±0.2 mm + 0.2 mm/km. The instrument and reflector constants will be determined before each network measurement. In order to avoid. excessive errors resulting from atmospheric refraction, special care will be taken in the choice of the timing of the network measurements and in the meteorological corrections employed. Simulations of this network using a standard deviation for the distance measures of 0.3 mm have shown that the positional uncertainty in x, y of these surface monuments should not exceed 0.8 mm. 10.3 The In-Tunnel Network and Positioning of the Machine Elements The most important factor influencing the alignment procedures to be used with the KAON Factory is the crowded nature of the tunnels. This is particularly important when one considers that the positions of beam elements will certainly have to be redetermined on a regular basis and clements may occasionally have to be replaced. In the Main tunnel the Collector ring and the Driver ring are stacked vertically above the other, and the Extender ring is offset horizontally from the Driver ring. The BC transfer line runs above the Collector ring for about 140 meters and the DE transfer line crosses the walkway between the Driver and the Extender Rings. This crowding, as well as the cable trays and other services, prohibits the use of wall mounted survey brackets. The small vertical spaces between the tunnel roof, the BC transfer line, the Collector ring, and the Driver ring are not sufficient to place survey instruments directly on the machine elements. This precludes using the machine elements as a platform for the in-tunnel geodetic network. Removable rigid steel tripods, bolted to the floor in the walkway, will be used as the network base. When the tunnel is complete and has settled, the in-tunnel network will be measured. It will be measured in sections between two PEP style floor monuments, which are supported from the surface network, · by traversing along the steel tripods. Each traverse will be adjusted by a least squares fit holding the PEP style monuments as known points. The rigid steel tripods will then be used in the positioning, alignment and smoothing of the magnet rings. The surface planimetric x, y coordinates will be transferred to the tunnel level using an op-tical plummet accurate to 1:200,000. This will insure adequate planimetric liaison between surface and the tunnel control points approximately 11 m below. 10-4 The sequence of measurements of the tunnel geodetic network traverse is outlined in Fig. 10.3.1. Angles will be measured with Wild T3000 theodolites, taking measures on both faces, and with two circle readings. Distances will be measured at the same time with the Wild DI2000 electronic distance measuring apparatus (EDM). The EDMs used in these and other distance measurements will be calibrated frequently for instrument and prism constants, including cyclic error. It is expected that systematic errors will be minimized by using this procedure. The great-est uncertainty will be in the positions of the tripods due to the random errors associated with their removal and reinstallation. Simulations of the traverse outlined in Fig.lO.3.l using a standard error of 3.0 gon-seconds for the angle measurements and 0.5 mm for the distance measurements indicate that the positional uncertainty of the geodetic tripod network will not exceed 0.45 mm, and that the absolute long range tunnel geodetic net-work error envelope (i.e. the combination of the uncertainties in the surface network, the transfer of the surface coordinates to the tunnel level and the traverse through the tunnel) should not exceed 1.0 mm. The KAON Fact~ry site is sufficiently small (less than 1 km2 ) that a spherical earth model will be employed for all altimetric calculations and transformations between the z ordinates of the accelerator coordinate system and the true heights above the reference spheroid. High precision levels, invar scales and geodetic level traverses will ensure an altimetric accuracy of 0.2 mm/km. Where magnets are required to be tilted a high precision tilt meter such as the Talyvel level will be used. Smoothing and final positioning of the elements will follow the scheme outlined in Fig. 10.3.2. Once again using standard deviations of the angle and distance measurements of 3.0 gon-seconds and 0.5 mm respectively, simulations of this scheme show that the absolute long range error envelope of the magnet ring positions should not exceed 0.8 mm. More im-portantly, however, they show that short range radial (i.e. perpendicular to the beam direction) relative errors between neighboring elements will not exceed 0.1 mm. Once again, the use of high precision levels and invar staffs will insure that the vertical toler-ances between adjacent magnet elements of 0.1 mm will be met. 10-5 Tripod i+3 Tripod 1+2 Tripod i+4 -------@ Tripod occupied by theodolite. EOM. target, and prism. o Tripod occupied by target and prism only. ~ Angle and distance measurements. Figure 10.3.1: The measurement scheme of the in-tunnel rigid steel tripod network • Tripod occupied by theodolite and £OM. Quadrupole (and other machine components) . Angle and distance measurement. 4.ngle measurement only_ Figure 10.3.2: The measurement scheme of the positioning and smoothing of the magnets 10-6 10.3.1 Fiducialization of the Magnets The liaison between the geometric axes of the magnets and the fiducials used in the align-ment of the machine will be accomplished with a laser tracking system. This system employs laser interferometers and a digital angle encoding device. It is expected that this operation will give results in the order of 0.05 mm. 10.3.2 Data Collection and Instrument Calibration The huge amount of data to be collected and adjusted necessitates the use of automation and computer control of instruments and recording procedures. This is particularly impor-tant in reducing errors and optimizing the rate of installation and alignment. Computer programs are available to perform the measurement adjustments, the earth curvature cor-rections, the position calculations and the error analysis. Such programs are now being used at Stanford and CERN. A 20 m to 30 m bench equipped with an interferometer will be built for the calibration of the prism and EDM constants of the Kern Mekometer ME5000 and the Wild DI2000, and for the determination of the cyclic error associated with the DI2000. 10.3.3 Survey Assistance to the Civil Engineering The survey/alignment group will assist in the control of the posi tioning of the tunnel as it is constructed and will provide the coordinates of preliminary survey marks that will be placed at regular intervals along the interior of the tunnel. The tunnel axis and the floor altitude will be controlled by profiles and levels from these points. They will later be used to position the rigid steel tripods that will be used in the placement and alignment of the machine components. References [1] D.G. Martin and G.S. Clark ,The Survey and Alignment of the KAON Factory Project, TRI-DN-89-K103, 1989. 