UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

RF control of the M9 separator at TRIUMF Burge, R. 1990

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

Item Metadata

Download

Media
831-UBC_1990_A7 B87.pdf [ 3.25MB ]
Metadata
JSON: 831-1.0302158.json
JSON-LD: 831-1.0302158-ld.json
RDF/XML (Pretty): 831-1.0302158-rdf.xml
RDF/JSON: 831-1.0302158-rdf.json
Turtle: 831-1.0302158-turtle.txt
N-Triples: 831-1.0302158-rdf-ntriples.txt
Original Record: 831-1.0302158-source.json
Full Text
831-1.0302158-fulltext.txt
Citation
831-1.0302158.ris

Full Text

RF CONTROL OF THE M9 SEPARATOR AT TRIUMF by RAYMOND STANLEY BURGE B.A.Sc, University of B r i t i s h Columbia, 1984 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE i n THE DEPARTMENT OF ELECTRICAL ENGINEERING We accept t h i s thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA Ap r i l 1990 (E)Raymond S t a n l e y Burge, 1990 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of £7ccrVic<J f w c j M e c r M ^ The University of British Columbia Vancouver, Canada Date Ktxy HlO DE-6 (2/88) Abstract High voltage RF systems are used to accelerate proton beams f o r nuclear physics experiments. The acceleration process shapes the proton beam into a t r a i n of narrow pulses with the same period as the RF. This bunched beam structure i s used to separate and i d e n t i f y secondary p a r t i c l e s that are produced when the proton beam i s directed at a "target". An RF c o n t r o l l e r for a system that separates secondary p a r t i c l e s was b u i l t . Control of high power RF c a v i t i e s that operate near resonance i s discussed. The emphasis i s on developing a control model f o r resonant systems and building a control system based on hardware and software modules that can be e a s i l y configured f o r d i f f e r e n t RF systems. i i Table of Contents page Abstract < i i Table of Contents i i i L i s t of Figures v Acknowledgement v i i 1. Introduction 1.1 RF Systems and P a r t i c l e Accelerators 1 1.2 An RF Separator f o r Muon Studies 4 1.3 System Requirements 6 2. The RF Cavity and Power Amplifier 8 3. Transfer Functions f o r RF Regulation 3. 1 An RLC Model 12 3.2 A Simple Transfer Function f o r the Separator 13 3.3 Matrix Transfer Functions for Amplitude and Phase Regulation 16 3.4 P a r t i c l e Beam Loading as System Disturbance 21 4. Closed Loop Control 4.1 Phase and Amplitude Loops 25 4.2 The Cavity Tuning Loop 30 i i i page 5. Hardware Implementation of the Controller 5. 1 Overview 35 5.2 System Description 35 5.3 Self-Excited Operation 38 5.4 Self-Excited and Driven Tuning Systems 40 5.5 The Frequency Comparator 43 5.6 Spark Detection 44 5.7 System Hardware Configuration 45 5.7 Regulator Electronics 47 6. Software 6.1 The Main Program Loop 51 6.2 Task Communication 52 6.3 Control Tasks 53 6.4 Front Panel Input 54 6.5 Auto Start 57 7. Conclusions 60 8. References 61 iv L i s t of Figures page 1. Drift-tube Linac structure 2 2. Time structure of the TRIUMF beam 4 3. Relative phase of TT, U, and e beam components 5 4. Schematic of the separator cavity structure 8 5. Push-Push model of the cavity and plate c i r c u i t 9 6. Push-Pull model of the cavity and plate c i r c u i t 10 7. Equivalent c i r c u i t of high Q elements 11 8. Lumped RLC resonator 12 9. Pulsed power tests 14 10. One dimensional transfer function for the RF system 15 11. Linearized model for high Q c i r c u i t elements 21 12. Beam loading seen by a single voltage probe 22 13. Block diagram of the regulator loop 26 14. RF control space 27 15. Reguator simulation showing cross term r e j e c t i o n 30 16. Cavity tuning mechanism 31 17. Cavity tuning loop 31 18. Tuner small signal model 33 19. Basic c o n t r o l l e r concept 35 20. Control modules 36 21. Front panel display 37 v page 22. Self-excited configuration 39 23. P r i n c i p a l system states 39 24. S e l f - e x c i t e d tuning loop 41 25. Driven tuning loop 42 26. Signal switch f o r tuner and phase regulator 43 27. Frequency comparator 44 28. Spark detector 44 29. Separator RF control modules 46 30. PID Regulator and RF System Model 48 31. S i m p l i f i e d regulator schematic 49 32. Discrete component d i f f e r e n t i a t o r 49 33. D i f f e r e n t i a t o r simulation 50 v i Acknowledgement I am much indebted to Dr. R. Ward for her guidance i n my studies and her encouragement i n the preparation of t h i s thesis. Thanks are also due to the members of the TRIUMF RF group and to MR. L Durieu of CERN and MR. P. Sigg of PSI. Special thanks are due to Mr. B. Chow and Mr. G. Dennision who helped b u i l d t h i s control system and without whose support and diligence t h i s project could not have succeeded. v i i Chapter 1 Introduct ion 1.1 RF Systems and P a r t i c l e Accelerators High power RF systems are used to c o n t r o l and acce lerate p a r t i c l e beams i n nuclear phys ics l a b o r a t o r i e s . In t h i s process i t i s d e s i r a b l e that the RF system be as easy to operate as any power supply i n an i n d u s t r i a l environment. Before d i s c u s s i n g the requirements for r e g u l a t i o n and c o n t r o l , i t i s useful to introduce the bas i c func t ion o f RF systems in modern p a r t i c l e a c c e l e r a t o r s . P a r t i c l e a c c e l e r a t o r s are machines that generate ion beams with s u f f i c i e n t k i n e t i c energy to produce nuclear reac t ions . U s u a l l y the acce l era ted p a r t i c l e s are e lec trons or protons and the k i n e t i c energy of a p a r t i c l e i s measured i n e l e c t r o n v o l t s (eV). An e l e c t r o n vo l t i s the energy gained by an e l e c t r o n (or any p a r t i c l e with the same magnitude charge) i n pass ing through a p o t e n t i a l drop o f one v o l t . Modern proton machines can produce p a r t i c l e s wi th energies of many b i l l i o n e l e c t r o n v o l t s (GeV). Charged p a r t i c l e s are acce l era ted i n vacuum by e l e c t r i c f i e l d s . The f i r s t high energy a c c e l e r a t o r s were DC machines us ing Cockcrof t -Walton and Van de Graaf f generators . These h igh vol tage sources can produce p a r t i c l e beams with energies of a few m i l l i o n e l e c t r o n v o l t s (MeV). The s t a t i c f i e l d s encountered i n DC machines l i m i t the k i n e t i c energy gained by the p a r t i c l e to the p o t e n t i a l 1 energy of the system. The maximum energy of an e l e c t r o s t a t i c accelerator i s determined by the voltage that can be maintained at the high voltage terminal. With careful design t h i s can be as high 7 as 10 vol t s . C y c l i c accelerators were developed to overcome the l i m i t s imposed by e l e c t r o s t a t i c f i e l d s . In these machines, p a r t i c l e s are made to follo w a path where they p e r i o d i c a l l y encounter an accelerating voltage which i s a small f r a c t i o n of the f i n a l energy gained by the p a r t i c l e . The p a r t i c l e trajectory can be straight as i n li n e a r accelerators (Linacs), or the trajectory can be curved as i n cyclotrons and synchrotrons. High k i n e t i c energies are b u i l t up as the p a r t i c l e encounters the accelerating voltage a large number of times. This process develops a gradual acceleration which i s not li m i t e d by the voltage drop i n the machine. o -> to target fig. 1 Drift-tube Linac structure. Large energy gains are possible i n c y c l i c accelerators because the e l e c t r i c f i e l d s i n the machine change with time. Such systems are nonconservative i n that i t i s possible to f i n d a closed path along which the k i n e t i c energy gained by the p a r t i c l e i s not zero. ion source RF System 0 o o o -» J L 2 Figure 1 shows the structure of a drift tube Linac. It demonstrates that, with correct phasing of the accelerating voltage, one can inject charged particles at zero volts and later extract them into a zero volt region with a higher kinetic energy. Acceleration occurs in the gap between the drift tubes. While the particle is inside the drift tube, it is in an equipotential region and does not experience any change in energy. Cyclic accelerators use equipotential regions, such as the drift tube, to allow the voltage time to achieve the correct amplitude and phase as the particle approaches the next acceleration gap. In modern machines, the particle velocity quickly approaches the speed of light and, in order to keep the size of the machine small, it is advantageous to use high frequency voltages to accelerate charged particles. This minimizes the drift tube length and allows resonant RF structures to be used as the accelerating elements. Most nuclear reactions of interest to researchers occur as relatively rare events. A great number of particles with a well defined kinetic energy are needed to study these events. Since there is an equivalence between mass and energy it is important to know the threshold energy needed to produce the exotic particles that make up the nucleus of the atom. To get meaningful statistics an experiment can run for a year at low event rates. The task of the accelerator designer is to produce a machine that delivers as much beam as possible at a very well defined energy. In cyclic accelerators a well defined energy occurs over a small range of RF phase and voltage. The result is that accelerators produce pulses or buckets of particles at a given energy. This is true for a l l 3 c y c l i c accelerators including the TRIUMF cyclotron ( f i g . 