May 2021: Test Operation and First Real Bunch-to-Bucket SIS18 -> ESR

Table of Contents

Introduction

The bunch-to-bucket transfer system has been used for almost one week in operation. From 11-16 May the focus has been on beam operation to experiments. The planned bunch-to-bucket experiment took place on 17 May.

Setup

See here.

B2B v00.02.40, branch b2b_dietrich_2020-oct-26, commit 4de28554

Log

date	Olog time	what	remark
2021-05-11	08:30	SIS18 and ESR kicker trigger changed from 'Timing Generator' to b2b system
	09:00	first beam extracted from SIS18 with b2b system
	15:00	HKR: can't trigger BI equipment properly, see olog	see here
		the problem is, that B2B triggers extraction ~1ms after EVT_KICK_START, but the integration window of BI ends ~500us after EVT_KICK_START
	16:30	solution: apply correction of -600 us for both extraction (SIS18) and injection (ESR) kickers at the b2b system
til 2021-05-17	08:00	6 days of routine operation with krypton beam SIS18 -> ESR (bunch to coasting beam), experiment at ESR	no issues, users are happy
til 2021-05-18	08:00	7 days of routine operation with xenon beam to HHT (fast extraction) experiment with PHELIX	no issues, users are happy
		first synchronization of a plasma physics experiment with an extracted bunch to 1 ns
2021-05-17	14:30	setup ESR with xenon beam; begin injection as coasting beam	Sequence ID 1, 124Xe46+ @SIS18, stripper foil, 124Xe54+ @ESR
	16:00	switch to mode buncht-to-bucket; but SIS18 has h=1 and ESR h=2; this does not fit!
	16:35	no beam in ESR
	17:00	one of the injection kicker modules at ESR failed, work with remaining two modules at higher energy
	17:20	SIS18 changed to h=2 and ESR to h=1 (a bit dirty)
		set revolution frequency in SIS18: cavity DDS 1979732.979 Hz (506.118 ns)
		measured revolution frequency in ESR: cavity DDS 1975118.183 Hz (506.3 ns), T_beat 216.7 us
	17:45	begin bunch-to-bucket test '
	18:45	end bunch-to-bucket test (experimentalists urgently need to enter ESR)

Table: Log of beam time. Hours are tentative.

Routine Operation

The bunch-to-bucket transfer system has been used for routine operation serving two different experiment for about 6 days.

• krypton beam SIS18 -> ESR (as coasting beam)
• xenon beam with fast extraction to cave HHT (PHELIX)

The main caveat has been triggering of beam instrumentation devices in the transfer lines. Despite all the efforts in retrofitting and modernization, the triggering of a large amount of diagnostic devices along the transfer lines is still based on the MIL event bus.

• it is not straight forward to trigger on event numbers other than EVT_KICK_START1/2
• it is not possible to do pre- or post-triggering

As a workaround, an ugly trick has been applied within the bunch-to-bucket transfer system to 'fix' the problem temporarily.

As the bunch-to-bucket system is not yet integrated in to the control-system. While the kicker timing could be set via the ParamModi parameter 'Kicker Offset', the revolution frequency of the ring must be set 'by hand' using a CLI of the b2b system. This was only done while setting up the experiments on 11 May 2021. After that, the b2b transfer system worked without any further intervention until the end of the experiments on 17 May (ESR) and 18 May (HHT).