10-7 PARAMETER TABLES Chapter 11 11 PARAMETER TABLES Table 1.1.1 Table 1.2.1 Table 2.1.1 Table 2.1.2 Table 2.1.3 Table 2.1.4 Table 2.1.5 Table 2.2.1 Table 2.2.2 Table 3.1.1 Table 3.1.2 Table 3.1.3 Table 3.2.1 Table 3.2.2 Table 3.3.1 Table 3.3.2 Table 4.2.1 Table 5.1.1 Table 5.1.2 Table 5.1.3 Table 5.2.1 Table 5.2.2 Table 5.3.1A Table 5.3.1B Table 6.2.1 Table 7.4.1 Table 7.5.1 Table 8.1.1 Table 8.2.1 Table 8.3.1 Table 8.3.2 Table 9.3.1 Medium-Energy Proton Synchrotrons Synchrotron Design Parameters Lattice Parameters for the Five Rings Emittances and Momentum Spread Beam Stay-Clear Apertures Emittance Growths Due to Synchro-Betatron Resonances in the Booster Misalignment Orbit Excursions, and Corrector Strength Magnet Parameters for Injection System Slow Extraction Simulation Magnet Quantities Magnet Parameters for the Rings Prototype Booster Magnet Parameters Resonant Circuit Parameters Power Supplies for Accelerator Magnets Injection and Extraction Kicker Magnet Parameters Diagnostic Kicker Magnet Parameters Summary of RF Systems and Feedback Systems Transverse Impedances (in MQ/m) and Thresholds for the Fast Instability Impedances and Growth Rates for the Resistive Wall Effect Transverse Impedances and Growth Rates Due to the Kicker Magnets Summary of Horizontal Dampers Summary of Vertical Dampers Summary of Beam Instrumentation Summary of Beam Instrumentation Beam Pipe and Vacuum Equipment List Collimator Shield Dimensions Operational/ Commissioning Beam Dumps Distribution of the Current from TRIUMF Cyclotron to Achieve 100 /-lA at 30 Ge V Specifications for Extraction Devices Extracted Radial Emittances for Various Conditions Extracted Beam Parameters Controls System Component Breakdown TRllJotF KAON Factory 11-APR-9O ACCUMULAT~ BOOSTER COlLECT~ DRIVER EXTENDER Ci rcunf. (m) 215.6600 215.6600 1078.3000 1078.3000 1102.2600 Berd Radius (m) 3.8390 11.4080 15.2800 74.6900 59.5900 Ganma trans. 4.0000 15.0000 1000.0000 -36.0000 -36.0000 Hanronic No. 45.0000 45.0000 225.0000 225.0000 230.0000 Qx 3.7400 5.2400 13.2300 13.2300 13.3300 Qz 5.7200 7.2200 14.1900 14.1900 14.2000 Chamer ~./2 (m) 0.0590 0.0620 0.0620 0.0670 0.0560 Chamer H./2 (m) 0.0430 0.0440 0.0350 0.0410 0.0210 ~ll Z/n «(ffl) 5.0000 5.0000 5.0000 5.0000 5.0000 Frep (Hz) 50.0000 50.0000 10.0000 10.0000 10.0000 Vrf (kV) 514.0000 750.0001 1846.0001 2550.0000 849.0001 RMS Disp. (m) 2.8090 3.0460 2.1190 2.3540 2.2110 Rise Fraction 0.0000 0.5000 0.0000 0.7500 0.0000 Ranp Type 0.0000 2.0000 0.0000 2.0000 0.0000 Tinj (GeV) 0.4502 0.4502 3.0000 3.0000 30.0000 Final T (GeV) 30.0000 GLOBAL PARAMETERS: 0.312E+12 Particles per bunch SEXN (pi 1TTlt"1Tr) 30.00 SEZN ( 1\) 3O.GO EL (eV-s) 0.048000 EL blDW..4' factor 4.000000 in COLLECT~ ACCUMULATOR Frep= 50.Hz, rise=.oo, C= 216.m, h= 45., Gt= 4.0, indZ/n= 5.00hm, Nb:O.312E+12~ Qx= 3.74, Qz= 5.72J HHt=.043m lmage coefrs: e1z,e2z,x1x,x1z= .13 .41 .22 .56 T B Frf Vrf Phis Phib lac Pow Phi1 Phi2 Bf Pf dp/p Qs srkt srkc srkb jZ/n IIIIIIIIIIIII -dQ \\\\\\\\\\\\\ (Gev) (T) (MHz) (kV) (deg)(deg)(~)(MW) (deg) (deg) (%) (ohm) xinc zinc xcoh zcoh xcloc zcloc 0.45 0.89 46.11 520. 0.0 0.03.4 0.00-100.3 100.3 .350 .737 .348 .0402 0.912 .862 0.739347. 0.163 .172 0.008 .023 .0122 .0204 0.45 0.89 46.11 520. 0.0 100.3 .350 .737 .348 .0402 0.912 .862 0.739347. 0.163 .172 0.008 .023 .0122 .0204 BOOSTER Frep= 50. Hz, rise=.50, C= 216.m, h= 45., Gt= i5.0, indZln= 5.00hm, Nb=O.312E+12~ Qx= 5.24, Qz= 7.22J HHt=.044m Ramp=sinusoldal image coefrs: elz,e2z,xlx,xlz= .13 .41 .20 .57 T B Frf Vrf Phis Phib lac Pow Phi1 Phi2 Bf Pf dp/p Qs srkt srkc srkb jZ/n IIIIIIIIIIIII -dQ \\\\\\\\\\\\\ (GeV) (T) (MHz) (kV) (deg)(deg)(~)(MW) (deg) (deg) (%) (ohm) xinc zinc xcoh zcoh xcloc zcloc 0.450.30 46.11 588. 0.0 .351 .740 .347.04590.914 .861 0.739389. 0.141 .168 0.004 .018 .0077 .0154 0.450.30 46.17 588. 1.9 -97.1 104.0 .351 .761 .346 .04570.914 .861 0.739388.0.140 .1670.004 .018 .0076.0153 0.46 0.30 46.34 592. 3.8 -93.3 106.9 .349 .780 .345 .0454 0.913 .862 0.738 384. 0.139 .166 0.004 .018 .0075 .0151 0.470.31 46.61 615. 5.5 -88.8 1OB.1 .344 .789 .346 .0454 0.914 .8650.741377.0.138 .1650.004 .017 .0074 .0148 0.48 0.31 46.99 651. 6.9 9.1 3.60.18 -84.3 107.9 .336 .790 .348 .0456 0.914 .870 0.747369. 0.136 .163 0.004 .017 .0072 .0145 0.500.3247.45 681. 8.3 -80.3107.6 .329 .790 .347.04530.915 .8740.752358. 0.133 .160 0.003 .016 .0069 .0139 0.520.33 47.98 705. 9.5 -76.5 107.3 .323 .789 .345 .0444 0.914 .878 0.756 346.0.129 .1560.003 .016 .0066 .0133 0.550.34 48.58 723. 10.713.73.80.33 -73.0 107.1 .316 .789 .342 .0431 0.914 .882 0.758332. 0.125 .1520.003 .015 .0062 .0125 0.58 0.35 49.23 736. -69.6 106.9 .310 .789 .337 .0415 0.913 .885 0.761 317. 0.121 .146 0.003 .014 .0058 .0117 0.62 0.36 49.91 745. -66.3 106.6 .305 .789 .331 .0396 0.912 .888 0.762 301.0.116 .1400.002 .013 .0054 .0108 0.66 0.38 50.60 750. 14.417.94.1 0.47 -63.1 106.4 .299 .789 .325 .03760.911 .891 0.764 285.0.111 .134 0.002 .012 .0049.0099 0.70 0.39 51.30 750. 15.6 -60.0 106.3 .293 .789 .317 .0354 0.909 .894 0.764 269. 0.106 .1280.001 .011 .0045 .0091 0.75 0.41 52.00 750. 16.820.64.30.56 -56.9 106.0 .288 .788 .310 .0333 0.908 .896 0.765 254.0.101 .121 0.001 .010 .0041 .0082 0.80 0.43 52.68 750. 17.921.84.40.61 -53.9105.5 .282 .786 .304 .03120.906 .899 0.766 238.0.095 .1150.001 .010 .0037.0075 0.86 0.45 53.34 750. -51.0104.7 .275 .782 .297 .0292 0.905 .9020.768 223.0.090 .109 0.001 .009 .0034 .0068 0.92 0.47 53.97 750. -48.2 103.8 .269 .776 .291 .02740.903 .905 0.770208.0.086 .1030.000 .008 .0031 .0062 0.98 0.49 54.57 750. 20.924.84.60.73 -45.5 102.8 .263 .769 .285 .0256 0.901 .908 o.m 195. 0.081 .0980.000 .007 .0028 .0056 1.05 0.5155.13 750.21.725.64.70.76 -42.9 101.6 .257 .761 .280 .0240 0.900 .9120.775 182. 0.077 .0920.000 .007 .0025 .0051 1.12 0.53 55.65 750. 22.5 -40.5 100.3 .251 .751 .274 .0224 0.898 .915 0.778 170. 0.073 .0870.000 .006 .0023 .0046 1.190.56 56.14 750.23.1 -38.3 98.9 .244 .740 .269 .02100.896 .9180.781 158.0.069 .083 0.000 .006 .0021 .0042 1.260.58 56.59 750. 23.727.35.00.85 -36.2 97.4 .238 .728 .265 .01970.895 .922 0.784 148. 0.065 .079 0.000 .005 .0019 .0038 1.34 0.6157.00 750. -34.2 95.8 .232 .715 .261.0185 0.893 .925 0.788 138.0.062 .075 0.000 .005 .0017.0035 1.420.63 57.38 750.24.527.95.1 0.89 -32.4 94.1 .227 .701 .257.01740.892 .9280.792129.0.058 .071 -.001 .005 .0016.0032 1.500.66 57.73 750. 24.828.05.20.91 -30.8 92.4 .221 .686 .253 .0164 0.890 .9320.795 121. 0.055 .067 -.001 .004 .0014 .0029 1.58 0.68 58.05 750. 24.928.0 5.2 0.92 -29.4 90.6 .216 .670 .250 .0154 0.889 .935 0.799 113. 0.053 .064 -.001 .004 .0013 .0027 1.66 0.71 58.34 750. -28.1 88.8 .211 .653 .247.0146 0.887 .938 0.802 107.0.050 .061 -.001 .004 .0012 .0024 1.74 0.73 58.60 749. 24.927.75.30.93 -27.0 87.1 .206 .637 .244 .0138 0.886 .941 0.806 100. 0.048 .059 -.001 .004 .0011 .0023 1.83 0.7658.84 744. -26.0 85.6 .201 .623 .241 .0131 0.884 .944 0.808 94. 0.046 .056 -.001 .003 .0010 .0021 1.91 0.79 59.06 737. -25.0 84.2 .198 .610 .238 .01240.883 .946 0.810 89. 0.044 .054 -.001 .003 .0010 .0019 1.99 0.81 59.25 726. -24.2 83.0 .194 .598 .235 .01170.881 .948 0.811 84. 0.042 .052 -.001 .003 .0009 .0018 2.070.84 59.43 713. -23.5 82.0 .191 .588 .232 .0111 0.880 .9490.812 80.0.040 .049 -.001 .003 .0008 .0017 2.15 0.86 59.59 697. 0.88 -22.9 81.1 .188 .579 .228 .0105 0.878 .951 0.813 76. 0.038 .047 -.001 .003 .0008 .0016 2.22 0.88 59.74 679. 0.86 -22.4 80.3 .186 .571 .225 .0099 0.876 .952 0.813 72. 0.037 .046 -.001 .003 .0007 .0015 2.30 0.91 59.87 657. 0.83 -22.0 79.7 .185 .565 .221 .0093 0.874 .953 0.812 69. 0.036 .044 -.001 .002 .0007 .0014 2.370.93 59.99 633. 25.0 27.1 5.60.80 -21.7 79.3 .183 .560 .217 .0088 0.872 .954 0.811 66. 