2). fig. 2 Typical time structure of the TRIUMF proton beam. C y c l i c accelerators produce beam pulses at the RF r e p e t i t i o n rate with a t y p i c a l pulse occupying approximately 30° of an RF cycle. Control of high intensity, high energy p a r t i c l e beams requires careful control of the RF phase and amplitude. 1.2 An RF Separator f o r Muon Studies P a r t i c l e s extracted from the accelerator are collimated into a beam and are directed to a "target" where the desired nuclear reactions occur. Secondary p a r t i c l e s produced i n the reaction can be selected and used i n further studies. One of the p a r t i c l e s produced i s the pion, an exchange p a r t i c l e that operates inside the nucleus. An atomic nucleus i s made up of protons which have a po s i t i v e charge and neutrons which have zero charge. The net charge of the nucleus i s positive and since l i k e charges repel, the nucleus should f l y apart. The pion i s an exchange p a r t i c l e that transmits a short range a t t r a c t i v e force which allows stable nuclei to e x i s t . This p a r t i c l e i s one of a group of p a r t i c l e s c a l l e d mesons which are the object of much study at TRIUMF. 4 The name pion i s a short form of ir-meson. Outside the nucleus t h i s p a r t i c l e has a high pr o b a b i l i t y of quickly decaying to a jx-meson (muon). After a longer time, the muon w i l l decay into an electron. Pions are produced when a pulse of high energy protons (>100Mev) h i t s a Beryllium target i n the beam l i n e . The groups of pions leaving the target have the same time structure as the high energy protons from the TRIUMF cyclotron. A cloud of muons w i l l begin to form around the pions as they decay i n f l i g h t from the production target. The p a r t i c l e s have d i s t i n c t and d i f f e r e n t energies such that the pions, muons and electrons form separate groups i n f l i g h t . Pions and electrons can be suppressed to less than 1% of the beam by a s t a t i c magnetic f i e l d and a perpendicular RF e l e c t r i c f i e l d . P a r t i c l e s are passed through the separator when the RF e l e c t r i c f i e l d cancels the vxB force of the horizontal magnetic f i e l d . By adjusting the phase of the RF voltage, physicists can l e t pions, electrons, or muons into the experiment. T i m e (ns) fig. 3 Relative phase of n, u, and e beam components. 5 1 . 3 System Requirements TRIUMF i s si m i l a r to a u t i l i t y company that delivers high energy protons to i t s customers and also provides technical services to the various experiments that use the proton beam. A device to separate decay products i n f l i g h t from the pion production target i s one of the pieces of equipment b u i l t for muon studies. The muon physics experiment i s a large and complex i n s t a l l a t i o n of which the RF separator i s only a small part. The separator c a v i t y i s driven at 23 MHz, the cyclotron operating frequency, and i t i s phase locked to the primary proton beam. It must be simple to operate; an on-off switch and a knob to select d i f f e r e n t p a r t i c l e s , n, u, or e. The RF requirements f o r the separator can be l i s t e d i n three sections. 1. RF Generator, Cavity, and Deflection Plates • a frequency of 23MHz. • a power output of 120Kw • a peak plate-to-plate cavity voltage of 360Kv 2. RF Regulator • phase locked to the proton beam extracted from the cyclotron • phase s t a b i l i t y better than 1° 3 • amplitude s t a b i l i t y better than 1 part i n 10 3. RF Controller • one button autostart • automatic cavity tuning 6 • automatic spark recovery and f a u l t handling • a s e l f - e x c i t e d , i d l e state when the cyclotron RF i s o f f • lo c a l display of the system state • manual control of a l l loop parameters and system states This report describes the analysis and design of the regulator, the c o n t r o l l e r , and the RF signal modules that were b u i l t and i n s t a l l e d on the RF separator. A control model i s developed f o r regulation and tuning RF systems operated near resonance. 7 Chapter 2 The RF Cavity and Power Amplifier The power amplifier, transmission l i n e , and cavity were designed and commissioned by Bob Worsham and Vojta Pacak who are p r i n c i p a l members of the TRIUMF RF group. Their work provides the basis f o r the lumped model which i s used to develop a transfer function describing the system response to control modulations. transmission line beam .1 1 1 1 X /A. privity high voltage gap A/4 cavity fig.4 Schematic of the separator cavity structure. The RF separator consists of two A/4 short c i r c u i t coaxial transmission l i n e c a v i t i e s that are t i g h t l y coupled through the t i p capacity at the high voltage electrodes. Two fundamental resonant modes resul t from the close coupling between the ca v i t i e s . These are a push-pull resonance, which i s the desired mode, and a push-push resonance which occurs at the frequency 8 defined by the A/4 c a v i t i e s . At the push-push resonance, the cav i t y Q and the shunt resistance measured at the high voltage gap are: Q = 4800 cavity parameters Rsh = 141k« A 2X long, 50Q transmission line connects the upper cavity to the high Q plate c i r c u i t of a class "C" grounded gr i d power amplifier. This introduces an 87ns delay between the plate c i r c u i t and the cavity. The Q and shunt resistance of the plate c i r c u i t are: Q = 3000 plate circuit parameters R P = 12000 The response of the separator RF system i s dominated by high Q elements i n the plate c i r c u i t of the power amplifier and by the cavity. A lumped model of t h i s system also includes the transmission l i n e and the power tube represented by an RF current source. Starting from the push-push model, one can arrive at a simple equivalent c i r c u i t for the dominant RF elements. fig. 5 Push-Push model of the cavity and plate circuit. 9 The push-push resistance measured at the gap i s the p a r a l l e l combination of the shunt resistance from each cavity. The desired push-pull mode operates at a frequency below the p a r a l l e l resonance defined by the push-push mode. At th i s frequency, each ca v i t y appears inductive and resonance i s formed by the gap capacitance i n series with one side. A capacitive voltage probe at one side of the high voltage gap sees a lumped model s i m i l a r to that i n the figure below. I P © fig. 6 Push-Pull model of the cavity and plate circuit. The amplitude and phase of the gap voltage i s regulated using the signal measured by a capacitive voltage probe on one side of the gap. To develop a transfer function for the system i t i s useful to transform the plate c i r c u i t to the high voltage gap. In the c i r c u i t below, Zi i s the transformed plate impedance, Zz i s the - s T cavity impedance seen from the deflection plate, and e i s the t o t a l delay i n the transmission l i n e s , from the plate c i r c u i t to the measurement point; a distance of 4A. 10 -sT <- V(s) Ks) O Z l Z2 fig. 7 Equivalent circuit of high Q elements. A control model f o r the RF cavity and the amplifier plate c i r c u i t i s developed from the admittance of these elements. The rela t i o n s h i p between the transformed plate current and the gap voltage i s given by: Ks) = V(s) [ J . + g l ) e s T = V(s) Y(s) When the plate c i r c u i t i s matched to the cavity, the e f f e c t i v e plate resistance seen by the cavity w i l l be equal to the cav i t y shunt resistance, Rsh. The t o t a l shunt resistance seen by the RF current generator w i l l be 1/2 Rsh. 11 Chapter 3 Trans fer Funct ions for RF Regulat ion 3.1 An RLC Model To develop a t r a n s f e r f u n c t i o n f o r c a v i t y r e g u l a t i o n , i t i s use fu l to cons ider a s imple lumped resonator dr iven by an RF current source . fig. 8 Lumped RLC resonator. The admittance, Y = l e t w = wo + Aw 1 Z where: wo R 1 Q = woL LC woRC R [ wo wo+Aw JJ 12 This approximation, with less than 5% error for Aw ^ wo/10, i s very useful for calculations within the control bandwidth. It represents the cavity as a f i r s t order pole about the resonant frequency, wo. The RLC c i r c u i t responds to control modulations l i k e a simple RC low pass f i l t e r with a time constant: 2Q T = RF cavity time constant Wo 3.2 A Simple Transfer Function f o r the Separator A s i m i l a r transfer function f o r the separator cavity and power amplifier can be evaluated by expanding the equivalent admittance expression about resonance. For maximum power transfer, the shunt resistance of the transformed plate c i r c u i t i s equal to the shunt resistance of the cavi t y so that the admittance of the separator power amplifier (PA), transmission l i n e , and cavity becomes: ^ e Zi Z2 sT r (1+STl) ( 1 + S T 2 ) Rsh Rsh , 9600 _„ 6000 where: T 2 = = 66us , x i = = 42us Wo Wo T * 200ns « T 1 . T 2 •••=» e s T* 1+sT Approximating the delay by a f i r s t order expansion gives: y ( s, 1 ••<!