Performance of a Digital LLRF Field Control System for the J ... - Cern
Performance of a Digital LLRF Field Control System for the J ... - Cern
Performance of a Digital LLRF Field Control System for the J ... - Cern
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
THP006 Proceedings <strong>of</strong> LINAC 2006, Knoxville, Tennessee USA<br />
PERFORMANCE OF A DIGITAL <strong>LLRF</strong> FIELD CONTROL SYSTEM FOR<br />
THE J-PARC LINAC<br />
S.Michizono # , S.Anami, Z.Fang, S.Yamaguchi, KEK, Tsukuba, Japan<br />
T.Kobayashi, H.Suzuki, JAEA, Tokai, Japan<br />
Abstract<br />
Twenty high-power klystrons are installed in <strong>the</strong> J-<br />
PARC linac. The requirements <strong>for</strong> <strong>the</strong> rf field stabilities<br />
are +-1% in amplitude and +-1 degree in phase during a<br />
500 μs flat-top. In order to satisfy <strong>the</strong>se requirements, we<br />
have adopted a digital feedback and feed-<strong>for</strong>ward system<br />
with FPGAs and a commercially available DSP board.<br />
The FPGAs (Virtex-II 2000) enable fast PI control <strong>for</strong> a<br />
vector sum <strong>of</strong> two cavity fields. The measured stability<br />
during an rf pulse was +-0.06% in amplitude and +-0.05<br />
degree in phase. The tuner control was successively<br />
operated by way <strong>of</strong> <strong>the</strong> DSP board by measuring <strong>the</strong><br />
phase difference between <strong>the</strong> cavity input wave and <strong>the</strong><br />
cavity field. Beam loading effects were emulated using a<br />
beam-loading test box. By proper feed-<strong>for</strong>ward, <strong>the</strong> rf<br />
stability was less than +-0.3% and +-0.15 degree.<br />
INTRODUCTION<br />
We have installed 20 high-power klystrons (324 MHz,<br />
max. 3 MW) in <strong>the</strong> J-PARC linac. These klystrons drive<br />
RFQ, DTLs and SDTLs [1]. The required stabilities <strong>of</strong> <strong>the</strong><br />
rf outputs are less than +-1% in amplitude and less than +-<br />
1 degree in phase. In order to satisfy <strong>the</strong>se requirements, a<br />
digital feedback system using a compact PCI (cPCI) has<br />
been developed. The system consists <strong>of</strong> a FPGA board, a<br />
DSP board, a mixer and down-converter, an rf&clock<br />
generator and an I/O board [2,3]. The FPGA board has<br />
two FPGAs (VirtexII2000), four 14-bit ADCs (AD6644)<br />
and four 14-bit DACs (AD9764), which works as a<br />
mezzanine card <strong>of</strong> a commercially available DSP board<br />
(Barcelona by Spectrum). The DSP board is used as a<br />
tuner control <strong>of</strong> <strong>the</strong> cavities and communication between<br />
<strong>the</strong> FPGAs and <strong>the</strong> local host [4].<br />
Intermediate-frequency (IF; 12 MHz) signals from<br />
pick-ups are directly detected by ADCs with 48 MHz<br />
sampling. Two cavity pick-ups are processed in FPGA1<br />
and <strong>the</strong> <strong>for</strong>ward powers are processed in FPGA1. Cavity<br />
signals are separated into I and Q components, and a<br />
feedback (FB) algorithm (PI-control) is operated with a<br />
vector sum <strong>of</strong> two cavities. A feed-<strong>for</strong>ward table is added<br />
in order to compensate <strong>for</strong> any transient behaviour and/or<br />
beam-loading.<br />
VECTOR SUM CONTROL<br />
A klystron drives two cavities in <strong>the</strong> case <strong>of</strong> <strong>the</strong> SDTL<br />
module, and <strong>the</strong> vector sum <strong>of</strong> two cavity signals is<br />
calculated. The operated vector sum control at a SDTL is<br />
shown in Fig. 1. The set-values are 6,000 in amplitude<br />
and 0 degree in phase. Since <strong>the</strong> FPGA handles integer<br />
___________________________________________<br />
# shinichiro.michizono@kek.jp<br />
numbers, <strong>the</strong> integer set-value is used. The proportional<br />
and integral gains are 4 and 0.01, respectively. The<br />
stability <strong>of</strong> <strong>the</strong> vector sum during <strong>the</strong> flat-top is less than<br />