0.034 .042 -.001 .002 .0006 .0013 2.44 0.95 60.10 607.25.027.1 5.60.77 -21.5 79.0 .182 .556 .214 .0083 0.870 .954 0.810 63.0.033 .041 -.001 .002 .0006 .0012 2.51 0.9760.19 583.24.726.85.60.73 -21.5 78.4 .182 .550 .210 .0079 0.868 .955 0.809 61.0.032 .039 -.001 .002 .0006 .0011 2.570.99 60.28 558.24.426.55.60.70 -21.8 77.9 .181 .545 .207.0075 0.866 .955 0.808 59.0.031 .038 -.001 .002 .0005 .0011 2.63 1.01 60.36 533. -22.2 77.3 .181 .539 .203 .0071 0.864 .956 0.807 57.0.030 .037 -.001 .002 .0005 .0010 2.69 1.02 60.43 508. 23.4 25.3 5.6 0.61 -22.8 76.6 .181 .532 .200 .0068 0.862 .956 0.806 55. 0.029 .036 - .001 .002 .0005 .0010 2.74 1.04 60.49 485. 22.6 24.4 5.6 0.56 -23.8 75.5 .181 .523 .197 .0065 0.860 . C157 0.806 54. 0.029 .035 - .001 .002 .0005 .0009 2.79 1.05 60.54 463. 21.5 23.2 5.6 0.51 -25.0 74.3 .181 .513 .194 .0062 0.858 • C157 0.805 52. 0.028 .034 - .001 .002 .0004 .0009 2.83 1.0760.59 444. 20.1 -26.6 72.6 .181 .500 .192 .0060 0.857 .958 0.806 51. 0.028 .033 -.001 .002 .0004 .0009 2.871.08 60.63 426. 18.4 -28.5 70.7 .181 .486 .190 .0058 0.855 .C159 0.806 50. 0.027 .033 -.001 .002 .0004 .0008 2.90 1.09 60.66 411. 16.5 -30.7 68.3 .181 .470 .188 .0056 0.854 .960 0.807 49. 0.027 .032 -.001 .002 .0004 .0008 2.93 1.1060.69 397. 14.3 -33.3 65.8 .181 .453 .187.00550.853 .961 0.808 .48. 0.026 .032 -.001 .002 .0004 .0008 2.96 1.11 60.71 386. 11.812.75.60.24 -36.1 62.9 .181 .435 .186 .0054 0.852 .962 0.809 48.0.026 .031 -.001 .002 .0004 .0008 2.98 1.11 60.73 377. 9.1 -39.2 59.8 .181 .417.185 .00530.852 .963 0.810 47.0.026 .031 -.001 .002 .0004 .0008 2.99 1.1260.74 371. 6.1 -42.5 56.4 .181 .399 .184 .00520.851 .963 0.811 47.0.026 .031 -.001 .002 .0004 .0008 3.00 1.1260.75 367. 3.1 -45.9 53.0 .181 .381 .184 .00520.851 .964 0.811 47.0.026 .031 -.001 .002 .0004 .0008 3.00 1.1260.75 365. 0.0 -49.4 49.4 .181 .365 .183 .00520.851 .964 0.811 47.0.025 .031 -.001 .002 .0004 .0008 COLLECTOR Frep= 10.Hz, rise=.OO, C=1078.m, h=225., Gt=*****, indZ/n= 5.00hm, Nb:O.312E+12~ Qx= 13.23, Qz= 14.19 HHt=.035m Image eoefts: e1z,e2z,x1x,x1z= • ~6 .41 .11 .59 T B Frf Vrf Phis Phib lac Pow Phi1 Phi2 Bf Pf dp/p Qs srkt srke srkb jl/n 1111111111111 -dQ \\\\\\\\\\\\\ (GeV) (T) (MHz) (kV) (deg)(deg)(AIlp)(~) (deg) (deg) (X,) (ohm) xinc zinc xeoh zeoh xeloc zcloc 3.00 0.83 60.75 1846. 0.0 -49.2 49.2.180 .377 .184 .0283 0.889.964 0.848 35.0.148 .195 -.013 .031.0032 .0158 ***Blow Lp: EL SEXN SEZN = 0.192 30.0 30.0 3.00 0.83 60.~ 1846. 0.0 0.04.5 0.00 -99.7 99.7 .349 .759 .374 .03130.984 .863 0.799 35.0.048 .099 -.014 .023 .0016 .0081 DRIVER Frep= 1o.Hz, rise=.75, C=1078.m, h=225., Gt=-36.0, indZ/n= 5.00hm, Nb=O.312E+12~ Qx= 13.23, Qz= 14.19 HHt=.041m Ramp=Slnusoldal image eoefrs: e1z,e2z,x1x,x1z= .~5 .41 .14 .58 T B Frf Vrf Phis Phib lac Pow Phi1 Phi2 Bf Pf dp/p Qs srkt srke srkb jl/n 1111111111111 -dQ \\\\\\\\\\\\\ (GeV) (T) (MHz) (kV) (deg)(deg)(Amp)(~) (deg) (deg) (%) (ohm) xinc zinc xeoh zeoh xeloc zeloc 3.00 0.1760.75 1847. 0.0 .350 .762 .373 .0315 0.983 .862 0.797 38. 0.042 .098 -.016 .022 .0015 .0059 3.03 0.1760.78 1919. 3.8 5.1 4.60.39 -92.0 105.3 .345 .793 .376 .03170.983 .8650.800 37.0.042 .097 -.016 .022 .0015 .0058 3.100.1860.852091. 7.0 -83.2 106.7 .333 .804 .382 .0320 0.983 .873 0.808 36.0.041 .096 -.015 .021 .0014 .0057 3.23 0.1860.95 2225. 9.912.74.81.17 -75.3107.0 .320 .809 .384 .03140.983 .880 0.814 34.0.039 .093 -.015 .021 .0013 .0053 3.41 0.1961.08 2332. 12.615.84.91.55 -67.8106.5 .307.809 .384 .0300 0.983 .8870.821 -.014 .020 .0012 .0049 3.65 0.20 61.23 2396. 15.3 18.95.0 1.93 -60.7106.0 .294 .809 .380 .0279 0.983 .894 0.827 28. 0.034 .083 -.014 .018 .0011 .0043 3.93 0.21 61.38 2433. 18.021 .95.1 2.31 -53.7105.4 .281 .809 .373 .0255 0.983 .899 0.832 24. 0.030 .076 -.013 .017 .0010 .0038 4.26 0.23 61.53 2447. 20.824.95.22.67 -46.8 104.8 .268 .809 .366 .0229 0.983 .905 0.835 21. 0.027 .070 -.012 .015 .0008 .0032 4.64 0.25 61.66 2450. 23.727.95.33.03 -40.1 104.3 .256 .809 .357 .0204 0.983 .909 0.838 18. 0.024 .064 -.012 .014 .0007 .0027 5.070.2761.79 2452. 26.530.95.43.38 -33.6 103.7 .243 .809 .347 .0181 0.983 .913 0.841 15. 0.021 .058 - . 011 .013 .0006 .0023 5.54 0.29 61.90 2456. 3.71 -27.5 103.0 .231 .809 .338 .0159 0.983 .9170.843 13. 0.019 .052 -.010 .012 .0005 .0019 6.06 0.31 61.99 2457. -21.6 102.6 .220 .809 .329 .0140 0.984 .920 0.845 10. 0.016 .047 -.009 .011 .0004 .0016 6.62 0.33 62.072465.34.538.95.64.34 -16.1 102.0 .210 .809 .320 .01240.984 .923 0.847 8.0.014 .042 -.009 .010 .0003 .0013 7.22 0.36 62.14 2473. -10.9 101.4 .200 .809 .311 .0109 0.985 .925 0.848 7. 0.012 .038 -.008 .009 .0003 .0011 7.85 0.3962.20 2484. 39.343.65.7 4.f!B -6.2 100.9 .190 .809 .302 .00970.986 .9270.850 5. 0.011 .035 -.008 .008 .0002 .0009 8.520.4262.25 2495.41.445.65.85.14 -1.8100.4 .182 .809 .294 .0086 0.988 .929 0.852 4.0.009 .031 -.007.008 .0002 .0C08 9.22 0.45 62.29 2507. 43.447.5 5.85.37 2.3 100.0 .174 .809 .286 .0077 0.990 .931 0.855 3. 0.008 .029 -.007 .007 .0002 .0007 9.95 0.48 62.32 2518. 45.349.35.85.57 6.0 99.5 .166 .809 .279 .0069 0.992 .9320.857 2. 0.007 .026 -.006 .006 .0001 .0006 10.70 0.5262.35 2527. 9.4 99.1 .159 .809 .271 .0062 0.994 .934 0.860 1. 0.006 .024 -.006 .006 .0001 .0005 11.48 0.55 62.38 2539. 48.452.1 5.95.92 12.4 98.3 .153 .805 .265 .0056 0.997 .935 0.866 O. 0.006 .022 -.005 .006 .0001 .0004 12.280.5962.40 2545.49.753.45.96.06 15.1 97.8 . 147 .803 .259 .0051 1.001 .9370.870 0.0.005 .020 -.005 .005 .0001 .0003 13.100.63 62.42 2550. 50.854.36.06.17 17.5 97.1 .142 .798 .253 .00471.005 .9390.877 -1. 0.004 .019 -.005 .005 .0001 .0003 13.93 0.66 62.43 2550. 51.855.26.06.26 19.7 96.4 .137 .793 .248 .0043 1.009 .940 0.884 -1. 0.004 .017 -.005 .005 .0001 .0003 14.77 0.70 62.44 2550. 52.655.86.06.32 21.5 95.4 .132 .784 .243 .0040 1.014 .943 0.f!B3 -1. 0.004 .016 .-.004 .004 .0001 .0002 15.61 0.7462.452550. 53.056.1 6.06.36 22.9 94.1 .128 .770 .239 .00371.019 .946 0.905 -2. 0.003 .015 -.004 .004 .0001 .0002 16.46 0.78 62.46 2550. 24.0 92.5 .123 .751 .236 .0035 1.024 .9490.918 -2. 0.003 .014 -.004 .004 .0000 .0002 17.31 0.81 62.472550. 24.7 90.7 .119 .727 .233 .0033 1.030 .9530.933 -2. 0.003 .013 -.004 .004 .0000 .0002 18.160.85 62.48 2548.52.655.06.16.33 25.1 88.6 .115 .699 .231 .00321.035 .958 0.949 -2.0.002 .012 -.004 .003 .0000 .0001 19.00 0.f!B 62.49 2532.52.454.66.16.26 25.5 87.0 .111 .678 .228 .0030 1.041 .961 0.962 -3.0.002 .012 -.003 .003 .0000 .0001 19.83 0.93 62.49 2506.52.1 54.1 6.1 6.18 25.8 85.5 .108 .659 .225 .0029 1.047 .964 0.974 -3.0.002 .011 -.003 .003 .0000 .0001 20.64 0.96 62.50 2470.51.853.76.16.07 26.1 84.2 .106 .642 .222 .00271.054 .966 0.985 -3.0.002 .011 -.003 .003 .0000 .0001 21.44 1.00 62.50 2424. 51.553.36.1 5.93 26.3 83.1 .103 .627 .219 .0026 1.060 .968 0.996 -3. 0.002 .010 -.003 .003 .0000 .0001 22.22 1.0362.502368. 51.353.06.1 5.77 26.4 82.0 .101 .614 .216 .0025 1.067.970 1.006 -3. 0.001 .010 -.003 .003 .0000 .0001 22.98 1.0762.51 2302. 5.59 26.5 81.1 .100 .603 .213 .0024 1.074 .971 1.016 -3.0.001 .009 -.003 .003 .0000 .0001 23.71 1.1062.51 2227. 50.752.36.1 5.39 26.5 80.3 .098 .594 .210 .0023 1.081 .972 1.025 -3.0.001 .009 -.003 .003 .0000 .0001 24.42 1.1362.51 2142. 50.452.06.1 5.16 26.5 79.7.097 .587 .206 .0022 1.088 .973 1.034 -4.0.001 .009 -.003 .003 .0000 .0001 25.09 1.1662.51 2048. 4.92 26.4 79.1 .096 .581 .203 .0021 1.096 .974 1.043 -4. 0.001 .008 -.003 .003 .0000 .0001 25.73 1. 19 62.52 1945. 49.9 51.4 6.1 4.65 26.2 78.6.096 .577 .199 .0020 1.103 .975 1.052 -4. 0.001 .008 -.003 .002 .0000 .0001 26.33 1.22 62.521831.49.751.26.14.37 26.1 78.3.096 .577 .195 .00191.111 .975 1.060 -4.0.001 .008 -.002 .002 .0000 .0001 26.f!B 1.2462.52 1718. 