A major purpose of the Techni-
cal lilformation Center is to provide
the broadest dissemination possi-
ble of information contained in
DOE’S Research and Development
Reports to bu=iness, industry, the
academic cor,.
mity, and federal,
state and local governments.
Although a small portion of this
report is not reproducible, it is
being made available to expedite
the availability of information on the
research discussed herein.
LA-UR -83-755
1
l,?,-lJR--(I3-755
13Es13 009973
TITL~:
PROTOTYPE PHASE AND AMPLITUDE FEEDBACK CONTPOI. SYSTEMS
FOR THE FMIT ACCELERATOR
AUTHOR(S):
M. V. Fazio and R. D. Patton
p(:~l #>{
,.............=-
NOHCE’”’ “.’”
?OR’TIOW8 OF lltt8
REPOttY ARE IUEMLE:
It
has baan ra~roducad frcm tha best
●allabla copy
to permit the broadest
msslbla avallabllitg.
SUBMITTEDTO
1983 Particle Accelerator Conferenc~,
March 21-23, Santa Fe, NM
I)Is(”I,AIMIW
‘1Ili.f rclmrl ww+ pIcpHIal W ml mxrmnl M work h[mn~wml hy mr Hrncm.y w[ the 1JnilrAl Stntm
( iuverrrulcnl.
NclllicrI% 1Inntxl SIiI\Cs ( iovcrrwncnl nm nn} 1IECIK%Ilwmlf, wor wry or Ihcir
CIUpklVCCS, IWHkCnN y w nrr[mly, cxrwm% IM wlilll~l.
.Ir Ilmntilllcs Imy ICBIII Iinhility wr rcapmwi.
I,ili[y r,w IIIC Ilccur:lcy, corrlplc.lcnchs, m dldllr~(If iw inhnnmlirm. qrpnrHlun, frrrrrlucl, or
fwtwrm dim%wel, JI rcplrwnts Ihm ils uw
WIIUIIIIWII
mrrmnc privulcly ownctl righls. Ncrcr-
CIWC hcrein III Uny Sfmclf’1.’ CWIIIIICWIIII p! IahlI.1, p in’c~%
w scrvicc hy
Irmlcnumc, lr~~cm$r~l
Illflnufwdurrl,
UI ,IIIICI wire’ dImb nor nrscw+urily constiltllc w imply ils CIHlll”Wlllcllll rcwm-
,Ilcn(ltilll)ll,,)r~H\t)r@ IV
Ihr I Iuiird SIAM (iovcrwwleol w IIIIY nmcnL’Yllmlalr.IIlcVlcwn
LIOII
~jplnitnN III ,Iulhtw$ chplc**l
hrlcin
AI IrIll Ilwcsx:ll Ily slwh’ or Icflcd III(IW d IIIC
●
[ Inilcd SInlm ( hwrl wmrnl III IUIy Iqwm”) lhclcIIl
ILOSA[
annl(xs
Los Alamos National Laboratory
Los Alamos,New Mexico 57545
““’’’’’wmwak ~“”
t
II 1!11:;IWIW’IW l“;Itwl’111111t 1’ )
P?dTOTYPE PHASE AND AMPLITUDE FEEDBACK-CONTROL SYSTEMS FOR THE FMIT ACCELERATOR*
M. !.
Fazio and R. D. Patton, AT-5, MS H877
Los Alamos National Laboratory, Los Alamos, NM 87545
W!!!?Q!
The phase and ampl ’tude feedback-control systems
for the Fusion Materials Irradiation Test (FMIT)
accelerator” have been successfully prototype and
tested,
The testing was performed at low power with
two 1OO-W rf sytems driving a high-q resonant cavity
at 80 MHz. The control systems can maintain the cav-
ity field amplitude to within *1% and the pha~e to
w~thin 21°
of the set-point values.
When there are
multiple rf systems independently driving a resonant.
cavity through
individual drive loops, amplitude
matching and proper phasing between the outputs of
ewh rf system are essential for proper system opera-
tion. Experimental results are presented.
Introduction
For
nmnal
operation,
the FMIT accelt?rator
requires, simultaneously, as many as six high-power rf
systems tc drive a single Alvarez structure
at full
beam current.