Measurements with Bunch-to-Bucket Transfer

UTC	phase difference [ns]	phase difference [degree]	cavity gap	remark
16:00:23.693	272	193.4	650
16:00:55.989	147	104.5	650
16:01:28.280	147	104.5	650
16:02:00.568	147	104.5	650
16:02:32.861	397	282.3	650
16:03:05.150	357	253.8	650
16:03:37.439	317	225.4	650
16:04:09.728	437	310.7	650
16:04:42.030	437	310.7	650
16:05:14.328	417	296.5	650
16:05:46.628	417	296.5	650
16:06:18.319	397	282.3	650
16:06:50.617	397	282.3	650
16:07:22.921	387	275.2	650
16:07:55.233	387	275.2	650
16:08:27.373	387	275.2	650
16:08:58.921	387	275.2	650
16:09:31.075	387	275.2	650
16:10:03.203	387	275.2	650
16:10:35.335	640	95.1	650
16:11:07.464	640	95.1	650
16:11:39.614	640	95.1	650
16:12:11.738	630	88.0	650
16:12:43.872	630	88.0	650
16:13:15.444	620	80.8	650
16:13:47.576	650	102.2	650
16:14:19.704	660	109.3	650	phase max destructive
16:14:51.830	660	109.3	650
16:15:23.993	660	109.3	650
16:15:55.572	407	289.4	650	phase optimal (destructive – 180 degree)
16:16:27.752	407	289.4	650
16:17:00.062	407	289.4	650
16:17:32.237	407	289.4	650
16:18:03.945	660	109.3	650
16:18:36.085	407	289.4	650
16:19:08.202	407	289.4	650
16:19:40.350	407	289.4	650
16:20:12.480	507	0.5	650
16:20:44.612	507	0.5	650
16:21:16.745	307	218.3	650
16:21:48.881	307	218.3	650
16:22:21.017	407	289.4	650
16:22:53.175	407	289.4	650
16:23:25.302	407	289.4	650
16:23:57.453	407	289.4	650
16:24:29.592	407	289.4	650
16:25:01.725	407	289.4	650
16:25:33.865	154	109.5	650
16:26:06.008	154	109.5	650
16:26:38.152	154	109.5	650
16:27:10.319	154	109.5	650
16:27:42.470	407	289.4	650
16:28:14.632	407	289.4	650
16:28:46.781	407	289.4	650
16:29:18.931	407	289.4	650
16:29:51.082	407	289.4	650
16:30:23.233	407	289.4	650
16:30:55.372	407	289.4	650
16:31:27.531	407	289.4	650
16:31:59.677	407	289.4	650
16:32:31.844	407	289.4	650	changing gap voltage @optimal phase
16:33:03.408	407	289.4	163
16:33:35.553	407	289.4	163
16:34:07.714	407	289.4	163
16:34:39.855	407	289.4	163
16:35:11.998	407	289.4	163
16:35:44.145	407	289.4	163
16:36:16.310	407	289.4	163
16:36:48.449	407	289.4	163
16:37:20.584	407	289.4	163
16:37:52.719	407	289.4	163
16:38:24.878	407	289.4	163
16:38:57.012	407	289.4	1281
16:39:28.570	407	289.4	1281
16:40:00.732	407	289.4	1281
16:40:32.894	407	289.4	1281
16:41:05.036	407	289.4	1281
16:41:37.203	407	289.4	320
16:42:09.351	407	289.4	320
16:42:41.493	407	289.4	320
16:43:13.636	154	109.5	320	phase destructive (optimal – 180 degree)
16:43:45.782	154	109.5	320
16:44:17.928	154	109.5	320
16:44:50.088	407	289.4	320	phase optimal again
16:45:22.223	407	289.4	320
16:45:54.382	407	289.4	320
16:46:26.525	407	289.4	320
16:46:58.665	407	289.4	320
16:47:30.798	407	289.4	320
16:48:02.932	407	289.4	320

Table Data. Shown are time of measurement as UTC (first column), phase difference of the two h=1 group DDS in nanoseconds (second column, revolution frequency ESR is 506.3 ns), phase difference in degree (third column) and rf-cavity gap voltage (fourth column).

%IMAGEGALLERY{size="medium" maxwidth="1000" maxheight="800" sort="name" include="images*.*"}%
Figures: 'Waterfall' images sorted by time. The images have been taken during approximately one hour of measurement time. Numbers in the figure denote time of EVT_KICK_START1 as UTC, phase difference between h=1 Group DDS of SIS18 and ESR in degree and rf-gap-voltage in Volts. Each images shows the signal from FCT GE02DT1FP: Bunch from SIS18 captured in an ESR bucket. x-axis: bunch time [us] relative to h=1 (1.975 MHz), y-axis: time [ms] after acquisition start from bottom to top. Shown are data of ~ 3 ms after acquisition start. Data acquisition, processing and image are done by our colleagues from beam instrumentation.

RF-Phase Difference

Injection as Function of Phase Difference

Figure: 218 degree phase difference.

Figure: 225 degree phase difference.