4.07 25.6 77.7.095 .572 .191 .00191.119 .976 1.069 -4. 0.001 .007 -.002 .002 .0000 .0001 27.41 1.2762.521606.48.349.76.1 3.75 24.6 76.7.095 .562 .188 .0018 1.126 .976 1.078 -4.0.001 .007 -.002 .002 .0000 .0001 27.f!B 1.29 62.521496. 23.2 75.2.095 .547.185 .00171.133 .978 1.088 -4.0.001 .007 -.002 .002 .0000 .0001 28.32 1.31 62.521388.45.1 46.36.1 3.07 21.2 73.1 .095 .527.183 .00171.139 .979 1.098 -4.0.001 .007 -.002 .002 .0000 .0000 28.71 1.3262.521284.42.643.76.1 2.72 18.6 70.3 .095 .502 .181 .00161.145 .981 1.108 -4.0.001 .007 -.002 .002 .0000 .0000 29.05 1.34 62.52 1184. 39.440.46.1 2.35 15.2 66.9.095 .472 .179 .0016 1.150 .983 1.118 -4. 0.001 .007 -.002 .002 .0000 .0000 29.34 1.3562.53 1092. 35.336.1 6.1 1.97 10.9 62.5 .095 .438 .177 .0016 1.155 .985 1.127 -4.0.001 .007 -.002 .002 .0000 .0000 29.571.36 62.53 1008. 1.59 5.7 57.1 .095 .401 .176.0015 1.158 .987 1.135 -4.0.001 .006 -.002 .002 .0000 .0000 29.761.3762.53 937.24.1 24.66.1 1.20 -0.7 50.7.095 .363 .175 .0015 1.161 .988 1.142 -4.0.001 .006 -.002 .002 .0000 .0000 29.f!B 1.38 62.53 883. 16.817.26.1 0.80 -8.2 43.1 .095 .325 .174 .0015 1.163 .9f!S 1.147 -4.0.001 .006 -.002 .002 .0000 .0000 29.971.38 62.53 848. 8.7 -16.7 34.6 .095 .289 .174 .0015 1.165 .990 1.150 -4.0.001 .006 -.002 .002 .0000 .0000 30.00 1.38 62.53 836. 0.0 0.0 6.1 0.00 -25.6 25.6.095 .256 .174 .0015 1.165 .990 1.151 -4. 0.001 .006 -.002 .002 .0000 .0000 EXTEM)ER Frep= 10.Hz, rise=.OO, C=1102.m, h=230., Gt=-36.0, indZ/n= 5.00hm, Nb:O.312E+12~ Qx= 13.33, Qz= 14.20J HHt= . 021m Image eoefrs: e1z,e2z,x1x,x1z= .19 .41 .04 .60 T B Frf Vrf Phis Phib lac Pow Phi1 Phi2 Bf Pf dp/p Qs srkt srke srkb jl/n 1111111111111 -dQ \\\\\\\\\\\\\ (GeV) (T) (MHz) (kV) (deg)(deg)(AIlp)(~) (deg) (deg) (X,) (ohm) xinc zinc xeoh zeoh xcloc zeloc 30.00 1.73 62.53 849. 0.0 0.06.1 0.00 -25.6 25.6.095 .257.174 .0015 1.173 .990 1.158 -4. -.006 .013 -.009 .009 .0000 .0002 30.00 1.73 62.53 849. 0.0 0.06.1 0.00 -25.6 25.6.095 .257.174 .0015 1.173 .990 1.158 -4. -.006 .013 -.009 .009 .0000 .0002 Table 1.1.1: Medium-Energy Proton Synchrotrons Average Rep. Protons/ Circulating Energy Current Rate Pulse N Current I (GeV) (JLA) (Hz) (x 1013) (A) Fast Cycling Argonne IPNS 0.5 14 30 0.3 4.0 Rutherford ISIS 0.75 130(200) 50 1.6(2.5) 4.0(6.1) AGS Booster (1.5) (20-40) (7.5) (1.8-3.5) ( 4- 8) Fermilab Booster 8 7 15 0.3 0.3 SSC LEB (11) (8) (10) (0.5) (0.5) Slow Cycling KEK PS 12 0.32 0.6 0.4 0.6 CERN PS 26 1.2 0.38 2 1.5 Brookhaven AGS 28.5 0.9 0.38 1.6 0.9 - with Booster (4) (0.38) (~) ( 4) Kaon Factories TRIUMF KAON 30 100 10 6 2.8 KAON Booster 3 100 50 1.2 2.7 Moscow KF 45 125 6.25 12.4 3.2 Moscow Booster 7.5 250' 50 3.1 3.2 1 Table 1.2.1: Synchrotron Design Parameters BOOSTER DRIVER Energy 3 GeV 30 GeV Circumference 4.5 X 211" Rr = 215.66 m . 22.5 X 211" Rr = 1078.30 m Current 100 J.LA = 6 X 1014/s 100J.LA = 6xl014/s Repetition Rate 50 Hz 10 Hz Charge/Pulse 2 J.LC = 1.2 X 1013 ppp 10 J.LC = 6xl013 ppp Long Straights 2 X 156 m No. Arc Superperiods 6 12 Lattice } { Focusing FODO FODO Structure Bending OBOBBOBO BBBB No. Focusing Cells 24 68 Maximum {3x X {3y 17.5 m X 15.4 m 38.0 m X 31.8 m Dispersion 1Jmax 7.2 m 7.4 m Transition it 15 30 i Tunes Vx X Vy 5.23 X 7.22 13.23 X 14.18 Space Charge 6.Vy -0.17 -0.10 Emittances }{ f nx X f ny 6011" X 6011" (J.Lm)2 10011" x 10011" (J.Lm)2 at Injection fl ong 0.048 eV-s 0.192 eV-s Harmonic 45 225 Radiofrequency 46.1-60.8 MHz 60.8-62.5 MHz Energy gain/turn 316 keY 2040 keY Max RF Voltage 750 kV 2550 kV RF cavities 12 x 75 kV 18 x 150 kV 2 Table 2.1.1: Lattice Parameters for the Five Rings Accumulator Booster Collector Driver Extender Energy (GeV) 0.45 0.45-3 3 3-30 30 Circumference (m) 215.66 215.66 1078.30 1078.30 1102.26 Repetition rate (Hz) de 50 de 10 de Intensity (JLC/pulse) 0-2 2.0 0-10 10 10-0 Number of cells/arcs 24 24 48 48 48 Number of cells/straight 10 10 8 Number of superperiods/arcs 6 6 12 12 12 Horizontal tune 3.73 5.24 13.23 13.23 13.23 Vertical tune 5.71 7.22 14.19 14.19 14.19 Max. hor. beta/arcs (m) 17.8 17.5 36.5 38.0 38.0 Max. vert. beta/arcs (m) 17.8 15.5 31.4 31.8 32.9 Max. hor. beta/str. (m) 36.5 38.0 100.8 Max. vert. beta/str. (m) 25.4 28.3 40.0 Max. dispersion (m) 5.2 7.2 5.2 7.3 6.9 Transition Energy It 4.0 15.0 00 36i 11-30i Dipoles Number 24 24 96 96 96 Length (m) 1.01 2.99 1.0 4.89 3.90 Min. field (T) 0.88 0.30 0.83 0.17 1.75 Max. field (T) 0.88 1.12 0.83 1.38 1.75 Quadrupoles (in arcs) Number 51 48 94 94 94 Length (m) 0.3 0.36/0.46 0.2/0.4 0.94-1.69 0.82-1.8 Strength dB/dx(T /m) 2.8-4.2 12.5/8.5 7.3-9.8 14.7 8.76-19.46 Quadrupoles (in straights) Number 42 42 34 Length (m) 0.2 0.75- 1.55 0.70- 1.60 Strength dB/dx(T /m) 7.5-1004 14.20-15.10 11.42-24.06 Long straight sections Number 6 6 2 2 2 Length (m) 12.5 8.8 154 154 154 Correction elements: Sextupoles Number 24 24 48 48 48 Length 0.2 0.2 0.2 0.2 0.2 Horiz. orbit correctors 68 68 64 length 0.1 0.2 0.2 Vert. orbit correctors 24 24 68 68 64 length 0.1 0.1 0.1 0.1 0.1 3 Table 2.1.2: Emittances and Momentum Spread A B C D E 4fq ,.1:( 1rmm-mrad) 45 60 20 24 25 4fq ,y( 1rmm-mrad) 45 60 20 24 3 6.p/p x 103 3.48 3.48 3.73 3.85 1.64 Bucket height X loJ 4.66 4.68 4.88 4.90 6.14 Table 2.1.3: Beam Stay-Clear Apertures (half-widths in em) A B C D E II V H V H V II V H V Arc quadrupoles j21Jf. 4.0 4.0 4.6 4.3 3.8 3.5 4.2 3.5 4.4 1.4· 1]( 6.p / p )max 2.4 3.4 3.0 2.5 1.1 c.o.d. 0.4 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 Total 6.9 4.5 8.5 4.8 7.3 4.0 7.2 4.3 6.0 1.9 Dipoles j21Jf. 3.2 3.6 4.0 3.8 3.1 2.7 3.9 3.6 3.9 1.2 17( 6. p / p )max 1.4 1.3 1.7 2.2 0.9 c.o.d. OA 0.5 0.5 0.5 0.4 0.3 0.5 0.4 0.4 0.4 Total 5.0 4.1 5.8 4.3 5.2 3.0 6.6 4.0 5.2 1.6 *) based on slow extra.ction dynamics 4 Table 2.1.4: Emittance Growth Due to Synchro-Betatron Resonances in the Booster m Horizontal Vertical 1 13% 7.5% 2 4.5% 2.5% 3 12% 6.7% 4 5.0% 2.8% 5 7.0% 4.0% Table 2.1.5: Misalignment, Orbit Excursions, and Corrector Strength Small rings Large rings units Displacement of elements rms 250 250 Jlm Roll error of elements rms 1 1 mr 6B/Bo rms 0.1 0.1 % Noise of monitors rms 250 250 Jlm max uncorrected orbit, hor. 50%c.1. 8.8 24.1 mm max uncorrected orbit, vert. 50%c.1. 9.3 14.4 mm max uncorrected orbit, hor. 95%c.1. 11.2 51.2 mm max uncorrected orbit, vert. 95%c.1. 14.7 24.5 mm max corrected orbit, hor. 50%c.1. 2.3 0.9 mm max corrected orbit, vert. 50%c.1. 1.0 1.4 mm max corrected orbit, hor. 95%c.1. 3.5 1.5 mm max corrected orbit, vert. 95%c.l. 1.2 2.2 mm max corrector strength, hor. 50%c.1. 4.4 0.2 mr max corrector strength, vert. 50%c.1. 0.3 0.2 mr max corrector strength, hor. 95%c.1. 5.7 0.2 mr max corrector strength, vert. 95%c.1. 