+-0.06% in amplitude and +-0.05degree in phase.<br />
EXTERNAL MONITOR<br />
In order to confirm <strong>the</strong> stability <strong>of</strong> <strong>the</strong> feedback, we<br />
measure <strong>the</strong> amplitude and <strong>the</strong> phase with an external<br />
monitor. Figure 2 shows a schematic <strong>of</strong> <strong>the</strong> external<br />
monitor. The amplitudes are measured by diodes, and <strong>the</strong><br />
phases are measured with two mixers. The signals are<br />
stored in a computer by way <strong>of</strong> a commercially available<br />
FPGA board (Xilinx XtremeDSP) with two 14-bit ADCs.<br />
Data acquisitions are carried out every 1 μs. The obtained<br />
data are shown in Fig. 3. The set-values are 6,000 in<br />
amplitude and 0 degree in phase, respectively and<br />
stability is less than +-0.2% and +-0.2 degree. Since <strong>the</strong><br />
band-width <strong>of</strong> <strong>the</strong> cavity is less than 20 kHz, <strong>the</strong> fast<br />
noises in Fig. 3 are included in <strong>the</strong> monitor. It is<br />
considered that better stabilities would be accomplished<br />
in <strong>the</strong> cavities.<br />
7000<br />
6000<br />
5000<br />
4000<br />
3000<br />
2000<br />
1000<br />
(a)<br />
0<br />
0 100 200 300 400 500 600 700<br />
6100<br />
6050<br />
6000<br />
5950<br />
5900<br />
4<br />
3<br />
2<br />
1<br />
0<br />
-1<br />
-2<br />
-3<br />
-4<br />
(b)<br />
Cavity 1<br />
Vector Sum<br />
Cavity 2<br />
100 200 300 400 500 600 700<br />
(c)<br />
Cavity 1<br />
Vector Sum<br />
Cavity 2<br />
100 200 300 400 500 600 700<br />
Time [us]<br />
Figure 1: <strong>Per<strong>for</strong>mance</strong> <strong>of</strong> <strong>the</strong> vector sum FB control. (a)<br />
and (b), amplitude; (c) phase. Set-values are 6,000 in<br />
amplitude and 0 degree in phase. The proportional gain <strong>of</strong><br />
4 and integral gain <strong>of</strong> 0.01 <strong>for</strong> a 48 MHz clock<br />
accumulation are applied.<br />
574 Technology, Components, and Subsystems<br />
<strong>Control</strong> <strong>System</strong>s
Figure 2: Schematic <strong>of</strong> <strong>the</strong> external monitor system.<br />
TUNER CONTROL<br />
The phase difference between <strong>the</strong> <strong>for</strong>ward rf power and<br />
<strong>the</strong> cavity is related to cavity tuning. Tuner control is<br />
carried out by monitoring <strong>the</strong> phase difference in order to<br />
maintain <strong>the</strong> set-tuning angle. The measured phase<br />
difference is compared with <strong>the</strong> set-detuning angle. These<br />
calculations are made by one <strong>of</strong> <strong>the</strong> four DSPs on<br />
Barcelona. When <strong>the</strong> measured tuning is different from<br />
set-value, <strong>for</strong> example, by more than 1 degree, tunercontrol<br />
starts. The control continues until <strong>the</strong> difference<br />
becomes less than 0.2 degree. An example <strong>of</strong> <strong>the</strong> tuner<br />
controls is shown in Fig. 4. In this example, we changed<br />
<strong>the</strong> tuner position by about +1 mm (cavity 1 at around 1<br />
min.) and -1 mm (cavity 2 at around 4 min.), manually.<br />
The corresponding detuning was about 15 degrees. This<br />
detuning was much larger than <strong>the</strong> expected value during<br />
operation (typically around 1 degree), but <strong>the</strong> tuner<br />
position stabilized. During <strong>the</strong> tuner control, <strong>the</strong> vector<br />
sum was still stable owing to <strong>the</strong> cavity feedback control<br />
even though <strong>the</strong> amplitudes and phases <strong>of</strong> two cavities<br />
were different due to <strong>the</strong> detuning.<br />
1.005<br />
1<br />
0.995<br />
0.6<br />
0.4<br />
0.2<br />
0<br />
-0.2<br />
-0.4<br />
Cavity 1<br />
100 200 300 400 500 600 700<br />
time [us]<br />
Cavity 1<br />
+-0.5%<br />
+-0.2deg.<br />
Cavity 2<br />
Cavity 2<br />
100 200 300 400 500 600 700<br />
time [us]<br />
1.005<br />
1<br />
0.995<br />
Figure 3: Amplitude and phase stability during an rf pulse<br />