•«»> )( i . s i ] 2 r 1 ^ s ( x i + T 2 ) ^ . . . . . Rsh"^ - 2 J over p r a c t i c a l control bandwidths 13 The r e l a t i v e l y small delay i n the transmiss ion l i n e al lows the two resonant s t r u c t u r e s to be modeled as a s i n g l e R L C c i r c u i t . The t ransmiss ion l i n e adds a second pole at 1MHz. Z(s) « R 1+ST where: R = R s h T = ^ (T1+T2) The c a l c u l a t e d time constant of the model i s 54us, the average of the two time constants . Open loop pulsed power t e s t s gave a measured value of 55us f o r the time constant of the RF a m p l i f i e r s and c a v i t y . fig. 9 Pulsed power tests. RF s o u r c e . p u l s e w i d t h m o d u l a t o r p o w e r amp1 I f l e r c a v l t y amp 111 u d e d e t e c t o r -> to scope The photo shows two f a l l i n g edges of the detected c a v i t y vo l tage captured on a d i g i t a l o s c i l l o s c o p e . One trace shows the * 50us time constant of the R F system. The other trace records a c a v i t y spark which e f f e c t i v e l y short c i r c u i t s the c a v i t y and power 14 amplifier. This avalanche condition remains i n effect as long as the RF drive i s present. The 20Hz, 5% pulse modulator turns o f f the drive and the short c i r c u i t mechanism i s extinguished i n the period between pulses. Other tuned c i r c u i t s i n PA cathode c i r c u i t and the driver stage use discrete elements with Q's less than 100. This i s considerably less than the Q's achieved with d i s t r i b u t e d resonant structures In the c a v i t y and PA plate c i r c u i t . The effect of the energy stored i n the discrete coupling c i r c u i t s of the driver stage i s to s l i g h t l y increase the time constant of the system measured during pulsed power tests. These tests show that the delays and energy storage i n the driver stages and transmission l i n e can be described as a transconductance of the form: Ga Ga Ga a a j | (1+ST I ) 1+sJ^Ti 1+STa Vln Ga R V 1+STa > 1 + ST 7 Ta - 2flS T - 54US fig. 10 One dimensional transfer function for the RF system. Open loop pulsed power measurements indicate that a second order model i s s u f f i c i e n t to describe the dominant response of the RF system. The approximate pole locations determine adjustment ranges i n the regulator compensation networks. An estimate of Ga i s not needed i f the RF regulator design allows some gain adjustment. 15 This i s discussed l a t e r i n chapter 5. 3.3 Matrix Transfer Functions f o r Amplitude and Phase Regulation Regulation of both amplitude and phase of the separator gap voltage i s a two dimensional problem i n which a current vector from the power tube i s used to control the output voltage vector. The current vector can be described by a steady state term and a small time varying, modulation term. I + d/(t) = I e j 9 l + l e j 9 l [ ^ l + Jd6i(t)] The gap voltage i s the product of the generator current and the load impedance. The impedance can be written i n polar form as Zz<p where: |Z| = ^ tan# £ - T A W / 1 + ( T A W ) 2 At resonance, Aw=0 and the tube works into a r e s i s t i v e load. Vibration and other sudden changes can move the resonant frequency and cause large impedance changes i n high Q systems driven at a f i x e d frequency. When the gap voltage i s adjusted by small modulations of the cavi t y current, the effect of the detuned c a v i t y i s to cross couple the variations i n the phase and amplitude. To calculate t h i s effect i n high Q systems, i t i s desirable to express the detuned c i r c u i t impedance at the modulation sidebands. S t a r t i n g from the previous impedance approximation, the admittance can be written i n terms of a displacement from the c a r r i e r 16 frequency rather than a displacement from the resonant frequency, l e t : Aw = t o t a l frequency displacement from resonance wrf = displacement of the c a r r i e r frequency from resonance w = modulation sideband frequencies 1 ' ~ R " R R ~ Zrf R where: Aw = wrf+w Zrf = c i r c u i t impedance at the c a r r i e r frequency The control modulations w i l l appear as RF sidebands. The Laplace variable for the RF control is defined as s=jw, where w=Aw-wrf. The admittance of the c i r c u i t at these control frequencies i s : Y(s) = — + — a - L f 1 + S T — 1 Y l S ; Zrf R Z r f L 1 S T R J » 1 f i + S T -v „ _1_ |"(1+ST) - jtanBzl Zrf I- l-jtan9z J Zrf [ 1 - jtanBz J where: tan9z = -TWrf wrf ^ constant The impedance at the modulation sidebands for a cavity detuned to an a r b i t r a r y angle, 0z, from resonance i s : Z(s) s Zrf [ l- j tan9z 1 (l+ST)-jtan9zJ ^ r f l+ST+tan29z) -JsTtan0z] (1+ST) 2+ tan 29z This expression describes the impedance of an "off center" Q 17 curve. I f there are no control modulations then s = 0 and the c i r c u i t impedance i s Z(0) = Zrf. The steady state and modulation terms i n the gap voltage are given simply by Ohms law; the product of the current vector and the c i r c u i t impedance. V + d l^t) = V e j e v - V e j 6 v [ *™ + JdBv(t)] , ,, J6v „ T J9l where Ve = Zrf Ie In the Laplace domain (s=jw), the var i a t i o n i n the gap voltage can be expressed i n terms of the current modulation. The expressions are s i m p l i f i e d i f the amplitude variations of the current and voltage vectors are expressed as f r a c t i o n a l changes. ^ d W U + J d 0 v ( t ) j = v ( s ) + J e v ( s ) = [sx+(l+tan 28z) - JsTtanGz][i(s)+jei(s)) (l+sx) 2+tan 29z Separating the real and imaginary parts, t h i s expression can be rewritten i n matrix form to show the transmission of phase modulations and f r a c t i o n a l amplitude modulations through the high Q section. v(s) 9v(s) (l+sx)2+tan 29z sx+(l+tan9z) srtanGz i(s) -srtan9z sx+d+tan 9z) 9i(s) Cross Coupling Matrix for Amplitude and Phase Modulations 18 Driving the c i r c u i t off resonance causes the cross terms i n the matrix to increase by a factor tanGz. This detuning can come from several sources: • random vibrations and dimensional changes i n the plate and cavit y c i r c u i t s . • changes i n the cyclotron operating frequency. • The separator i s locked to the beam phase which changes due to a 5 Hz mechanical vibrat i o n i n the cyclotron resonators. The e f f e c t i v e impedance of the separator RF system i s the p a r a l l e l combination of the plate c i r c u i t and the cavity structure. No means i s presently available to automatically tune the plate c i r c u i t . The cavity, which exhibits considerable frequency d r i f t , i s independently adjusted by an automatic tuning loop that i s described later. It i s l i k e l y that the plate c i r c u i t and the cavity w i l l become tuned to different frequencies. In t h i s case, the e f f e c t i v e admittance of the high Q c i r c u i t can be rewritten as: Y(s) = — f 1 + s(xi+x 2) 1 K J ZrfL 2-j(tan9i+tan82) •* where - L = ^ + ^ Independent tuning effects of the plate c i r c u i t and the cavi t y are e a s i l y described by adjusting the terms i n the coupling matrix. tanBz -> }r (tan6i+tan02) and x -» i ( x i + X 2 ) 19 When automatic tuning i s r e s t i c t e d to the cavity, the regulator must be able to handle A.C. disturbances i n both the plate c i r c u i t tune, 9 i , and the cavity tune, 92, and D.C. errors i n the plate c i r c u i t tune, 9 i . The cyclotron usually operates between 23.05 MHz and 23.07 MHz, a 20 KHz range. With a plate c i r c u i t Q of 3000 and a c a v i t y tuner that works perfectly, the plate c i r c u i t can be detuned as much as | 9 i | s 70 which translates to a system detuning angle of |9z| s 54°. Before the separator i s run for an extended period of time, the plate c i r c u i t i s tuned to match the e x i s t i n g cyclotron frequency. If the cyclotron magnet i s stable and does not suffer any power bumps, the cyclotron frequency can usually be held within 5KHz of the i n i t i a l tune which l i m i t s the possible separator detuning range to |9z|s34° when the cavity tuning loop i s active. After the separator system has been manually adjusted, the expected detuning angle, 9z w i l l be zero. If the impedance matrix 2 3 i s expanded about tan 9z=0 and terms of tan 9z and greater are neglected then the transfer function matrix can be s i m p l i f i e d to: v(s) 9v(s) 1+ST . tan 2Gz 1 + ST U + S T T sxtanGz 1 + S T 1 + S T sTtan9z 1 + S T tan 8z K s ) 9i (s) ( 1 + S T ) ' This gives a convenient form to study the tuning loop as a perturbation of the resonant system. 20 i ( s ) 91 (s) i nput vec tor 1+ST srtanGa 1+ST tanGz 1+ST -1 tan9z 1+ST v(s) 9v(s) output vector fig. 11 Linearized model for high Q circuit elements. 