0.4 0.2 mr 5 Table 2.2.1: Magnet Parameters for Injection System Magnet DCl,2 HKl,2,3,4 VKl,2,3,4 Powered dc dc Programmable Length m 1.0 0.5 0.4 Vertical Aperture mm 10 50 50 Horizontal Aperture mm 70 140 140 Angular Deflection mr 116 42.3 15 Magnetic Field T 0.396 0.278 0.126 Ampere Turns A-turns 3124 11065 14070 Table 2.2.2: Slow Extraction Simulation Achromatic Chromatic extraction extraction units Tune range /).v 0.011 0.032 Power-Supply noise dVnoi3e 10-4 10-4 Chromaticity of v ~ 0 -10.7 " of /3 (d/3jdp)po 667 667 m " of 0' (dO'jdp)po 0 0 Synchrotron tune V3 0.0016 0.0016 Duty Factor 55±4 64±4 % Septum Hits 0 1 (of 2000) Fraction not extracted 1.8±0.3 3.7±0.4 % Extracted Emittance w/o scattering in septum (100%) f 0.1911" 0.1711" mm-mrad with " (87%) f 0.0611" mm-mrad with " (98%) 4.211" mm-mrad f Momentum Bite dp/po 0.21 0.088 %FWHM Bunch Length d<l> 2.4 1 ns 6 Table 3.1.1: Magnet Quantities Dipoles Quadrupoles Sextupoles OeD's Total I-A Transfer 11 40 12 63 A-Ring 24 55 24 48 151 A -B Transfer 6+2S 8 16 B-Ring (ac) 25 52 24 24 125 B-C Transfer 5+2S 37 12 56 C-Ring 96 140 48 136 420 C-D Transfer 2+2S 8 12 D-Ring (ac) 96 140 48 128 412 D-E Transfer 2+6S 6 14 E-Ring 96 132 48 136 412 Switchyard and Primary BL's 31+4S 72 40 147 Secondary Beam Lines 35 87 6 128 TOTALS 445 777 198 536 1956 Quantities do not include spares Septa are indicated by +NS, where N is number needed 4 Skew quadrupoles have been included in the total for each ring All sextupoles and OeD's listed may not be needed for initial operations. 7 Table 3.1.2: Magnet Parameters for the Rings DIPOLES Effective Vertical Max. Pole Ring Length Aperture Field Width Quantity (m) (m) (T) (m) A de 1.00 0 .092 0.882 0.24 24 8 50 Hz 2.99 0.108 1.118 0.28 25 C de 1.00 0 .070 0.834 0.18 96 D 10 Hz 4.89 0 .100 1.381 0.28 96 E de 3.90 0 .052 1.731 0.25 96 One extra B-ring magnet ill included for power lIupply purpotlell QUADRUPOLES Effective Bore Pole Tip Max. Ring Type Length Radius Field Gradient (m) (m) (T) (TIm) A FandD 0 .3 0.067 0.284 4.25 8 F 0.46 0.071 0 .604 8.51 D 0.36 0 .071 0.890 12.53 C Fand D 0.2-0.4 0.065 0.64 9.78 SS 0.2 0.050 0.52 10.39 D FandD 0 .94-1.69 0.074 0.98 13.24 SSI 1.09-1.24 0 .052 0.98 18.84 SS2 0.68-0.94 0.061 0.98 16.06 E F 1.4-1.8 0.043 0 .521 12.12 D 0 .82 0.062 0 .99 15.96 SSI 1.1-1.6 0 .035 0.842 24.06 SS2 0.7-1.3 0.049 0.942 19.22 SEXTUPOLES Effective Bore Pole Tip Maximum Ring Length em) Radiu8 Field Gradient (m) (T) (TIm') A F 0.2 0.073 0.021 3.937 D 0.2 0 .073 O.ot5 2.760 8 F 0 .2 0.073 0 .079 14.89 D 0 .2 0.073 0.056 10.44 C F 0 .2 0 .086 0.103 13.93 D 0 .2 0 .058 0.050 14.94 D F 0 .2 0.086 0.834 112.8 D 0.2 0 .058 0.407 121.0 E F 0 .2 0 .070 0 .553 112.9 D 0.2 0.058 0.407 121.0 ORBIT CORRECTION DIPOLES Effective Maximum Vertical Bend Ring Length Field Aperture Angle (m} (T} (m} (mr) A 0 .18 0.01 0.12 0.53 B 0.18 0.028 0.072 0.40 C 0.18 0.016 0.102 0.22 D 0.18 0.127 0.102 0.22 E 0.18 0.127 0.102 0 .22 8 Quantity 51 24 24 96 40 96 12 28 48 48 12 20 Quantity 12 12 12 12 24 12 24 12 24 12 Quantity 24 24 136 136 136 Table 3.1.3: Prototype Booster Magnet Parameters Physics Energy Range Field Rise Frequency Field Fall Frequency Maximum Pole Tip Field Minimum Pole Tip Field Effective Length Pole Gap Field Uniformity Good Field Width Number Required Power Supply Maximum Current Maximum Voltage/Magnet (peak) Maximum Inductance No. Magnets/PS Cell Maximum Voltage/PS Cell (peak) 450-3000 MeV 33.3 liz 100 liz Dipole 1.05 T 0.271 T 3.18 m 10.68 em B/Bo ~ 1 X 10-4 ± 5.0 em 9 25 5000 A 3 kV 5.75 mil 5 15 kV Quadrupole 0.70 T 0.175 T 0.8 m 13.2 em CN/C2 < 2 X 10-3 6.6 em 48 1600 A 0.4 kV 1 mil 24 10 kV Table 3.2.1: Resonant Circuit Parameters B D Ring Energy .45-3 Ge V, 50 Hz 3-30 GeV, 10 Hz Dipoles 1. No.of resonant cells 5 10 2. No.of magnets/cell 5 10 3. Magnet inductance/cell 25mB 90mB 4. Choke inductance /cell 25mH 150 mH 5. Capacitance fixed/cell 680 uF 704 uF 6. Capacitance switched/cell 5600 uF 7. DC bias current 2850 A 3110 A 8. Peak magnet current 4500 A 5553 A 9. Peak choke current 4500 A 4576 A 10. Peak switch current 2170 A 11. Peak voltage to ground 7.8 kV 14 kV 12. Magnet peak stored energy 253 kJ/cell 1.39 MJ/cell 13. Choke peak stored energy 253 kJ/cell 1.57 MJ/cell 14. Cap. fixed stored energy 83 kJ/cell 276 kJ/cell 15. Cap. switched stored energy 244 kJ/cell Quadrupoles Focusing 1. No.ofresonant cells 1 2 2. No.of magnets/cell 25 24 3. Magnet inductance/cell 100 mIl 278 mil 4. Choke inductance/cell 100 mH 278 mH 5. Capacitance fixed/cell 202.8 uF 413 uF 6. Capacitance switched/cell 3295 uF 7. DC bias current 1000 A 1000 A 8. Peak magnet current 1600 A 1718 A 9. Peak choke current 1600 A 1718 A 10. Peak switch current 1276 A 11. Peak voltage to ground 9.41kV 12.6 kV 12. Magnet peak stored energy 128 kJ /cell 410 kJ/cell 13. Choke peak stored energy 128 kJ/cell 410 kJ/cell 14. Cap. fixed stored energy 8.96 kJ /cell 130 kJ/cell 15. Cap. switched stored energy 115 kJ/cell Quadrupoles Defocusing 1. No.of resonant cells 1 2 2. No.of magnets/cell 25 2 3. Magnet inductance/cell 100 mH 172.5 mil 4. Choke inductance /cell 100 mH 172.5 mH 5. Capacitance fixed/cell 202.8 uF 738.5 uF 6. Capacitance switched/cell 5,908 uF 7. DC bias current 1000 A 1000 A 8. Peak magnet current 1600 A 1768 A 9. Peak choke current 1600 A 1768 A 10. Peak switch current 1365 A 11. Peak voltage to ground 9.41kV 8.3 kV 12. Magnet peak stored energy 128 kJ/cell 270 kJ/cell 13. Choke peak stored energy 128 kJ/cell 270 kJ/t::ell 14. Cap. fixed stored energy 8.96 kJ/cell 102 kJ/cell 15. Cap. switched stored energy 91 kJ/cell 10 Table 3.2.2: Power Supplies for Accelerator Magnets Magnet Type Rating Quantity Regulation Volts Amps % OCD 60 12 48 0.1 60 200 184 0.1 100 200 264 0.1 Transfer line 20 500 30 0.1 40 500 4 0.1 150 500 2 0.1 20 1000 36 0.01 A & C quads 30 1000 4 0.01 40 1000 13 0.01 50 1000 6 0.01 15 1000 4 0.01 150 1000 1 0.01 E quads 350 1000 5 0.01 A,C,E dipoles 450 1000 22 0.001 B & D quads 450 1000 18 0.001 B & D dipoles 600 3500 9 0.1 PFN 2500 80 4 0.1 2000 500 1 0.1 2500 3200 1 0.1 Septum* 6.4 4100 6 0.01 1000 100 4 0.01 25 2600 2 0.01 41 2600 2 0.01 10 4400 1 0.01 10 5000 2 0.01 33 9550 2 0.1 6.5 8315 2 0.01 13.2 5000 2 0.01 682 * Does not include voltage drop in bus. Some septum magnets will be pulsed 11 Table 3.3.1: Injection and Extraction Kicker Magnet Parameters Kicker /I ~xtr 1 A B B C C D D D E E Location Extr 2 Inj Extr Ioj Extr Ioj Extr Fast Inj Abort Momentum (GeV Ic) 1.026 1.026 1.026 3.825 3.825 3.825 3.825 30.92 30.92 30.92 30.92 Kick angle (mrad) 1.76 8.44 10.0 7.7 4.0 4.0 4.0 2.5 2.5 2.0 2.0 Horizontal [H] or Vertical [V] V V V V V V V H H H H x Aperture (mm) 79 115 146 135 76 76 86 37 37 56 56 y Aperture (mm) 73 73 81 70 68 68 76 24 24 32 32 Frequency (Hz) 50 50 50 50 50 10 10 10 10 10 10 Flat Top (ns) 867 867 867 659 659 3625 3625 3522 3522 3522 3522 Rise Time (ns) 108 108 10ms 82 82 82 25ms 80 80 lOllS 160 Fall Time (ns) 1 ms 1 ms 108 10ms 82 lms 82 25ms 25ms 160 lOllS Fill Time maximum (ns) 79 79 79 53 53 53 53 51 51 132 132 Fill Time actual (ns) 49 78.2 64.6 47.3 53 53 53 51 51 132 132 'LVPFN (kV) 54.0 54.0 86.0 291.0 131.1 131.1 146.4 374.5 374.5 175.2 175.2 Number Modules 1 1 . 2 5 3 3 3 7 7 4 4 VPFN (kV) 54.0 54.0 43 58.2 43.7 43.7 48.8 53.5 53.5 43.8 43.8 Pulse Current (A) 1080 1080 1720 4656 874 874 976 1070 1070 876 876 Flux Density in Aperture (mT) 17.2 11.8 14.8 43.3 14.5 14.5 14.3 56.0 56.0 35.5 35.5 Characteristic sic Impedance (0) 25 25 12.5 12.5 25 25 25 25 25 25 25 Actual Magnetic Length (m) 0.35 2.45 2.314 2.27 3.535 3.535 3.579 4.607 4.607 6.0 6.0 Insertion Length (m) 0.9 3.0 2.904 2.904 4.165 4.165 4.209 5.397 5.397 6.67 6.67 Available Length (m) 0.9 3.0 2.904 2.904 7.77 7.77 7.147 7.0 7.0 11.9 11.9 12 Table 3.3.2: Diagnostic Kicker Magnet Parameters II Kicker Location 1/ Momentum (GeV Ic) 1.026 3.825 3.825 30.92 30.92 Kick angle (mrad) 0.2 0.3 0.2 0.2 0.1 x Aperture (mm) 70 80 77 86 49 Y Aperture (mm) 70 80 77 86 49 Rise and Fall Time (ns) 200 200 200 200 200 PFN Volts (.kV) 13 18 18 25 25 Impedance (ohms) 25 25 25 12.5 12.5 Magnetic Length (m) 0.05 0.35 0.23 0.7 0.2 Insertion Length (m) 0.45 0.75 0.63 1.1 0.6 Number 1 Vertical 1 Vertical 1 Vertical 1 Vertical 1 Vertical Modules 1 Horizontal 1 Horizontal 1 Horizontal 1 Horizontal 1 Horizontal 13 Table 4.2.1 Summary of RF Systems and Feedback Systems Ring Accumulator Booster Collector Driver Extender Carrier Frequency [MHz] 46.11 46.11-60.75 60.75 60.75- 62.53 62.53 Harmonic number 45 45 225 225 230 Revol'n frequency [MHz] 1.025 1.025-1.350 0.270 0.270-0.2779 0.2719 1hx. Detuning [MHz] .0461 0.1958 0.1519 0.4107 0.1351 Cavi ty Q (bare) 5000 3000-5000 5000 3000 5000 R/Q per cavity 100 35 (Q vs fo linear.) 