Each ampliflar is designed to produce
600 kW, cw at 80 MHz.l All of these amplifiers must
be phase and amplitude controlled to withfn tl” and
21%, respectively, of the set-point values for proper
accelerator operation,
These tolerances are maintained
with analog feedback-control systems.
The complete
analog feedback-control system and low-power rf system
was tested in the configuration depicted in Fiq. 1.
An experiment was breadboarded in the lahnratory
to test viablllty of the multiple rf-drive concept.
The experiment consisted of two rf amplifie, chains,
each with 1OO-W cutput, driving a resonant load with a
~(1 of
~ooo.
The resonant load was a coaxial, half-
w~ve,en!lth, capacitively end-loaded cavity.
The rf system Is driven by a frequency synthe-
sizer at 80 MHz.
The synthesizer is followed by t,hc
cavity phase shifter module.
Beyond this polrt, tt,e
rf siqnal
IS split and distributed to two identir,ll
rf ~mplifier chains.
Chail 1 is designated th[ master
chdin,
#nri Chain ? is the :slave.
,..! ,,, .
i“”;’’’:!”’”
1.
“1
I
,,,,,,
,,,.
.!
-/.’:::,1
.,,.!, ,,, ,.,
,,,,, ,,.
“,,., ,,’
~
1!1 [“’’”-- !!
1:
1,J{,,.;~
I
I
.’h
,,.,..,1.1.:(;:?.]
.,
.1,
!, .,....
I
I “
-1 r
1 ! !,, ,.
~ ii ‘“’’”’--
l::~~id
II I
.,,
,
,!.,.,,, ,,,
The cavity and the chain phase-shifter modules
are alike. The phase shifters are of the varactor-
tuned circulator type.’
The actual phase shifting
is done at 400 MHz because of th~ unavailability of
circulators at 80 MHz and the physical size of tl,e
strip-line circuitry at that frequency. Each phase-
chifter Dackage includes an active up-converter to mix
the 80-MHz input with a 320-MHz source to reach the
400 MHz required by the circulator/varactor “t of
the circuit, which is followed by an activ(
down-
converter to get back to 80 MHz. Phase information is
preserved by using the same 32(1-MHz signal for both Iip
and down conversions. The phase-shifter module includes
an automatic-level control (ALC) amplifier whose o~t-
put is held constant to compensate for the phase-
shffter’c insertion-loss variation with phase shift.
The phase shifter
is followed by a 30-dB, 100-mW
drive-level control (DLC) amplifier whose output power
depends on a dc control voltage. Because of unavail.
ability of 80-MHz isolators,
d 10-dB pad is included
between the ALC and DLC amplifiers to provide some
isolation between stages.
The final stage is a 1OO-W
rf amplifier, loop coupled into the cavity resonator.
In the FMIT system, the 1OO-W amplifler will be fol-
lowed hy a three-stage, 600-kk linear amplifier
with an EIMAC 0973 final.
Control l’hiloso~
—————.
The basic control philosophy accomplishes a dutil
purpose. The first is tn rriaintairlthe cauily field’s
required Fh~se and amplitude.
Th, second is fl) to
estah!ish zer~ phase shift brtwer,l both I lain’s rf
outputs and (7) to maintain the slave chain’s power
olltput equal to that of the master chain.
Phase and
amplitude tracking hetwe~n the two rf chains is impor-
tant tn achiev(, qood overall syst~~m strrhilit,y, It
will become ~ven nlor~ important in thp case of I“MII
with six 600-kW chains.
I
! , . .
.,,,....
!,.!.
,, , ,,,
h!
To achieve phase tracking, each chain has an
automatic phase-control loop that maini.sins a constant
phase shift across the chain regardless of operatinq
conditions.
The chain phasu detectors are double-
balanced mixers.
The cavity phase detector is a
device designed by Los Alamos that has a linear output
over a 360° range and a 100-kHz bandwidth.3
Amplitude trackinq is accomplished by making the
slave chain match the master chain in output power by
using an amplitude controller.
The cavity phase and
cavity amplitude control 10CJPS have d 20-kHz bdnd-
width, determined primarily by the cavity bandwidth.