3.4 P a r t i c l e Beam Loading as a System Disturbance The separator uses the RF e l e c t r i c f i e l d to select a given periodic group of charged p a r t i c l e s and reject others. As pa r t i c l e s enter the high voltage gap they d i s t o r t the e l e c t r i c f i e l d and so load the RF system. By superposition, one can determine the voltage disturbance induced on the deflection plates by a periodic group of charges t r a v e l l i n g between them. A test charge located near a grounded metal surface w i l l induce an image charge on the metal surface. S i m i l a r l y , a beam bunch t r a v e l l i n g across the gap electrode w i l l induce an image current on the electrode. The fundamental component of t h i s bunched beam induces an alternating current on the deflection plates which, i n turn, creates a disturbance voltage on the high impedance resonant cavity. Ideally, the cavity presents a low impedance to other frequency components of the bunched beam and, as a res u l t , these 21 components generate negligible disturbance voltage. The maximum induced current that can disturb the RF system i s equal to the fundamental component of the beam current i n the gap. Evaluating the spectrum of a periodic series of narrow pulses shows that the fundamental component of beam current i s approximately twice the average value of beam current. fig. 12 Beam loading seen by a single voltage probe. Approximately equal currents are induced on each electrode which produces a common mode disturbance voltage on the deflector plates. To a single voltage probe, the beam disturbance appears as an asymmetrical current source. A d i f f e r e n t i a l voltage measurement would be needed to cancel the common mode interference. In a system with only one measurement probe, both the beam and the generator currents affect the measured voltage but only the generator amplitude and phase are available as control variables. I f the beam loading current i s large then the control loops can become further cross coupled due to the steady state vector geometry seen by the single probe. Writing down the steady state 22 current that determines the measured voltage: I = J [ig COS*g + IbL COS * b L ) 2 +[lbL S in * b L + Ig sin$g)' f IbL SJnSbL + Ig SJnflg 1 [ Ig COS*g + IbL COS*bL J 0i= arc tan I The RF generator current provides the control input f o r the system. As with the previous matrix, the beam loading e f f e c t s are si m p l i f i e d i f the variations i n amplitude are expressed as fr a c t i o n a l changes. "dl " I _ u I cos(0i -$g) sin (0 i -*g) dig' Ig d8i -sin (0 i -*g) cos(0i -*g)_ d$g Beam loading e f f e c t s are ea s i l y demonstrated with $g=0 and $bL=-90 . For t h i s worst case, the control of cavity current by the RF generator vector can be written as: dl_ I d8i Ig+IbL Ig -IbL dig' Ig IbL Ig d$g where: Ig = amplitude of the generator current IbL = amplitude of the beam loading current It i s apparent that the cross terms i n the matrices increase as the beam loading current increases while the overall control gain decreases. When the beam loading current i s greater than the generator current, the phase and amplitude controls become reversed. 23 In the separator, the loading current has three components due to electrons, pions, and muons as indicated i n figure 3. I f a l l three p a r t i c l e groups were i n phase, the t o t a l beam loading current i s s t i l l less than luA. To produce a cavity voltage of 130KV into a r e s i s t i v e load of 75KIJ (Rsh/2), the RF generator must supply about g 1.7A of current; more than 10 times the beam loading current. In t h i s environment, the single voltage probe voltage produces v i r t u a l l y no d i s t o r t i o n of the gap voltage measurement. The only s i g n i f i c a n t cross coupling terms are introduced by high Q elements i n the RF amplifier and cavity. 24 Chapter 4 Closed Loop Control 4.1 Phase and Amplitude Loops The control vector for the separator RF system becomes rotated as the resonance of the output c i r c u i t d r i f t s with respect to the driv i n g frequency. This i s described by a perturbation matrix, A. v(s) <pvis) where: 1 + ST (1+STa)(1+ST) STtan6z ac( s ) ^ c ( s ) 2 tan 0z (1+ST)' STtanGz 1+ST 1+ST 1+ST tan2 8 z (1+ST)' Orthogonal phase and amplitude controls, ac(s) and <j>c(s), can be bu i l t to operate over a wide range when the inverse transfer matrix i s known. v(s) <pv(s) Ka (1+STa)(1+ST) ac(s ) 0 c ( S ) Very l i t t l e cross coupling i s introduced i n the low Q RF driver stages. The transfer function f o r these r e l a t i v e l y wide band sections can be approximated by the product of one dimensional expressions s i m i l a r to the system i n figure 10. An inverse matrix, 25 A~ provides the control decoupling at low signal levels so that the RF phase and amplitude can be regulated by two independent control loops. It i s not convenient to implement a solution in t h i s form because the A matrix requires an updated estimate of tanGz. The system becomes further complicated i f one must estimate cross terms i n the RF drive r stages. If a l l the cross coupling terms are treated as system disturbances, then the transfer function of the RF system can be described by a simple second order expression l i k e that shown i n figure 10. Even without the decoupling matrix i t i s s t i l l possible to b u i l d a co n t r o l l e r for t h i s system around two independent second order regulators i f the cross terms are treated as system disturbances. For control purposes, i t i s convenient to represent the entire RF system as a 2x2 matrix that maps the control input vector into the output space. Measurement and control functions are also represented by 2x2 matrices while the system signals are described by two dimensional vectors. Vref r e g u l a t o r RF sys tem c Gc measurement w Vout fig. 13 Block diagram of the RF regulator loop. 26 The gap voltage vector, Vout, can be written i n terms of the set point vector, Vref as: Vout = [Q+PGCMJ-VGC Vref = [W+(PGc) _ 1] _ 1 Vref I f the system i s controlled by two independent regulators then Gc becomes a diagonal matrix. The control loop contains an amplitude detector and a phase detector, both of which can be constructed to provide independent measurements thus eliminating cross terms i n the measurement matrix. Only the RF matrix, P, contains cross terms. Gc = g u 0 0 g 2 2 M = mil 0 0 m 2 2 P = p i l p i 2 p 2 1 p 2 2 where: pn= p 2 2 and p i 2 = - p 2 i The plant matrix, P describes the rotatio n of orthogonal phase and amplitude modulation vectors. At low modulation frequencies ( ST«1), the rot a t i o n i s given by tan# sxtanSz and at higher frequencies (sx»l) the rotation approaches <f> = 0z. 27 Phase and amplitude changes i n the gap voltage are shown along the v e r t i c a l and horizontal axis. The matrix c o e f f i c i e n t pn i s the projection of the amplitude control vector, v i , on the amplitude axis and p22 i s the projection of the phase control, V2 on the phase axis. The determinant of the IP matrix, p n p 2 2 - p i 2 p 2 i i s the area of the parallelogram defined by the input vectors. After some algebra, one can write the closed loop transfer function f o r the 2x2 system. [M+(PGc) ] i - i - i bn 0 0 b22 forward terms 0 bi2 b2i 0 cross terms The forward transfer functions are: b n = mi l + k22p22+l g l l pi 1(k22p22 + l)-k22pi2p21 mi 1 + p22 g l l piIp22-p2ipi2 where k22 = g22m22 and k22p22 » 1 b22 = m22 + k i l p i l + l g22 p 2 2 ( k l i p i l + l ) - k l i p i 2 p 2 1 m22 + pi 1 g22 p22piI~p2ipi2 where k n = gu m i i and k n p n » 1 These forward transfer functions are of the usual form: Hii = Gi i 1 + mi iGi i where Gn = gi kjjpjj » 1 PJ j = g i i p i i , kjjpjj « 1 When an amplitude change i s made i n the reference vector the regulators w i l l attempt to hold the gap phase constant. At 28 frequencies where the regulator gain i s large, the projection on the y axis, p22 can be held constant and the e f f e c t i v e amplitude gain i n the plant becomes the projection on the x axis. This projected amplitude gain i s given by the area of the control vector parallelogram, |P|, divided by p22. ( f i g . 14) Within the regulator bandwidth, the cross coupling i s reduced by the loop gain and a factor equal to the area spanned by the control vectors. b l 2 _ g22pi2 a P H . ( k l l p i 1 + 1)(k22p22 + l ) - k l I k 2 2 p i 2 p 2 1 kl1m22(pi 1p22-pi2p21) b21 = g " P 2 1 - P l i -( k l l p i 1 + 1 ) (k22p22+l )-kl Ik22pi2p21 k22IM 1 (pi 1p22~pi2p21 ) Two independent PID regulators are used to cancel the dominant poles i n the phase and amplitude control loops. Figure 15 shows a simulation of the small signal step response for a detuned RF o system with 0z ±27 . When the loops are closed, cross coupling between phase and amplitude control i s p r a c t i c a l l y eliminated. Simulations show that two independent regulators for phase and amplitude can reject cross coupling terms introduced by an RF system operated near resonance ( |Sz |50°). A c a v i t y tuning loop i s included i n the separator system. While the tuner does reduce cross coupling between the phase and amplitude control signals, i t s chief task i s to match the c a v i t y 29 impedance to the transmisson l i n e and minimize ref l e c t e d power seen by the tube. 0 1.1 0 1 .9 .8 .7 .6 .5 .4 .3 .2 .1 S»all Signal Step Besponse of Detuned BF Sustea open loop cross tern closed loop forward tern with PID regulator legulator Gain = 1BE6 Detuning Angle = 27 deg. note rejection of cross term in closed loop system v closed loop cross ten with PID regulator I V ' v r - ' i i i i t • i • • i i i i i — i — e .a 1 Ti«e t»l&-5) fig. 15 Regulator simulation showing cross term rejection. 4 . 