100 100 100 . lIr! Volts/cavity [kV] 172 30.4-62.5 100-154 46.4-142 142 Max. h cos (h/ [0 9.88 32.2 25.0 39.4 21.6 Mill . FFD Gain 14 46 35 56 31 Suggested Gain 19.8 64.4 200 110 196 Cavity ~ DW (bare) [kHz] 4.61 6.08-7.69 6.07 10.1-10.4 6.253 FFB ! BW [MHz] 0.092 0.495 (track fo) 1.214 1.146 (track fo) 1.226 Max. allowed delay [ns] 65 70 65 70 65 Single turn delay FF no not likely likely yes v.likely Coupled-bunch mode FB no yes likely yes likely Injection compensation no yes yes yes yes fs y1tch (bare) [kHz] 41.2 47.3- 7.0 7.8 8.53- 0.44 0.44 Phase loop yes yes yes yes yes Amplitude loop yes yes yes yes yes Radial loop no yes no yes no Synchronization loop yes yes,2 no yes yes Quadrupole loop no yes TBD yes TOD N UIII bcr of Cavities 3 12 12 18 6 Lcng t.h (em) 225 123 171 169 166 Number of Amplifiers 3 12 24 36 6 Total Volts/ring [kV] 514 365-750 1300-1846 836-2550 847 Po/ring [kW] 88.8 31.7-185 284.6 64.7- 602 119.6 Pb/ring [kW] - 234-926 - 1984- 6350 -l'r / ri ng [k W] - 1725 2600 7770 1300 Tuning range [MHz] .0461 14.84 0.152 2.191 0.1351 Fast Tuning yes yes yes yes yes Max . t uning angle [deg] 84.2 88.2 87.7 88.6 87.3 14 Table 5.1.1: Transverse Impedances (in Mn/m) and Thresholds for the Fast Instability Ring ZBB Zsc ZTh e=O e = -1.3 A 0.35 66. 240. 240. B (at 3 GcV) 0.35 17. 28. 33. e 1.7 41. 61. 65. D (at 30 GeV) 1.7 5.1 8.1 24. E (bunched,/'t = 30)) 5.4 5.1 14. 30. E (dcbunched,/'t = 10) 5.4 5.1 71. 55. Table 5.1.2: Impedances and Growth Rates for the Resistive Wall Effect Ring ZRW Freq. growth rate (8 1) (Mn/m) (MHz) e=O e = -1.3 A 0.22 0.30 1500. 1400. B (at 0.45 GeV) 0.13 0.80 710. 630. e 1.1 0.21 1400. 1000. D (at 3.0 GeV) 1.1 0.21 1400. 450. E (bunched) . 5.6 0.21 1000. 38. E (debunched) 5.6 0.21 1000. 1000. Table 5.1.3: Transverse Impedances and Growth Rates Due to the Kicker Mag-nets Ring ZKi Freq. growth rate (8-1) (Mn/m) (MHz) e=o e = -1.3 A 0.020 4.8 140. 120. B (at 0.45 GeV) 0.016 16. 72. 58. e 0.093 7.1 120. 85. D (at 3.0 GeV) 0.740 7.4 970. 270. E (bunched) 0.460 2.9 77. 3. E (debunched) 0.460 2.9 77. 77. 15 Table 5.2.1 Summary of Horizontal Dampers Ring A B C D E Inject error (mm) 0.2 0.2 0.2 0.2 0.8 LlQH[coh] (%) 1.21 0.77 0.32 0.15 < 0.01 (13,,) (m) 15 15 30 35 30 ne (S-I) 750 520 140 55 5.4 Peak Ve 7 5 7 3 6 Peak Vz 35 20 170 255 3300 Peak VTot 40 25 180 260 3305 CW VTot < 10 <5 < 40 < 50 < 200 Quantity 1 1 1 1 3 Length (m) 1.6 1.5 1.5 1.5 1.4 Impedance n 25 25 50 50 50 Pea.k Power 2 X (20 W) 2x(lOW) 2 X (165 W) 2 X (340 W) 6 x (5 kW) CW Power 2 X (1 W) 2 X (1 W) 2 X (10 W) 2 x (15 W) 6 x (20 W) Table 5.2.2 Summary of Vertical Dampers Ring A B C D E Inject error (mm) 0.2 2.46 1.9 2.3 0.2 LlQv[coh] (%) 2.03 1.54 1.57 0.58 0.02 (13v) (m) 15 15 25 30 30 ae(s-I) 2000 12300 1780 1060 4 Peak Ve 19 930 690 390 2 Peak Vz 35 180 1600 2300 830 Peak VTot 55 1110 2290 2690 830 CW VTot < 12 < 15 < 50 < 60 < 200 Quantity 1 2 3 3 1 Length (m) 1.6 1.5 1.5 1.5 1.1 Impedance n 25 50 50 50 50 Peak Power 2 x (30 W) 4 x (1.5 kW) 6 x (3 kW) 6x(4kW) 2x(4kW) CW Power 2 X (3 W) 4 X (1 W) 6 X (2 W) 6 X (2 W) 2 X (200 W) 16 Table 5.3.1A Summary of Beam Instrumentation Device type Accumulator Booster Collector Driver Extender COD Beam Position · 48 48 134 134 132 Length (em) 20 20 20 20 20 Sense at RF 46.1 46-61 60.7 60- 62 62.5 LF cut-off (Mllz) /0 = 1.0 /0 = 1.0 5/0 = 1.35 5/0 = 1.35 5/0 = 1.36 Diagnostic Kicker 2 2 2 2 2 Length (em) 50 100 85 180 80 rise/ fa.ll (ns) 217 165 741 720 720 Duration (JlS) ~ 0.976 0.74 - 0.98 ~ 3.7 ~ 3.7 ~ 3.68 LF Wide band Posn. 5 5 5 5 5 Length (em) 20 20 20 20 20 LF cut-off (kHz) q/o = 200 qfo ~ 200 q/o = 54 q/o ~ 54 qfo = 51 HF cut-off (Mllz) 2: 1.02 2: 1.35 2: 1.39 2: 1.39 2: 1.39 HF Wideband Posn. 2 2 2 2 2 Length (em) 20 20 20 20 20 LF cut-off (MHz) ~ 1.02 ~ 1.02 ~ 0.270 ~ 0.270 ~ 0.272 JIF cut-off (GHz) 2: 2.5 2: 2.5 2: 2.7 2: 3.0 2: 3.9 Quadrupole Mon. 2 2 2 2 2 Length (em) 20 20 20 20 20 Profile Monitor 3 3 3 3 3 Length (em) < 40 < 40 < 40 < 40 <10 Halo X& Y Scraper 1 1 1 1 1 Length (em) 10 10 10 10 10 LF Current Monitor 1 1 1 1 1 Length (em) 15 15 15 15 15 LF cut-off (liz) 100 100 20 20 10 lIF cut-off (kHz) 2: 10 2: 10 2:10 2:10 2: 10 HF Current Monitor 1 1 1 1 1 Length (em) 30 30 30 35 3.5 LF cut-off (MHz) /0 = 1.0 fo = 1.0 /0 = 0.270 fo = 0.270 fo = 0.272 JIF cut-off (GHz) 3-4 3-4 3-4 4-5 4-5 Phase probe 3 12 8 12 1 Length (em) 10 10 10 10 10 Polarimeter - 1 - 1 -Length (m) - 1.2 - 1.5 --17 Table 5.3.1B Summary of Beam Instrumentation Device type Slow loss Monitor Length Time-slice (ms) Fast loss Monitor Length Time-slice (ns) Longitudinal Damper Length (m) Number of gaps kV per gap II Transverse Damper V Transverse Damper Mechanical Length (m) Definitions Accumulator Booster Collector Driver Extender 48 48 128 128 128 1 1 1 1 1 24 24 64 64 64 ",1 1 1 1.6 ",1 2 1.5 4 ",1 1 2 1.5 ",1 1 3 1.5 ",1 5 1.5 10 ",1 1 3 1.5 '" 1 3 1 1.4 The number of COD monitors is the tota,l for both planes II and V. Each monitor mea.sllres displacement in only one plane, either II or V. The number of LF Position monitors is the total for both planes II a.nd V. Each monitor measllres displacement in only one plane, either II or V. The LF cut-off is the fractiona.l part of the betatron tune multiplied by the revolution frequency. The Profile Monitors are either of the fast-wire or residual gas ionization type. Space has been allowed for photo-multiplier tubes downstream of the flying wire, if this is used. The single Ha.lo Scraper works in both planes II and V. There are 4 motorised jaws, and the step-size is ~ 0.1 mm .. Each quadrupole (or non-intercepting width) monitor measures in only one plane, either J[ or V. The LF Current Monitor has a resolution of < 1%, and range> 4 decades. The Loss (or spill) monitors are placed away from the beam pipe and so are accorded zero-lellJ!;lh in the table.The Fast Loss monitor are place at inject/eject etc .. The purpose of Phase Probe is two-fold: for RF cavity/beam synchronisation, and for ill implementing the feed-forward loops for beam-loading compensation. The probes are placed close to the RF cavities. Each transverse da.mper consists of a double stripline; each line is placed diametrically opposite its partner in the va.cuum tank. It is not essential to place horizontal and vertical dampers in sepa.ra.te vacuum tanks, so that a total of 3 damper units utilize only 3 m of floor-space. 