monitored with an external monitor.(a):normalized<br />
amplitude and (b) phase. The FB parameters are same to<br />
Figure 1.<br />
Technology, Components, and Subsystems<br />
<strong>Control</strong> <strong>System</strong>s<br />
Proceedings <strong>of</strong> LINAC 2006, Knoxville, Tennessee USA THP006<br />
± 0.8%<br />
1<br />
0.75<br />
0.5<br />
Cavity1<br />
Cavity2<br />
0<br />
-0.25<br />
-0.5<br />
0.25<br />
-0.75<br />
0<br />
0 2 4 6 8 10 12<br />
-1<br />
14<br />
6300<br />
Cavity2<br />
6200<br />
6100<br />
6000<br />
5900<br />
Vector Sum<br />
Cavity1<br />
5800<br />
0 2 4 6 8 10 12 14<br />
10<br />
5<br />
0<br />
-5<br />
Cavity1<br />
Vector Sum<br />
Cavity2<br />
-10<br />
0 2 4 6 8 10 12 14<br />
Time [min.]<br />
Figure 4: Tuner position, amplitudes and phases <strong>of</strong> cavity<br />
1 and 2 during <strong>the</strong> tuner control test. The tuner positions<br />
are changed +-1 mm (corresponding +-15 degree<br />
detuning) manually and <strong>the</strong>se returned in 2 minutes by<br />
tuner control.<br />
BEAM LOADING COMPENSATION<br />
In order to examine <strong>the</strong> beam-loading effects on <strong>the</strong><br />
feedback, we introduced a beam-loading test box [1]. A<br />
schematic <strong>of</strong> <strong>the</strong> beam-loading test is shown in Fig. 5. The<br />
test-box consists <strong>of</strong> phase-shifter and a mixer. The beamloaded<br />
cavity signal without FB is shown in Fig. 6. About<br />
a 15% decrease in amplitude is observed by <strong>the</strong> test-box<br />
in this case. The stabilities during feedback operation at<br />
<strong>the</strong> beginning and end <strong>of</strong> <strong>the</strong> beam pulse were around +-<br />
5% in amplitude and +-2 degree in phase because <strong>the</strong><br />
feedback algorism can not follow completely due to <strong>the</strong><br />
loop delay. By adding <strong>the</strong> proper feed-<strong>for</strong>ward<br />
synchronized with <strong>the</strong> beam pulse, <strong>the</strong> stabilities were<br />
improved drastically to +-0.3% in amplitude and +-0.15<br />
degree in phase, respectively, as shown in Fig. 7. It is<br />
confirmed that <strong>the</strong> feed-<strong>for</strong>ward is effective <strong>for</strong> a constant<br />
disturbance.<br />
I/Q out<br />
cPCI digital FB (324MHz)<br />
~0dBm<br />
Attenuator<br />
FB monitor<br />
Mixer in<br />
~0dBm<br />
trigger<br />
unit<br />
Ext.Trig.<br />
PC<br />
E<strong>the</strong>rnet<br />
XPORT<br />
RS232C<br />
Attenuator<br />
FPGA<br />
VurtexII3000<br />
beam<br />
loading<br />
test box<br />
Ext.Trig.<br />
14-bit<br />
ADC<br />
14-bit<br />
ADC<br />
External monitor<br />
Limit<br />
Amp. Klystron<br />
wave<br />
detector<br />
324 MHz<br />
cavity1<br />
Mixer<br />
Figure 5: Schematic <strong>of</strong> beam loading compensation<br />
examination.<br />
575
THP006 Proceedings <strong>of</strong> LINAC 2006, Knoxville, Tennessee USA<br />
e<br />
d<br />
u4,000<br />
p<br />
lit<br />
7,000<br />
6,000<br />
5,000<br />
3,000 m<br />
A<br />
2,000<br />
1,000<br />
0<br />
1.01<br />
1.005<br />
1<br />
0.995<br />
0.99<br />
0 100 200 300 400<br />
time [us]<br />
500 600 700 800<br />
1<br />
0.5<br />
0<br />
-0.5<br />
-1<br />
LINEARITY FOR PHASE-SCAN<br />
In commissioning <strong>the</strong> J-PARC linac, first <strong>of</strong> all, <strong>the</strong><br />
accelerated beam characteristics were compared with <strong>the</strong><br />
calculation under various rf parameters (phase-scan under<br />
a constant amplitude set-level). Thus, amplitude stability<br />
is required <strong>for</strong> <strong>the</strong> various set-phases. Since a distortion <strong>of</strong><br />
<strong>the</strong> IF signal (Mixer output) results in phase and<br />
amplitude errors, <strong>the</strong> amplitude is measured with a diode<br />
cPCI<br />
0 200 400 600 800<br />
NIM<br />
trigger<br />
unit a)<br />
Q<br />
Ext.Trig.<br />
PC<br />
100 200 300 400<br />
time [us]<br />
500 600 700 800<br />
Figure 7: Beam-loading compensation by FB with<br />