2 The Cavity Tuning Loop The model developed f o r the RF system allows both the power amplifier and the cavi t y to be tuned to resonant frequencies that d i f f e r s l i g h t l y from the dr i v i n g frequency. The ef f e c t i v e detuning fo r t h i s system, tanGz, i s the average of the cavity detuning and the plate c i r c u i t detuning. tanGz = ^  (tanGi + tan62) where: tan8i = plate c i r c u i t detuning tanG2 = cavity detuning A tuning loop i s used to control the cavity d r i f t and force a steady state value of tan62 = 0. The tuner mechanism produces 30 small mechanical deformations of the cavity which in turn produces a s h i f t i n the cavity resonance. A linearized transfer function for the tuner mechanism and the tuning loop i s of the form: KM d K2 AW2 T s(1+STM) s tan02 = TAW2 m o t o r + g e a r l n g c a v i t y s m a l l c a v i t y f r e q .a3 j u s t m e n t t ime c o n s t a n t fig. 16 Cavity tuning mechanism. KM * 2mm/s/V, K2 - 5KHz/mm - 104n/s/mm, T * 70us, TM 0. Is Cavity tune i s measured by a phase detector which compares the phase of the transmission l i n e voltage injected into the cavi t y and the gap voltage. At resonance, these two voltages are i n o phase. A 90 phase s h i f t i s introduced between the signals so that the phase detector w i l l r egister zero when the cavity i s driven at i t s resonant frequency. Operating near resonance (|tan92|s .5), such that t a n 8 2 62, the output of the phase detector w i l l give a good estimate of the c a v i t y tune. The tuning loop, shown below, i s designed to operate with the reference, R = 0. R = 0 - ^ E ) - 1. K C M t u n e r > D I d1st urbance c o n t r o l l e r Kd -> tan02 phase d e t e c t o r fig. 17 Cavity tuning loop. 31 Output from the phase detector i s a linear function of the phase difference between the RF injected into the cavity and the gap voltage. The detector responds much faster (lus) than the mechanical tuning system (100ms) with the result that when a simple proportional controller i s used, the loop transfer function i s dominated by the motor. Disturbance inputs such as s t a t i c tuning errors and long term d r i f t s are reduced by the closed loop. t a n 9 2 = H = K t 1 + H S(1+STM) The r a t i o of reflected power to forward power seen by the cavity transmission l i n e i s : f 2 = | Z - R o | 2 _ ( R L - RO) 2+ X 2 IZ+RoI2 (RL+RO) 2+ X 2 where Z = . , ? ° „ = RL + jX l+ j t a n 0 2 ° = transformed cavity impedance seen by the transmission line „2 tan 4 82+tan 2 62 tan 2 6 2 „2^ _ = — — - — ^ — for r ^ 0. 1 (2+tan 92) +tan 92 To keep the power reflected from the cavity to less than 1%, the tuning loop must keep |62|sl0°. This e a s i l y achieved over the control bandwidth however, the tuner must transmit mechanical motion through a vacuum seal and there i s reason to reduce the bandwidth of t h i s movement. Depending on the disturbance spectrum, the ref l e c t e d power can show s i g n i f i c a n t fluctuations while the average value i s regulated to less than 1%. 32 The c a v i t y time constant of about 70us produces a lag between a phase change injected into the cavity and the appearance of the change at the cavity gap. Phase differences measured across the cav i t y are the result of tuning disturbances, D, and the propagation delay of phase modulations through the cavity. The tuning loop treats the output of the phase detector as system disturbances and attenuates both signals by a factor (1 + H). In operation, the loop keeps the cavity tuned near resonance such 2 that terms of tan 82 can be neglected and the phase signals i n the ca v i t y can be represented as: <f>(s) cav1t y input phase , ^ ST2tan82 V i S j , , cross term 1+ST2 D d i s t urbance (pout ) cavity output phase u 62 to tuner phase measurement fig. 18 Tuner small signal model. 82 = D + — — f 0(s) + v(s) S T 2 t a n 6 2 1 - </>(s) 1+ST2 r 1+ST2 •> r = D + ST2 1+ST2 r . , ST2tan82 ,, C V ( S ) 1+ST2 " * ( S ) ) Both the cross modulation terms and tuning disturbances, D are attenuated by the tuning loop. Phase modulations are treated d i f f e r e n t l y . The cavity output phase i s 0out = 82 + #(s). Setting D and v(s) to zero such that only the phase control modulations interact with the tuning loop, one can write #out i n terms of the 82 #(s) and the closed loop disturbance of the tuning system, T—rj • 33 = C 1 - iSi T+H ) * ( s ) = T^— f 1 + TTu S T 21 *ts) 1+ST2 1+H J Over the range where |H| » 1, the tuning loop e f f e c t i v e l y f l a t t e n s the c a v i t y response to phase modulations such that (pout = 0(s). The bandwidth of the motor driven tuning loop i s of the order of a few hertz, much less than the cavity bandwidth of » 2.5KHz and much less than the regulator bandwidth. As a resu l t , the separator tuning loop i s not strongly coupled with the amplitude and phase regulation. Since cross coupling introduced by a detuned RF system i s treated as a disturbance by the regulator system, the tuning loop acts l i k e a slow moving, feedforward term that i s well within the bandwidth of the regulators. From the control point of view, the cavity should be tuned to be the complex conjugate of the transformed plate c i r c u i t impedance to give 8 i = - 6 2 and tan8z=i/2(tan8i+tan82)=0. In i t s present form, the tuning loop eliminates phase and amplitude cross coupling due to c a v i t y d r i f t s but i t has l i t t l e influence on effects introduced by a detuned plate c i r c u i t . It i s not d i f f i c u l t to change the e x i s t i n g resonance tuner to a conjugate tuning configuration but t h i s may change the power amplifier design. 34 Chapter 5 Hardware Implementation of the Controller 5.1 Overview The RF cont r o l l e r handles routine system f a u l t s , provides automatic st a r t up, and keeps the cavity tuned to the cyclotron frequency. It also regulates the phase and amplitude of the gap voltage, providing wideband control of the p a r t i c l e f l u x delivered to the experiment. The system i s b u i l t around analog control loops which were designed to have unity gain bandwidths greater than lOOKHz. These loops are supervised by a computer which controls the setpoints and the regulator parameters. C o m p u t e r ^ RF R e g u l a t o r ^ Ana l o g L o o p fig. 19 Basic controller concept. 5.2 System Description The RF con t r o l l e r was developed as a modular system. Individual system functions were i d e n t i f i e d and then hardware modules were RF S y s t e m } -> V R F 35 b u i l t to perform that single function. Such modules can be e a s i l y changed or upgraded and complex systems can be pieced together from these basic elements i n much the same way that they are assembled on paper using block diagrams. Beam S i g n a l s e 1 f - e x c i t e d \ P h a s e L o c k ) sn d r i vt r C y c l o t r o n RF RF s o u r c e swi t c h C a v l t y Vo 11 a g e Feedbac k s i g na1 s p l i t t e r amp 1i t ude d e t e c t o r p h a s e d e t e c t o r amp1i t ude r e g u l a t o r p h a s e r e g u l a t o r amp 1 1 mo du t ude a t o r p h a s e m o d u l a t o r P h a s e R e f e r e n c e Fig. 20 Control modules. RF O n / O f f s w i t c h — > RF Out to PA , t L i n t e r l o c k t r i p RF d u t y e y e 1 e 1 R e g u l a t o r I/O Bus -> t o o p t i c a l c o u p l e r s A dedicated computer provides access to the con t r o l l e r variables. It also handles graphic display of the system information and sequencing f o r automatic start-up and spark recovery. The computer permits the modular design to be extended to the user interface, sequencing operations, and upgrading to provide new system states. The regulator I/O lines connect to the same backplane as the computer but are o p t i c a l l y isolated from the bus. The software f o r automatic sequencing and system I/O was developed 36 on an MS DOS computer using compiled BASIC. This system had a l l of the graphics and timer support needed to produce the bar graphs, system p i c t o r i a l s , and sequencing operations. The controller was b u i l t using a 7MHz PC on STD bus with Prolog's System 2 operating system. This configuration worked well and could e a s i l y handle the necessary tasks using a simple p o l l i n g loop. A gas plasma display i s used as the front panel display. It i s compatible with IBM's EGA graphics but unlike a CRT i t i s not sensitive to magnetic f i e l d s or phosphor burn. The amplitude and phase control can operate i n open loop or closed loop with the state of each loop shown graphically on the front panel display. Set point and readback values are displayed numerically and with horizontal bar graphs. The modulator drive signals and an adjustable drive l i m i t value are displayed i n a v e r t i c a l bar graph. System status i s shown i n the right side of the display. setpoint 2843 AMPLITUDE LOOP e 2 4 6 6 18 8 2 4 6 1 1 1 8 18 C a i n : 58 32 (9 T a u l i l a u D : readkack 2843 18 C D 8 S 4 2 . 8 . axpl i tilde mdulator setpoint -1848 PHASE LOOP 1 1 I I I I I I I I I IB 8 18 I I I I I — I I I I I I -18 8 18 rea<U>aclc 14 control —i MANUAL interlocks OK source DRIVEN • Mode drive RF OH Fig. 21 Front panel display. 