18 Table 6.2.1: Beam Pipe and Vacuum Equipment List ITEM IA A AB B BC C CD D DE E Extr. Ceramic Beam Pipes (m) 0 0 0 128 0 0 0 651 0 0 0 SS Beam Pipes (m) 188 216 32 88 202 1078 42 427 51 1102 908 No. Ion Pumps (120i/s) 53 57 4 3 50 244 10 244 10 244 181 No. Ion Pumps (240i/s) 0 12 0 93 0 30 0 72 0 24 18 No. Sublimation Pumps 1 4 0 15 0 10 0 24 0 8 6 (1000i/s) No. Beam Pipe Gate Valves (rf) 0 4 0 4 0 8 0 8 0 8 0 No. Beam Pipe Gate Valves 3 2 2 1 3 0 3 0 2 1 6 (non-rf) No. Pumping Ports 3 6 1 5 3 8 2 8 1 9 6 No. High Vacuum Gauges 3 15 1 13 3 24 2 24 2 24 7 (Penning) No. Beam Position Monitors 53 51 4 48 50 136 10 136 10 128 181 19 Table 7.4.1: Collimator Shield Dimensions Storage Ring Iron Dimensions Iron Total Dimensions Concrete /Synchrotron Power diam. xlength Weight diam.xlength Volume kW m tonne m m3 Accumulator 5 1.5x2.0 26 2.0x2.5 4.3 Booster 10 2.0x2.5 58 2.5x3.0 6.9 Collector 10 2.0x2.5 58 2.5x3.0 6.9 .5 0.7x1.85 5 1.7x2.35 3.2 Driver 15 2.2x2.7 75 2.7x3.2 8.1 2 1.3x2.25 22 l.Sx2.75 4.0 Extender 30 2.5x3.0 lOS 3.0x3.5 10.0 2 1.3x2.25 22 l.Sx2.75 4.0 Table 7.5.1: Operational/Commissioning Beam Dumps Transfer Beam Line Power Iron Core or Spill Point Dissipation Weight Purpose kW tonne I-A Transfer Line 10 60 Chopper micropulse disposal Accumulator Injector 5 30 HO , H- beam spill dumps at injection A-B Transfer Line 10 60 Accumulator testing and disposal B-C Transfer Line 10 60 Booster testing and disposal C-D Transfer Line 10 60 Collector testing and disposal D-Extraction (Fast) 30 100 Driver testing and disposal E (Clearing) 30 100 Extender clearing, testing and disposal 20 Table 8.1.1: Distribution of the Current from TRIUMF Cyclotron to Achieve 100 pA at 30 Ge V Prestripper interception in cyclotron (available for research down beam line 1A Vacant buckets for kicker magnets 5/45 Total loss at injection Total loss at extraction Reset of painting parameters (f'V 2 ms/20 ns) Incomplete conversion H- to H+ Total loss at collimators Beam obtained at 30 GeV Total beam required from cyclotron Routine 11+ beam presently extracted (14 /-LA) 5/-LA 5/-LA 15/-LA 1.5/-LA 5/-LA 100/-LA 164.5 /-LA l60/-LA Table 8.2.1: Specifications for Extraction Devices Device Angle Leading Effective Edge Strength Gradient Length ( clv3) (m) Foil 239.1 RFD 136.75 0.2 MV/m 0.5 DCD1 114.1 3.9 MV/m 1.0 DCD2 123.6 3.9 MV/m 1.0 MC1 264.5 0.10 T 0.85 MC2 274.5 0.15 T 0.85 MC3 285.6 f'V 0.56 T 0.90 QMC1 293.0 0.0 4.0 T/m 0.30 MC4 296.0 f'V 0.57 T 0.90 MC5 320.0 0.50 T 0.7 T/m 1.20 21 b.R (mm) 30 50 140 300 400 RFB (kV) 0 0 0 150 300 Table 8.3.1: ExtraCted Radial Emittances for Various Conditions RFD E ~E P,Er (V/mm·m) (MeV) (MeV) (7r mm mrad) ±100 ±20° ±100 ±20° 55 452 1.1 1.5 1.6 1.8 82 452 1.1 1.5 2.0 2.5 110 452 1.1 1.5 2.7 3.3 110 463 2.5 3.5 4.5 6.5 110 473 4.0 5.0 6.0 9.0 Table 8.3.2: Extracted Beam Parameters Extraction Efficiency Horizontal Emittance Vertical Emittance Extraction Energy Energy Spread Phase Width Phase Jitter - ,...., 2 Hz -,...., 10 min Frequency 22 88% 3.5 7r mm mrad 2.0 7r mm mrad 452 MeV 1.5 MeV 20 deg @ 23 MHz (2.4 ns) ± 5 deg (± 0.6 ns) a further ± 5 deg (± 0.6 ns) 23.055 MHz ± 0.002 MHz Efficiency 84% 87% 90% 91.5% 93% Table 9.3.1: Controls System Component Breakdown Item Equipment Management System LAN cable (Data Interconnect) (2 !un) Transceiver and cable Sub-system Specialist W /S VME-based development system Logging/database engine Database software Storage server Controls diagnostic equipment Report printer Development Support System Software Development W /S VME-based development system 1Tansceiver + cable Report printer Equipment Protection System Misc. cables (2 km) Master Timing System Backbone timing cables Intra-ring timing controllers Diagnostic timing channels KF-equipment timing channels Timing channel cables Operator Interrace Operator W /S 10 Pc:rtable W /S Beam Control System KF equipment FEPs Additional FEPs FEP resident software LAN cable (Control Interconnect) (2km) 1 LAN cable (Fact. LAN) (2 km) 1 1Tansceiver + cable (Control Interconnect) 90 Modelling computer 1 VME crate + 2 LAN cards, etc 90 VME crate + I/O modules 108 VME crate + moni'toring equip 20 LAN bridges 2 Beam Instrumentation Support System KF-Diagn08tic FEPs 1Tansceiver + cable (Control Interconnect) VME crate + 2 LAN cards, etc Analogue Display System ADS FEP 1 Diagnostic Analogue Channels 855 KF Equip MUX channels 460 Backbone Analog Cables (1 !un) 6 Analog input cables (1/12 !un) 1315 50 MHz Digital Scopes 6 GHz cables (1/6 km) 8 GHz transient digitizer 23 Item Count 1 290 5 5 1 1 1 8 5 10 10 20 5 1 14 380 380 286 380 20 680 33 1 380 95 95 8 INDEX OF DESIGN NOTES Chapter 12 12 INDEX OF KAON FACTORY DESIGN NOTES 12-1 1 12 INDEX OF KAON FACTORY DESIGN NOTES Number Author Subiect Date TRI-DN-88-Kl J .R. Richardson A Possible Future Increase in the Oct 88 Research Capability of the TRIUMF KAON Factory (The TRI-IOO) TRI-DN-88-K2 T. Suzuki Emittance Growth and Beam Loss by Oct 88 Betatron Resonances in High Intensity Proton Synchrotrons. TRI-DN-88-K3 U. Wienands A "Clean" Method to Create the Kicker Oct 88 Hole for KAON Factory B,C,D,E, Rings. TRI-DN-88-K4 D. Wilkinson The KAON Factory and Oct 88 the Atomic Nucleus TRI-DN-88-K5 T. Suzuki Analytical Expressions for Spurious Oct 88 Dispersion. TRI-DN-88-K6 C.Oram, Vacuum Requirements for the KAON Nov 88 R. Baartman Factory TRI-DN-88-K7 G.H. Rees Magnet Lattices of a Compatible Form Nov 88 for the A and BRings TRI-DN-88-K8 G.H. Rees Note on the Coherent Motion of a Nov 88 Hollow Longitudinal Distribution TRI -D N -88-Kg G.H. Rees H- Injection Topics Nov 88 TRI-DN-88-KIO G.H. Rees Longitudinal Impedance Estimate for Nov 88 the Distributed Capacitors of the SAIC RF Shield TRI-DN-88-Kll G.H. Rees Kicker Requirements for Booster Dec 88 Extraction and for the A-Ring Injection Line TRI-DN-88-K12 Alan J. Otter LAMPF Stranded Magnet Cable Dec 88 TRI-DN-88-K13 R. Baartman, Beam Pipe Impedances for the KAON Dec 88 C.Oram Factory 12-1 Number Author Subject Date TRI-DN-89-K14 G. Clark Support Magnets Jan 89 TRI-DN-89-K15 S. Koscielniak Non-Linearities in the Phase Advance Jan 89 Equation TRI-DN-89-K16 S. Koscielniak Stability Analysis of RF Cavity Jan 89 Tuning Loop TRI-DN-89-K17 C.Oram Revised Estimate of the Longitudinal Jan 89 Impedance for the Distributed Capacitors of the SAIC RF shield TRI-DN-89-K18 A.J. Otter A Prototype for the Booster Dipole Feb 89 TRI-DN-89-K19 J. Doornbos, Layout of Secondary Beams at KAON Jan 89 J. Beveridge TRI-DN-89-K20 E.J.N. Wilson Sextupole Patterns and the 2v,x-2vlI Feb 89 Coupling TRI-DN-89-K21 E.J.N. Wilson An Elementary Approach to Synchro- Feb 89 betatron Resonances TRI -D N -89-K22 P. Schwandt Estimation of Specific Core Losses in Feb 89 DC-biased Laminated Magnets TRI-DN-89-K23 M.R. Harold Preliminary Designs for the Driver Feb 89 Quadrupoles QF and QD TRI-DN-89-K24 M.R. Harold A Preliminary Design for the Driver Feb 89 Dipole TRI-DN-89-K25 S. Koscielniak A General Theory of Beam Loading Mar 89 TRI-DN-89-K26 George S. Clark TRIUMF Shielding Cost Experience and Feb 89 KAON Factory Experimental Hall Shielding TRI-DN-89-K27 M. Butler, Scattering Cross-sections for 9Be, Mar 89 S. Koscielniak 12C 160 and 27 Al , 12-2 Number Author Subiect Date TRI-DN-89-K28 T. Enegren RF Booster Cavity Biasing Ferrite in Mar 89 Radial Direction TRI-DN-89-K29 H. Sasaki Design of the Energy-Storage Choke Apr 89 for the Booster Ring Magnet System TRI-DN-89-K30 B. Frammery Timing for the KAON Factory Apr 989 TRI-DN-89-K31 P. Schwandt Comparison of Realistic Core Losses Mar 89 in the Booster Ring Dipole Magnets for Grain-Oriented and Ordinary Lamination Steels TRI-DN-89-K32 P. Schwandt An Improved Magnetic Design for the Apr 89 Booster Ring Dipoles TRI-DN-89-K33 E.W. Blackmore Experimental Area Design Working Apr 89 Group Summary TRI -D N -89-K34 P.A. Reeve Preliminary Design of Prototype Apr 89 Booster Quadrupole TRI-DN-89-K35 C.Oram, Design of the Vacuum System Jul89 E. Jones TRI-DN-89-K36 E.W. Blackmore