The automdtic phase control and the slave amplitude
control loops were Sesiqned for a 100-kHz bandwidth.
The basic configuration for the feedback control-
lers <s shown in
Fig. 2.
It consists of an error
amplifier; proportional, integral,
and derivative
stages; and an output driver.
All the controllers are
basically alike,
although the gains In the various
stages depend on the specific application and typi-
cally are not the same from unit to unit.
the
(1)
(2)
(3)
(d)
Control System Operation
In normal operation of the rf amplifier chains,
following sequence occurs:
An “RF Enable” signal to the cavity amplitude
controller tells Chain 1 to turn on.
Inputs to
this controller are a detected cavity feedback
signal and a set point corresponding to the
desired cavity rf field amplitude.
As Chain 1 turns on, Chain 2 tracks it in output
power. The 11O-DS response time of Chain 7
is much faster than the cavity’s 50-IJS rise
time; thus,
both ri systems essentially are
rfrivinq the cavity sim~ltaneously.
The aut~.matic phase-control loop around each
chain holds Lbe phase shift across each chain to
a constant value, regardless of operatinq cnnrli-
tic)ns, This assures that both amplifiers are
dr!vinq the cavity in phase.
As soo’n as tne cavity amplitude-loop locks, the
Lavity phase controller locks dnd maintains the
cavity field’s proper ph,lsc.
Fy]erimental Tests
— — ...—. .__-—-—.
Sevrt-nl tests were ,lerformed on tllc compl:t~ s,ys-
tmn tn assure proll~r system perforrniincc over a wi[lIY
ran!]e of system rfistllrhances.
Step dlsturharrcs ws+ro
introduced through the “External Demand” inpul.s to thr
various controllers.
The follnwinq tests arc snme of thn<p success-
flllly prrformcrl:
(1) Chain 1 output wnz varieii from O to 11307 in r cw
mn[ir. I’hate an[l nmplit.udc track lllq o! ~l;aill ;’
w,as wll,hln t,ll~I 1“/1 l% tnler,lnrrm
[11,
I
/ ,.,~,,.~,
II
,,
I
.,,
””l !,, I
,,,,,,
-(;.
‘lb
;,,
‘ ‘ ,,,,,
hm,
I
.!:;,)
i
-,,,”, ”!1,
’1’
,“!!. !,”,,!-.,
,,,, )..!,
I
“08”,”
,.,,. ”,,
m❑ .”,,
,,
jl:~ll
,,,”,,. 11.,
(2)
With both chains operatinfj at half-maximum power,
an External Demand was applied to the slave
amplitude controller of sufficient level to drive
it from 10 to 90% of fu-ll output if the control
loop were open.
The controller held the cavity
amplitude constant to *1%.
Figure
3a shows
the 10 to 90% External Oemand input,
Figures do
and 3C show the slave error and cavity (mister)
error signals.
Figure 3d is the vertically
expanded cavity rf field that is being controlled
to t2.8% on the ini’:ial under- and overshoot
and to considerably
<1% after the transient
has damped away.
Figu.es +e and 3f shnw the tops
of the master and
Slilve rf output enve!opes cm
an expanded time and voltage scale.
(a)
(b)
(c)
(d)
(0)
(r)
I Iq. .1, (a) lxl PIII,IlJcmaml 011
flavo 1:111111.1111111.(;’ V,
?() lls/dl v).
(1)) flair ,Impllflldr P1’1’llr<I,il,ll
(1111)
mv,
;’[)
11.,/dlv).
(() [:flvlly alllplIf.11111,
rrl’ot’
\lllllal (loll Il!v, ;’(I ll,t/(llv).
(d)
lrll~ll, y
rf tlrlll (II)mV, ?II1,./Iilv). (P) Vwli,ally
Pxpallllod
rf l’l!vl~llji)llof Ills.,!ot’Ihalll.
11111
rllvoll)pl I.
;Iv l}_p,
(?{l~~mV. 1~1l,\/dlv),
(f) V! ’t-111’tllly ~xpllnllf~ll rf PIIVIIII)IW l)f \lnvll
I“ilnlfl.
11111
I’llwllllw 1< ?V I’-l’.