37 During s t a r t up, the RF voltage i s pulsed to overcome multipactoring i n the cavity. A peripheral card f o r the STD bus was b u i l t to handle t h i s operation. The computer writes the pulser frequency and pulse width to t h i s card. When the system i s operated i n pulsed mode, the pulse width can be adjusted by a front panel knob from 0% to 100%. In CW operation, the RF i s turned On or Off by writing a pulse width of 100% or 0%. Push buttons on the front panel are used to toggle the system states such as RF On/Off, RF Pulsed/CW, etc. Continuous parameters can be adjusted using f i v e o ptical shaft encoders on the front panel. Each loop has one shaft encoder that can be assigned to adjust either the regulator gain, the two transfer function zeros, or the drive l i m i t for the loop. The remaining three shaft encoders are not assignable but remain dedicated to the amplitude and phase set points and the RF pulser duty cycle. The assignable knobs are used mainly during setup and commissioning while the dedicated knobs are used during manual operation of the RF system. Keeping multiple assignments to a minimum reduces the complexity of the front panel controls. 5.3 Self-Excited Operation To st a r t the RF system, the cavity and power amplifier usually need to be tuned before f u l l power i s applied. Experience with the cyclotron cavity has shown that i t i s best started i n s e l f - e x c i t e d mode where the signal from the cavity pickup i s fed back to the low level RF driver amplifier. The sel f - e x c i t e d frequency i s 38 determined by the cavity resonance and the phase s h i f t around the loop. FEEDBACK RF SAMPLE FROM CAVITY SIGNAL AMPLIFIER POWER AMPLIFIER DC BLOCK -> BANDPASS FILTER fig. 22 Self-excited configuration. Self-excited operation provides an idl e mode where the RF system operates independent of a reference phase or frequency and can be tuned by a phase s h i f t e r i n the feedback loop instead of the usual mechanical tuning loop. This arrangement has proved to be useful fo r commissioning and debugging the various RF systems on s i t e and has become a standard requirement for TRIUMF RF systems. i n t e r l o c k s Ok d r i v e F r e q . Ok ) 1 >—— * Rf o f f s e l f e x c i t e d d r i v e n i ( > <— > fig. 23 Principal system states. I f beam production i s interrupted, the separator looses i t s RF reference signal and the system waits at f u l l power i n sel f - e x c i t e d mode u n t i l beam i s again delivered to the experiment. S i m i l a r l y , from a cold s t a r t , the system waits at f u l l power i n s e l f - e x c i t e d mode u n t i l the cav i t y frequency i s matched to the reference frequency. 39 Limiters and a bandpass f i l t e r are included i n the se l f - e x c i t e d signal path. The l i m i t e r s keep the signal amplitude the same as i t would be i n the driven mode while the bandpass f i l t e r ensures that the cavity i s se l f - e x c i t e d by the desired push-pull mode. The s e l f - e x c i t e d system behaves l i k e a c l a s s i c a l o s c i l l a t o r i n which the t o t a l phase s h i f t around the loop i s 2nir. When the loop i s i n i t i a l l y closed, system noise generates a current i n the power tube which excites the high Q plate c i r c u i t and cavity. I f the delay around the loop at the cavity resonant frequency, wo, i s 2rm-<p then a steady state frequency i s reached when the cavity i s excited at a frequency above resonance, w=wo+Sw where the cavity provides the extra phase lag to make the loop delay equal to 2n7r. The impedance of the cavity at t h i s frequency i s : z h = l z l ^ — l 20 At steady state, <t> = tan f — Sw") and the sel f - e x c i t e d frequency v. Wo -' i s given by: /-, tan®-> w = wo + 5w = Wo[1+ 9 n J 5.4 Self-Excited and Driven Tuning Systems D r i f t i n the cavity resonance or i n the phase delay around the loop w i l l cause the system to operate off resonance. In terms of refle c t e d power and load matching, the self- e x c i t e d mode, l i k e the driven system, needs an automatic tuning loop. When i n driven 40 mode, the mechanical tuning loop keeps the cavity tuned to resonance and minimizes reflected power. In s e l f - e x c i t e d operation, the tuning mechanism drives the cavity to the cyclotron frequency while the phase regulation loop minimizes reflected power. phase r e g u l a t o r RF s o u r c e s w l t c h phase modulator RF feedback loop RF amp 1 i f l e r TX 1 i ne c a v i t y phase d e t e c t o r t u n e r f r e q u e n c y c o mpar e AF Ok < d r i v e <-p r e s e n t RF d e t e c t manual d> O— \ auto r e f e r e n c e f r e q u e n c y fig. 24 Self-excited tuning loop. In the s e l f - e x c i t e d mode the frequency comparator allows the tuner to adjust the cavity resonance to within 1 KHz of the cyclotron RF. This module produces a signal that i s positive i f the s e l f - e x c i t e d frequency i s greater than the cyclotron frequency, negative i f the frequency i s less than the reference and zero i f the s e l f - e x c i t e d frequency i s within 1 KHz of the cyclotron frequency. The phase regulator maintains a delay of 2nrc around the s e l f - e x c i t e d loop. The bandwidth of the phase regulator i s very much greater than the mechanical tuning system and no ref l e c t e d power fluctuations are detected when the se l f - e x c i t e d tuning loop 41 i s functioning. Under automatic control, the system can be taken into the driven state when (AF_Ok AND Drive_Present) i s true. r e f e r e n c e f r e q u e n c y 1 RF s o u r c e swi t c h phase r e g u l a t o r TX 1 i ne phase n o d u l a t o r RF amp 1 i f i e r d r i v e <-p r e s e n t phase d e t e c t o r c a v i t y phase d e t e c t o r RF d e t e c t auto q / o -manual t u n e r r e f e r e n c e f r e q u e n c y fig. 25 Driven Tuning Loop In driven mode, the phase regulator locks the cavity signal to the RF reference signal. When Drive_Present i s False, the c o n t r o l l e r w i l l return to the RF Off state. Under automatic control, i t w i l l bring the system back to the self - e x c i t e d state and wait f o r the RF drive. To change from se l f - e x c i t e d to driven, the RF source i s switched from the cavity to the reference signal. A DPDT switch box configures the control signals for the tuner and for the phase regulator. 42 R e f e r e n c e - ) -phase d e t e c t o r C a v i t y - » -TX 1 lne4-X s w i t c h c o n t r o l dr i ven -O -> to Phase R e g u l a t o r phase d e t e c t o r Cavi ty-*-f r e q u e n c y c o mpar e Ref erence-)--> to Tuner Dr 1 ve s e l f Tuner S w i t c h Module fig. 26 Signal switch for tuner and phase regulator. 5.5 The Frequency Comparator The automatic c o n t r o l l e r must be able to tune the se l f - e x c i t e d cavity to the reference frequency before the system i s driven. Necessary information for t h i s task i s provided by the frequency o o comparator which uses 2 mixers, a 0 power s p l i t t e r and a 90 power s p l i t t e r to derive two signals, sin(wi-w2)t and cos(wi-w2)t. With respect to the cosine term, the sine term i s inverted when W2>W1. COs(wi-W2)t = COs(w2-Wl)t sin(wi-w2)t = -sin(w2-wi)t The quadrature signals are converted to d i g i t a l waveforms and decoded by simple logic c i r c u i t s . Experience has shown that when the two frequencies are matched to within 1 KHz, the system can be driven. A retriggerable one shot and a latch detects t h i s condition. The time constant of the one shot determines how clos e l y the frequencies are matched. 43 ml xer W2 s p l i t t e r low pass f 1 I t e r Sin(a>l-6)2)t mixer 90 s p l i t t e r low pass f 1 I t e r D Q C 0 -> <J1>U2 -> W2>W1 COs((Jl-W2)t T one shot -> Aw>-T fig. 27 Frequency comparator. 5 . 6 Spark Detection Sparks i n the RF cavity e f f e c t i v e l y short c i r c u i t the ca v i t y and power amplifier. The avalanche condition, which the spark i n i t i a t e s , i s extinguished when the RF drive i s removed. To prevent damage to the RF system, the drive i s turned o f f f o r a minimum of 1 second when a spark i s detected. A photograph of the dif f e r e n t waveforms for normal RF Off and a spark are shown i n figure 9. The spark detector responds to amplitude signals that f a l l 70% i n less than 5us. Normal RF Off signals decay to t h i s l e v e l i n about 50us and do not trigger the c i r c u i t . F ron Amplitude Detector •ne Shot C o n p o . r o . t o r fig. 28 Spark Detector 44 5.7 System Hardware Configuration Long cable runs (60m) between the cavity and the co n t r o l l e r can introduce ground loop noise into the system. The connection diagram f o r hardware modules includes DC blocks and op t i c a l couplers that isola t e the con t r o l l e r from 60Hz ground loops. The DC blocks are coupling capacitors that pass only the RF signals on the cable s h i e l d and center conductor. Connections to the tuning motor and to the computer I/O bus are o p t i c a l l y isolated. o Phase detectors operate over a r e s t r i c t e d range, usually ±90 . The con t r o l l e r contains 4 manual phase s h i f t e r s which are adjusted when the system i s i n s t a l l e d . One of the phase s h i f t e r s adjusts the phase delay i n the self- e x c i t e d loop. The other three set the operating range of the phase detectors. Bandpass f i l t e r s are included on the reference input and the cavity feedback. They have a pass band of ±1 MHz about the 23 MHz o center frequency and introduce less than ±1 phase error over the separator's 10 KHz operating range. The f i l t e r i n the cavity feedback path attenuates out of band noise before the signal i s presented to the phase and amplitude detectors. A f i l t e r on the reference input i s needed to condition the various signals that can provide the reference frequency. 