Radiation Levels in the KAON Factory May 89 Accelerator Tunnels TRI-DN-89-K37 P. Schwandt Magnetic Design of the Driver Ring May 89 Dipole - A Second Look TRI-DN-89-K38 W. Bothe Driver Dipole Circuits TRI-DN-89-K39 D. Pavlic A Review of Energy Storage Chokes May 89 for Rapid Cycling Synchrotrons TRI-DN-89-K40 R. Baartman Synchrobetatron Resonance Driven By May 89 Dispersion in RF Cavities: A Revised Theory 12-3 Number Author Subiect Date TRI-DN-89-K41 T. Eaton A Target Study for the Fast Jun 89 Extracted 10 Hz Beam TRI-DN-89-K42 G. Guignard Possible Lattice Design for the Jun 89 Extender Ring TRI-DN-89-K43 R. Laxdal H- Extraction Energy Jun 89 TRI-DN-89-K44 L. Root Preliminary Optical Analysis of the Jul89 H- Extraction System TRI-DN-89-K45 D. Fiander, Electric Field Deflector for the Jul89 G. Wait, 1 MHz Chopper System M. Barnes TRI-DN-89-K46 R. Baartman Optical Effects of Tuner Solenoids Jul89 TRI-DN-89-K47 H. Sasaki Design of the Energy-Storage Choke Jul89 for the Booster Ring Magnet System II TRI-DN-89-K48 R.V. Servranckx, In Search of a Booster Lattice Jul89 U. Wienands, M.K. Craddock TRI-DN-89-K49 M. Barnes, Capacitor Plates Measurements Aug 89 R. Barnes TRI-DN-89-K50 H. Schonauer Addition of Transverse Space Charge Aug 89 to ACCSIM code TRI-DN-89-K51 G. Rees Space for a Longitudinal Beam Loss Aug 89 Collector in the DRing TRI-DN-89-K52 G. Rees A to B Ring Transfer and B Ring Fast Aug 89 Extraction TRI -D N -89-K53 V. Radel, Cooling of the Long Transmission Line Aug 89 G. Wait of the 1MHz Chopper 12-4 Number Author Subject Date TRI -D N -89-K54 V. ROdel, Voltage Build-up in the Long Pulse Aug 89 G. Wait Generator Cable of the IMHz Chopper TRI-DN-89-K55 D. A. Reich E Ring Slow Extraction Magnetic Aug 89 Septum: Concept Design TRI -D N -89-K56 C.Oram Beam Pipe Options for the AC Magnets Oct 89 at the KAON Factory TRI-DN-89-K57 A. Otter A Conceptual Design for a D-Ring Aug 89 Orbit Correction Dipole TRI-DN-89-K58 R. Servranckx, Report on Lattices for Extender Ring Aug 89 U. Wienands of the KAON Factory TRI-DN-89-K59 G. Wellman Transverse Space Charge Simulation in DIMAD TRI-DN-89-K60 C.Haddock, Prototype Booster Magnet Aug 89 P.A.Reeve, Measurements D. Evans, D. Livesey TRI-DN-89-K61 D.R. Gill Superconducting Technology at KAON Aug 89 TRI -D N -89-K62 R. Baartman, Transport Equations for ACCSIM Sep 89 F.W. Jones TRI-DN-89-K63 E. De Vita A Proposed Design for Extraction Sep 89 Channel No.5 TRI-DN-89-K64 E. De Vita Increasing the Field in the CSC Sep 89 Dipole TRI-DN-89-K65 E. De Vita Attaining a Uniform Field Gradient in Sep 89 Extraction Channel 4 TRI-DN-89-K66 J. Lenz Some Properties of a Plasma Sprayed Sep 89 Aluminum Oxide Film 12-5 Number Author Subject Date TRI-DN-89-K67 R. Baartman Stability During Slow Extraction in the Moscow Meson Factory Storage Ring TRI -D N -89-K68 S. Koscielniak RF Noise Tolerances for KAON Factory Sep 89 TRI-DN-89-K69 M. Barnes, Attenuation and Dispersion of Pulses G. Wait in Low Loss Coaxial Cable TRI-DN-89-K70 M. Barnes, A 1MHz Chopper for the KAON Factory Nov 89 G. Wait TRI-DN-89-K71 G. Clark KAON Factory Mnemonic Device Oct 89 Naming Convention TRI-DN-89-K72 B. Zotter Stability of Hollow Beams Jul89 TRI-DN-89-K73 P. Schwandt ANew Magnetic Design of the Oct 89 Driver Ring Dipoles TRI-DN-89-K74 S. Koscielniak RF Synchronization During Transfer of Batches from Booster to the Collector Ring TRI-DN-89-K75 M. Barnes, Analysis of the Transient Response of Oct 89 G. Wait Magnetic Kickers for the KAON Factory TRI-DN-89-K76 M. Barnes, Influence of Stray Capacitance on Oct 89 G. Wait IMP's Chopper Performance TRI-DN-89-K77 J.R. Richardson Another Look at the TRI-lOO Oct 89 TRI -D N -89-K78 L. Criegee A Study of the Slow Extraction at the Oct 89 KAON Factory TRI-D N -89-K79 H. Lustig Status Report on the Control of Power Oct 89 Supplies 12-6 Number Author Subject Date TRI-D N -89-K80 David Orrell Designs for a Quadrupole in the Oct 89 Driver Ring TRI-DN-89-K81 D. Boussard Scaling up of CERN PS Controlled Oct 89 Blow Up to the Collector Ring TRI -D N -89-K82 D. Boussard Variable Phase Shifter for the Fast Oct 89 RF Feedback in the Booster TRI-DN-89-K83 D. Boussard Study of a Feedback System Against Coupled Bunch Instabilities in the Booster TRI-DN-89-K84 G.F. Wellman Transverse Space Charge Simulation in Aug 89 DIMAD TRI-DN-89-K85 M. Barnes, Effect of Thyratron Displacement Nov 89 G. Wait Current upon Kick Rise-Time in Magnetic Kickers. TRI-DN-89-K86 M. Barnes, Kicker Magnet Fill-Time and Nov 89 G. Wait Parameters TRI-DN-89-K87 M. Barnes, Estimating of Negative Mutual G. Wait, Coupling Between Adjacent R. Barnes Cells of a Kicker Magnet. TRI-DN-89-K88 M. Harold Increasing the Field in the Booster Nov 89 Dipole. TRI-DN-89-K89 M. Harold Summary of Extraction Channel from Nov 89 the Dring. TRI-DN-89-K90 M. Harold Extraction Channel from the Booster Nov 89 TRI-DN-89-K91 D. Orrell Programs to Generate Models for Dec 89 PE2D and TOSCA TRI -D N -89-K92 T. Hodges Beam Heating of Vacuum Components Dec 89 12-7 Number TRI-D N -89-K93 TRI-DN-89-K94 TRI-DN-89-K95 TRI -D N -89-K96 TRI-DN-89-K97 TRI-DN-89-K98 TRI-DN-89-K99 TRI-DN-89-KIOO TRI-DN-89-KI01 TRI-DN-89-KI02 TRI-DN-89-KI03 TRI-DN-90-K104 TRI-DN-90-KI05 Author K. Fong K. Fong J.Pearson J. Pearson R. Laxdal Subject Conceptual Design for Magnetic Channel 3 Conceptual Design for Magnetic Channel 4 Date Guide to Using the CHANNEL Group Dec 89 of Programs. Field Design of Magnetic Channel Dec 89 No.1 for the H- Extraction Scheme. H- Extraction Parameters Dec 89 C.W. Planner, A Separated H- Injection System Dec 89 G .H. Rees, In A Modified Accumulator G. Mackenzie D. Orrell TOSCA Modelling of Gower Dec 89 Quadrupole C. Haddock Evaluation of Eddy Current C. Haddock Comparison of PE2D Eddy Current Dec 89 Loss Predictions and a Semi-Analytical Calculation G. Rees, Modulation Alternatives for the S. Koscielniak Longitudinal Emittance Dilution Cavities. D.G. Martin, The Survey and Alignment of the Dec 89 G. Clark KAON Factory Project T.A. Enegren Measurement of RF Cavity Shunt T.A. Enegren Coupled TL Higher Order Mode Damper 12-8 Number Author Subject Date TRI-DN-90-K116 S. Koscielniak Hints to Those Who Would Feb 90 Simulate the Effects of Random Time Perturbations TRI-DN-90-K117 R. Helmer Neutrino Facility Feb 90 TRI-DN-90-K118 N. Wilkinson The VMEbus and Related Real-Time Technology 12-10 Number Author Subiect Date TRI-DN-90-KI06 G. Spinney, Engineering of H- Beam Extraction Jan 90 G. Stanford, Equipment M. Zach TRI-DN-90-KI07 G.R. Gathright, Booster Quadrupole Laminated Core P.A. Reeve Checks TRI-DN-90-KI08 H. Sasaki Separate Energy-Storage Choke for Feb 90 Driver Ring of KAON TRI-DN-90-KI09 H. Sasaki Addendum to "Design of the Feb 90 Energy-Storage Choke for the Booster Ring Magnet System II" (TRI-DN-89-K47) TRI-DN-90-Kll0 T .A. Hodges, Laboratory Prototype of Rotating Jan 90 R. R. Langstaff, Immersed Target M.S. Lenckowski TRI-DN-90-Klll T.A. Hodges, Prototype "Bare" Rotating Target Jan 90 R.R. Langstaff TRI-DN-90-K112 C.Oram Comparison of Major Cost Components TRI-DN-90-K113 R. Laxdal, Tolerances Associated With the P. Cronje, Fields of Extraction Devices L. Root, E. Roosen TRI-DN-90-K114 R. Laxdal, Emittances of Extracted P. Cronje, H- Beams L. Root TRI-DN-90-K115 M. Barnes, Assembly Procedure for the G. Wait Prototype Vertical Extraction Kicker Magnet for the KA 0 N Factory 12-9 


Citation Scheme:


Citations by CSL (citeproc-js)