(;’1)11Inv,
10 11%/dlv),
(3) ‘An External Demand was applied ‘o the automatic
#
phase-control loop sufficient to drive the phase
of the chain +90°.
if It were open
loop.
The
controller miiintained the looP phase shift across
the chain to less than *ln.
(4) Ulth the cavity field amplitude set so that each
amplifier was delivering one-third full power,
the cavity-amplitude f?edback signal was modu-
lated to simulate a 200% beam loadin~ rwherc
Per cent beam loading =
‘pbeam’pco90er
j
x-loo].
The cavltyfleld amplitude was maintained at Ii%.
See Flf 4 for trdaslent response.
(5) With the cavltyphtise set to the set-point value,
a f90° phase disturbance was
applied through
External Demand on the ~avity phase controller.
Tank phase was held to *1”.
See Fig. 5 for
transient response.
In gaining experience with the multiple-drive
configuration,
it became obvious that very tight
tracktng was required between the output
Dhase and
amplitude of each chain to maintain a low VSWR in each
output drive line. Oetailed measurements have not
been made yet, but as a general rule of thumh a phase
difference between chains >lf)J and an amplitude
difference >10% led to drive-line VSURS and cavtty
field instabilities that would not be acceptable in an
operating accelerator system.
In an operating accel-
erator, final amplifier gain depends strongly on load
VSWR; thus, every possible effort must be made to get
the VSWR to behave in a controlled manner during tran-
sient conditions. Because OUY
system successfully
held phase and amplitude differences to less than
tl” and 11% l~etween cha’ns, at this time detailed
measurements were not made on the effects of large
phase and ampl~tude tracking errors between chains.
Another control-system conflquration had been
tried previously, where
instead of a master/slave
arrangement, each rf cl,ln had an automatic gain con-
trol (AGC) around It to Keep the gain constant under
a variety of operating conditions and system rlisturh-
antes.’
It proved impossible to develop an AGC
loop with enough dynamic range to maintain a constant
qain over the rf output range of 20 to 100% of full
power.
Because of the lack of success in tr,vinq to
stahi
first
ize this configuration, we opted to go with t.hc
conflquration discussed.
rlq, 4. VPrtlcnlly ~xpand~(l clfvil,v rlrld wllh slmu-
Intod ?00% beam Iondinq. lull flllllal l\
;]4[1mV P-P.
(l(~mv, 1oo”ll~/(llV)m
6iims!m
Fig. 5. Cavity phase error 50 mV/deg.
(50 mV,
200 ps/div).
Conclusion
When multiple rf sources are required Tor driving
single rf accelerator cavity, very tlgh~ phase and
amplitude tracking is needed between the source’s out-
puts . The master/slave configuration for amplitude
c~fltrol,
coupled with automatic phasecontrol loops
arouno each chain, proved to be a very stable design
In the laboratory, mair,taining the cavity field to
within tl” in phase ard *1% in amplitude, in spit? of
a variety of imposed system disturbances.
Based orI
these results,
a similar control philosophy Is being
implemented on the FMIT accelerator and currently i,
under test at hiqh Power.
1.
2,
-i.
~eferences
M. Fazto, H. Johnson, and D. Rlggin, “Develop-
ments of the RF System for the Fusion Materials
lrraciiation lest Accelerator, ” Proc. 1979 Lifiear
Accelerator Conf., Montauk, New York, September
10-14,
1979,
Brookhaven National
Laboratory
report IINL-S1134, p. 356 (19f10).
R. F, Nylander, M.
V. Faz\o, F. Dacc!, and J. Il.
Rogers,
Ilperfomancc Tests of the ~~o-kw cw~
nn+liz,
Radio-Frerplency Systems for the TMIT
Accelerator,”
!.OS Alamos National LaboratorY
document 1.AllR-~3-753 (1903).
Donald W. RPid, Dennis Riq!lin, M{chael V. ta7io,
Rnhert D. Patton, nrwl IlarolrlA. Jackson, “A 3110”
Iliqital Phass! Iletectnr with 1(:[1-kl!~~!andwidth,i
Proc, 19nl P~rticle Accelerator Cmf., W(lshinq
ton, DC, March 11-13, 19/1;, 111”1 Trans. N,irl
sci. ?nJ p. 21511 ‘1111).
.-.