45 TRANSMISSION LINE EC-BLOCK 23 MHZ BANDPASS FILTER FEEDBACK RF SAMPLE FROM CAVITY DC BLOCK 23 MHZ BANDPASS FILTER 6 WAY POWER SPLITTER tuner odjuat —> 0 sample PHASE DET. I/O MANUAL TUNING sample, PHASE DET. TUNER SIGNAL SWITCH TUNER OPTO MOTOR CONTROL COUPLER DRIVE AF DET. 0 self-excited adjust AMP DET. SPARK DETECT RF PRESENT -> I/O detector zero <— 0 I/O RF SOURCE z> m m O o I* (/) r-•H C o PHASE MOD AMP. MOD. I/O 23 MHZ BANDPASS FILTER PVM 8. RF SWITCH DC BLOCK RF OUT MOTOR) 4 WAY POWER SPLITTER CYCLOTRON REFERENCE FREQUENCE PS 0 H L A - V external 23 MHZ BANDPASS FILTER DC BLDCK LOCK TD BEAM RF DET. BEAM LINE 1A CAP PROBE SIGNAL I/O fig. 29 Separator RF Control Modules 5.8 Regulator E l e c t r o n i c s A s i m p l i f i e d diagram of the regulator loop and i t s connection to the RF system i s shown i n figure 30. It i s a PID configuration i n which 8 b i t multiplying DACs are used to change the loop gain and the regulator zeros. A 12 b i t DAC i s used to control the set point and an 8 b i t ADC i s multiplexed to monitor the drive level and the detector output. Wide band phase and amplitude modulators and detectors add high frequency poles to the RF system. These effects and the high order poles i n the regulator are not included i n the s i m p l i f i e d diagram because the c i r c u i t i s designed to reach unity gain before these high order terms affect the loop s t a b i l i t y . Control signals to the RF modulators are generated by operational amplifiers and vary between ± 10V. Signals from the amplitude and phase detectors are processed by si m i l a r devices and are also i n th i s ±10V range. Within an order of magnitude, one can write the product of the plant gain and the measurement gain as KaKm « 1. If the regulator design cancels the two dominant poles in the second order plant and is unity gain stable then the regulator gain can be adjusted to be closed loop stable in all TRIUMF RF systems where KaKm ~ 1. An 8 b i t multiplying DAC i s used as a variable r e s i s t o r to adjust the controller gain, Kc, over 2 decades (48db) between 10 5 to 2 .5xl0 7 . T i s adjustable from 0 to 330us i n 255 steps. This range i s able to compensate for Q's usually attained i n copper c a v i t i e s . 47 R-P l o p e n l o o p s w i t c h XD-\ 1 Kc , . A. — ( 1 + S T ) s o ; > 1 o p e n (1+STf) c l o s e d l o o p s w i t c h c 1 o s e d l e a d (1+STa) Regulator f-Ka (1+ST)(1+STa) m e a s u r e m e n t Km (1+STm) -» Plant fig. 30 PID Reguiator and RF System Model When the RF system i s operated i n open loop, the PI term i s strapped f o r unity gain and follows the set point. This prevents the integrator from d r i f t i n g and permits a nearly bumpless transfer between open loop and closed loop control. Ideally, xe should be equal to T and, at the cost of increased complexity, i t i s possible to make T f adjustable so that i t tracks T . A simpler c i r c u i t configuration with a fixe d value of T f was used to make V the regulator unity gain stable under a l l settings of Kc and T . The derivative term i s r e s t r i c t e d to the feedback path. This configuration provides less DAC noise on the output as the set point changes. It i s adjustable from 0 to lOOus i n 255 steps. For most systems, the dominant pole can be canceled i n either path. 48 The s i m p l i f i e d schematic of the regu la tor shows an op-amp lead c i r c u i t i n the feedback path. T h i s c i r c u i t does not perform wel l at h igh frequenc ies . A bet ter d i f f e r e n t i a t o r can be b u i l t from d i s c r e t e components as shown i n f i gure 32. It i s a c a p a c i t o r i n s e r i e s wi th a common base stage with emit ter fo l lowers on the input and output. The current i n the common base stage i s 4.5mA which g ives an input re s i s tance of 5.5fl. The output r e s i s t a n c e o f the emi t ter f o l l o w e r feeding the capac i tor i s about 2.5fl. These values (8f2 and O.Oluf ) ind ica te that the d i f f e r e n t i a t o r phase o should f a l l to 45 around 2MHz. T h i s agrees with the s i m u l a t i o n . 12U-L 8.1UF 4—!•• 1 . 2 1 I 2 . 7 1 8. IUT 0 2 2 . 2 1 •4-120 fig. 32 Discrete Component Differentiator 49 Discrete Differentiator Response 100 IK 10K 100K 1M Frequency in Hz. fig. 33 Differentiator Simulation Several companies now provide operational amplifiers with bandwidths i n excess of 20 MHz. Most of these devices are transimpedance amplifiers which have a high impedance positive input and a low impedance negative input. The negative input i s t y p i c a l l y a common base stage which maintains a low input impedance over a wide frequency range. Bench tests show that a useful d i f f e r e n t i a t o r can be b u i l t using one of these devices. The transimpedance op-amp i s also well suited to the multiplying DAC c i r c u i t s i n the PID regulator. The DACs have a large output capacitance (120 pf) which can reduce the s t a b i l i t y of voltage op-amp c i r c u i t s unless the high frequency gain i s reduced. Bench tests indicate that the bandwidth and gain of the regulator can be considerably improved i f current input op-amps are used with the multiplying DACs. 50 Chapter 6 Software 6.1 The Main Program Loop A program was written to monitor system operation and supervise t r a n s i t i o n of the system from one sta t e to another. It also provides an operator interface and graphic d i s p l a y of the c o n t r o l l e r variables. Manual or automatic control can be selected from the front panel. The software permits more f l e x i b i l i t y under manual c o n t r o l . In automatic mode the system i s constrained to fo l l o w r i g i d rules. The c o n t r o l l e r follows a p o l l i n g loop of the form: Do While Control=True Scan Knobs Scan Buttons state transition requests Read_Inputs Fault Control { apply fault rules Control Devices { apply control rules Display_Data { update display End While To achieve a new state, the machine needs to consider two input vectors; the present state and external input. The present state i s available i n the machine memory while external input can come from the operator or from the RF system as a detected spark or loss of drive, etc. External input acts as a request to change the resident image of the system state and the appropriate hardware. Safe operation of the RF system does not permit a r b i t r a r y t r a n s i t i o n from a given state to any other state. In manual mode and i n automatic mode, the requests f o r state change are f i l t e r e d by control rules. As experience i s gained with the system, the control rules are changed to accommodate new functions. 6.2 Task Communication System information i s stored i n global variables, available to a l l tasks. Flags are used to communicate between tasks. The two p r i n c i p a l tasks are display and control and the flags associated with these tasks are: Amplitude_Display_Mail Phase_D i sp1ay_Ma i1 Sys t em_D i sp1ay_Ma i1 Pulser_Display_Mai1 Amplitude_Control_Mai1 Phase_Control_Mai1 Sys t em_Co nt ro1_Ma i1 Pulser Control Mail Each mail f l a g i s two bytes long and has an internal structure i n d i c a t i n g the individual requests. If the f l a g i s zero, the task i s not invoked. If the f l a g i s non-zero, b i t s are reset i n the f l a g as each request i s processed. 52 6.3 Control T a s k s A task such as Control_Devices is of the form: SUB Control_Devices Check_Control_Rules IF System_Control_Mail THEN System_Control IF Amplitude_Control_Mail THEN Amplitude_Control IF Phase_Control_Mail THEN Phase_Control IF Pulser_Control_Mail THEN Pulser_Control END SUB Check_Control_Rules filters the requests in individual control flags to make them compatible with the present system state. Changes are made to the system i f the validated control flag is non-zero. Program constants have been declared as integer masks that test individual bits in the appropriate display and control flags. If an individual bit tests TRUE then that specific action is taken and then the bit is reset. This process will clear all the bits in the control flag. Amplitude_Control is typical of the control modules. Sub Amplitude_Control IF Amplitude_Control_Mail AND Setpoint_Flag THEN OUT Amplitude_Setpoint_Port, Amplitude_Setpoint Amplitude_Control_Mail = Amplitude_Control_Mai1 XOR Setpoint_Flag END IF 53 IF Amplitude_Control_Mail AND Tau_D_Flag THEN OUT Amplitude_Tau_D Port, Amplitude_Tau_D Amplitude_Control_Mail = Amplitude_Control_Mai1 XOR Tau_D_Flag END IF END SUB Phase_Control and Amplitude_Control write new values to the phase and amplitude regulators. The loop variables that are changed by these routines are: • Setpoint • Limit - hardware l i m i t for the modulator drive • Gain - loop gain • Tau_I - regulator zero • Tau_D - regulator zero 6.4 F r o n t P a n e l Input A peripheral card was b u i l t to latch the front panel knobs and push buttons. When a button i s pushed or a knob turned, the event sets a single b i t i n one of two 8 b i t registers. These two registers are read during the p o l l i n g loop and ANDed with masks that enable inputs compatible with the present machine state. Action i s taken only i f the result i s non-zero. The hardware automatically clears the registers at the end of the read cycle, minimizing latency i n scanning the front panel. 