proper feed <strong>for</strong>ward. The stability is +-0.3% in<br />
amplitude and +-0.15 degree in phase during <strong>the</strong> flat<br />
top.<br />
I<br />
I/Q out<br />
(324MHz)<br />
~0dBm<br />
Attenuator<br />
Mixer in<br />
(324MHz)<br />
~0dBm<br />
XPORT<br />
E<strong>the</strong>rnet RS232C<br />
FPGA<br />
VurtexII3000<br />
Amp.<br />
Attenuator<br />
Ext.Trig.<br />
Time [us]<br />
14-bit<br />
ADC<br />
cavity simulator<br />
Cavity Amp.<br />
Kly Amp.<br />
-15%<br />
Figure 6: Effects <strong>of</strong> <strong>the</strong> beam-loading test box. Beam<br />
loading <strong>of</strong> 15% is generated by <strong>the</strong> test box. This signal<br />
is obtained at <strong>the</strong> cavity without any feedback (only<br />
feed-<strong>for</strong>ward).<br />
Figure 8: Schematic <strong>of</strong> <strong>the</strong> linearity test.<br />
-0.1%<br />
+0.1%<br />
Figure 9: Results <strong>of</strong> <strong>the</strong> amplitude errors with various<br />
phases. The stabilities in amplitude are less than +-0.15%,<br />
<strong>the</strong> comparable to <strong>the</strong> pulse-to-pulse stability.<br />
detector under various set-phases with feedback. If a<br />
distortion <strong>of</strong> <strong>the</strong> mixers and/or non-linearity <strong>of</strong> <strong>the</strong> ADCs<br />
exist, a difference in amplitude between <strong>the</strong> external<br />
monitor and <strong>the</strong> internal FB monitor should appear. A<br />
schematic <strong>of</strong> <strong>the</strong> linearity test is shown in Fig. 8. Figure 9<br />
shows <strong>the</strong> results <strong>of</strong> <strong>the</strong> phase-linearity. The obtained<br />
stability is less than +-0.15% under <strong>the</strong> various set-values.<br />
This error is <strong>the</strong> same as <strong>the</strong> pulse-to-pulse amplitude<br />
errors, and thus <strong>the</strong> linearity is satisfied during <strong>the</strong> phasescan.<br />
SUMMARY<br />
The per<strong>for</strong>mance <strong>of</strong> <strong>the</strong> digital FB system used at <strong>the</strong> J-<br />
PARC linac satisfies <strong>the</strong> requirements, and a stability <strong>of</strong><br />
less than +-0.06% in amplitude and +-0.05 degree are<br />
obtained in <strong>the</strong> internal monitor. The external monitor,<br />
where <strong>the</strong> amplitude and phase are measured by diodedetectors<br />
and mixers, show a stability <strong>of</strong> less than +-0.2%<br />
in amplitude and +-0.2 degree. The tuner control is<br />
calculated in DSP, and reaches its goal <strong>of</strong> within 0.2<br />
degree in detuning within 2 min. <strong>of</strong> operation. It was<br />
shown that <strong>the</strong> beam loading can be compensated with<br />
proper feed <strong>for</strong>ward. The linearity <strong>of</strong> <strong>the</strong> IF signals was<br />
confirmed by a comparison between external and FB<br />
monitors. The conditioning <strong>of</strong> <strong>the</strong> cavities in <strong>the</strong> J-PARC<br />
linac will start this September, and beam commissioning<br />
will start this year.<br />
REFERENCES<br />
[1] S.Yamaguchi et al., “Overview <strong>of</strong> The RF <strong>System</strong> <strong>for</strong><br />
The JAERI/KEK High-Intensity Proton Linac”,<br />
LINAC2002, Gyeongju, Aug. 2002, p.452.<br />
[2 S. Michizono et al., “ <strong>Digital</strong> Feedback Ssystem <strong>for</strong><br />
J-Parc Linac RF Source”, LINAC2004, Luebeck,<br />
Aug. 2004, p.742.<br />
[3] T. Kobayashi et al., “<strong>Per<strong>for</strong>mance</strong> <strong>of</strong> RF Reference<br />
Distribution <strong>System</strong> <strong>for</strong> <strong>the</strong> J-PARC Linac”, this<br />
conference.<br />
[4] S. Anami et al., “<strong>Control</strong> <strong>of</strong> <strong>the</strong> Low Level RF<br />
<strong>System</strong> <strong>for</strong> <strong>the</strong> J-Parc Linac”, LINAC2004, Luebeck,<br />
Aug. 2004, p.739.<br />
576 Technology, Components, and Subsystems<br />
<strong>Control</strong> <strong>System</strong>s