54 Push buttons are assigned to toggle boolean system variables. Both the display and control flags are set by front panel input that i s enabled by the appropriate masks. No problems have occurred with t h i s procedure, however, i t i s probably better to set the display f l a g when the actual control i s accomplished. SUB Scan_Buttons Push_Buttons = INPUT(Push_Button_Port) AND Button_Mask IF Push_Buttons THEN IF Push_Buttons AND On_Off_Flag THEN System_Display_Mail = System_Display_Mai1 OR On_Off_Flag System_Control_Mail = System_Control_Mail OR On_Off_Flag END IF o o o o o END IF END SUB Input from the front panel shaft encoders i s more complicated than the boolean information received from the push buttons. I f the Knob_Register i s non-zero then the Direction_Register i s read. I f a b i t i s set i n the Knob_Register then the Direction_Register b i t i s tested to see i f the associated variable should be incremented or decremented. The changed variable i s validated to ensure that 55 (0 £ value £ max_value) and then the appropriate b i t s are set i n the control and display flags. The system only responds to a person turning one knob at a time. SUB Scan_Knobs Knob_Register = INPUT(Knob_Port) AND Knob_Mask IF Knob_Register THEN Direction = INPUT(Direction_Port) Increment setpolnt step size ( i n i t i a l value = 0) IF Setpoint_Step < Max_Step THEN Setpoint_Step = Setpoint_Step +1 D IF Knob_Register AND Amplitude_Setpoint_Knob THEN IF Direction AND Amplitude_Setpoint_Knob THEN Amplitude_Setpoint = Amplitude_Setpoint + Setpoint_Step ELSE Amplitude_Setpoint = Amplitude_Setpoint - Setpoint_Step END IF • validate 0 S setpolnt S max_setpoint • set control , display flags 2) ELSE IF o Phase set point 3) ELSE IF o amplitude loop parameters 4) ELSE IF o phase loop parameters 5 ) ELSE IF o pulse width END IF no knob turned, relax setpolnt step size IF Setpoint_Step > 0 THEN Setpoint_Step = Setpoint_Step - 1 END SUB 56 One advantage of scanning the front panel knobs i n t h i s fashion i s that i t produces no large step functions outside the response time of the control and display tasks i n the p o l l i n g loop. The loop can be slow (>10ms) and ramping the voltage with a 12 or 16 b i t DAC becomes a tedious job with a system that looses shaft encoder counts. The apparent response i s changed by incrementing the setpoint by an amount that depends on how fast the front panel knob i s turned. If the program detects that each time through the p o l l i n g loop, the setpoint i s always flagged, then the setpoint knob i s being turned f a s t e r than the system can respond. The si z e of the setpoint step i s increased u n t i l i t reaches a maximum value or u n t i l the program detects that the setpoint i s not flagged and then the setpoint step i s decreased. In t h i s way the " f e e l " of the system i s t a i l o r e d to s u i t manual operation. 6.5 Auto S t a r t The i n i t i a l start-up condition i s : RF Off Phase and Amplitude loops open Amplitude setpoint = 0 mode = se l f - e x c i t e d When a spark i s detected, hardware immediately turns o f f the RF drive and sets an I/O b i t that i s scanned by the computer each 57 time through the p o l l i n g loop. The computer w i l l r estart the system i f i t i s i n automatic mode. IF Spark_Detected THEN Spark_Count = SparkjCount + 1 IF Spark_Count = 1 THEN Auto_End_Amplitude = Amplitude_Setpoint I n i t i a l i z e Display_Message("SPARK: waiting f o r vacuum") Pause (8) wait 8 seconds System_Control_Mail = Auto_Start_Flag Fault = True Sparks can occur during the auto sta r t process. If sparking occurs too many times the system w i l l abandon i t s attempts to st a r t and wait f o r an operator. Auto_Start can be halted by a system f a u l t or operator intervention. A s i m p l i f i e d auto s t a r t sequence i s of the form: END IF I n i t i a l i z e • Max_Sparks? =* Exit set Button Mask enable RF Fault or button pushed? Exit Pulse at 5% • set amplitude to 60% of target RF not detected? Exit • wait 2 seconds 58 Fault? => Exit • go CW Fault? =» Exit • wait 1 second Fault or button pushed? => Exit • Close Amplitude Loop Fault? =* Exit • wait 1 second • Close Phase Loop • ramp to target voltage Fault or button pushed? =* Exit • wait for beam Fault or button pushed? => Exit • match cavity freq. Fault or button pushed? =» Exit • go driven Fault or button pushed? => Exit • set Button_Mask • Spark_Count =0 Exit The auto s t a r t routine provides: - a means for the operator to turn on the RF system without special knowledge of the system. - a means to automatically recover from known f a u l t s and to re-establish operation of the separator. The present auto st a r t routine performs s a t i s f a c t o r i l y but i t i s 59 not well structured. During the wait sequences, the computer scans f o r f a u l t s and can i n i t i a l i z e the system i f a fa u l t i s detected. The code associated with t h i s routine needs work. In general, the perceived performance of the RF co n t r o l l e r rests with the software and the operator interface i t provides. For example, i f there i s low voltage from the screen g r i d power supply, i t i s reported as a con t r o l l e r f a u l t ; the system does not come up to voltage when the on button i s pushed. A successful operator interface requires more diagnostics and as much development as the does the regulator hardware. The TRIUMF Controls Group i s working to develop workstations for supervising s i t e processes. Such a system i s used at Los Alamos to supervise RF regulator loops and provide the user interface. This configuration w i l l improve the RF con t r o l l e r . 60 Conclusions A control system for the TRIUMF M9 Separator has been modeled, designed, and b u i l t . Experience with t h i s and other RF systems indicates that a second order model i s s u f f i c i e n t to control most, i f not a l l , of the TRIUMF RF systems. It i s not necessary to know the exact pole locations or the system gain. The regulator zeros can be adjusted over a range s u f f i c i e n t to cancel system poles introduced by the RF amplifiers and by copper c a v i t i e s . Within an order of magnitude, the product of the plant gain and the measurement gain i s 1 for the i n s t a l l a t i o n s at TRIUMF. The regulator design i s unity gain stable and the plant loop can be made stable given the 48db of gain adjustment i n the regulator. S i g n i f i c a n t cross coupling between the open loop phase and amplitude controls i s introduced when the plate c i r c u i t i n the power amplifier i s detuned. This can be almost eliminated i f the cav i t y i s tuned to the complex conjugate of the plate c i r c u i t . In systems where the transmission l i n e i s an integral number of wavelengths, a conjugate tuning scheme should present a r e s i s t i v e load to the tube. The present tuning system keeps the average ca v i t y tuning within 5° of resonance. The disturbance spectrum can sometimes exceed the bandwidth of t h i s loop, causing fluctuations i n r e f l e c t e d power greater than 1% of the forward power. 61 References 1) P. Sigg, A General RF Control System Concept TRIUMF Design Note TRI-DN-85-27 2) J. Cherix, RF Phase Detectors TRIUMF Design Note TRI-DN-86-17 3) J. Cherix, RF Control System Summary TRIUMF Design Note TRI-DN-86-18 4) L. Durieu, private communication 5) F. Pedersen, Beam Loading Effects i n the Cera Booster IEEE NS-22, June 1975, pl906 6) R. Burge, Control Options f o r Kaon Factory Beam Loading Los Alamos AHF Accelerator Workshop Proceedings February 1988, p298-307 7) R. Haussler, Modelluntersuchungen fur das Regelsystem der 150 MHz-Flattop-Anlage Swiss Nuclear Institute (SIN) Design Note TM-04-33, November 1974 8) S. Koscielniak, A General Theory of Beam Loading TRIUMF Design Note TRI-DN-89-K25 9) E. Blackmore et a l , An RF Separator for Cloud Muons at TRIUMF Nuclear Instruments and Methods, A235(1985) p235-243 10) T. Enegren, R. Burge, D. Dohan A Modular RF Control System at TRIUMF IEEE 1987 P a r t i c l e Accelerator Conference Proceedings, p532 11) T. Enegren, L. Durieu, D. Michelson, R. Worsham Development of a Flat-Topped Voltage for TRIUMF IEEE Transactions on Nuclear Science 1985, NS-32 62 12) D. Boussard, Control of Cavities with High Beam Loading IEEE Transactions on Nuclear Science 1985, NS-32 13) F. Pedersen, A Novel RF Cavity Feedback Tuning Scheme for Heavy Beam Loading IEEE Transactions on Nuclear Science 1985, NS-32 63 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0302158/manifest

Comment

Related Items