Figure: 275 degree phase difference.

Figure: 289 degree phase difference, optimum phase difference (see below).

Figure: 297 degree phase difference.

Figure: 310 degree phase difference.

Figure: 361 degree phase difference.

Determining the Best Phase Difference

Basically, we are novices here. As we tried to determine the best phase, the rf-amplitude was probably to large and it looks like the observation was hampered by quadrupole osciallations. As an alternative we tried to define the best phase, by first finding the worst (= most destructive phase) and then simply add 180 degree. All of the following measurements have been done with identical gap voltage of 650V in the ESR (ferrit cavity).

Figure: 88 degree phase difference.

Figure: 95 degree phase difference.

Figure: 102 degree phase difference.

Figure: 109 degree phase difference. Looks symmetrically worst!

It was decided that 109 degree looks most symmetrical. Thus, the best phase difference is expected at 109 + 180 = 289 degree.

Figure: 289 degree phase difference. This is supposed to be the best phase difference for the actual conditions.

Idea: Finding the best conditions involves multiple measurements like the one above but for different rf-gap-voltages.

RF Gap Voltage

It looks like an important parameter is a good value of the RF Gap Voltage. I (db) am not an accelerator physicist, but my naive view is that the 'trapping potential' of the rf-buckets in the injection ring shall match the one of the extraction ring when the bunches are transferred. All of the following measurements have been done with identical rf-phase difference of 289 degree.

Figure: 163V gap voltage.

Figure: 320V gap voltage.

Figure: 650V gap voltage.

Figure: 1281V gap voltage.

Reproducibility

Figure: Bunch-to-Bucket transfers at identical conditions. The black horizontal line serves to guide the eye.

The figure above compare transfers that have been done with identical conditions (relative phase 289 degree, gap 650V). Here, data acquisition has been started upon EVT_KICK_START1 + 1ms. However, the time of transfer may vary within a window defined by the beating method. At the three rightmost images, the beam is injected into ESR approximately 0.5ms after start of acquisition, while at the two left images the beam is transferred sooner. A horizontal black line is drawn ~0.5ms after data acquisition. It is interesting to note, that the outer boarders of the 'bunch shapes' almost look identical in all five images, only the 'position of highest intensity' seems to move.

Stability and Precision

Figure: On-line data by the b2b system. Stability is demonstrated by the good match of DDS signals (green boxes) and kicker timing (orange boxes). The phase difference of the DDS signals (blue box) is averaged over one hour. Details (see text). Statistical data (average, standard deviation, min, max) represents data of 96 bunch-to-bucket transfers.

The figure above shows the key parameters of the experiment for 96 buncht-to-bucket transfers.

• beating period ~216 us (~429 rf periods)
• extraction: h=2; injection h=1

The orange boxes shows the deviation from the measured kicker signals (electronics out) from the ideal value. The standard deviation is only about 1.1 ns for both SIS18 (extraction) and ESR (injection). This is about a factor of three better compared to the 'old' timing generator (see BunchBucketTestMeasurement5).

The blue box shows the phase difference between the two group DDS signals averaged over the full measurement time of about one hour. As the phase difference was the main tuning parameter and changed many times, the shown value is not suitable to estimate the precision.

The green box shows the precise matching of the two group DDS signals at extraction. The standard deviation is only ~ 0.5 ns, while the maximum deviation from the ideal value is only 2 ns.

• ext(traction): shows the deviation between measured and set-value; the set-value for kicker correction is already subtracted. In the ideal case, this should be 0 ns.
• inj(ection) : shows the deviation between measrued and set-value; the set-value fors kicker correction as well as the phase difference are already subtracted. A perfect phase match yields a value of 0 ns!

Thus, the difference between set-value and act-value for the relative phase of both h=1 groupDDS signal at transfer is only 0.5 ns @ 1 sigma. This corresponds to a precision of about (0.5 ns / 506 ns) ~ 1e-4 or 0.36 degree. Even the observed worst case deviation was only -2 ns, which corresponds to ~ 4e-3 or 1.4 degree only.

Please keep in mind, that the beating period was only 216 us and extremely short. The trick to achieve such a good precision at short beating times is to calculate the matching of both rf-signals for a couple of beating times in advance and simply pick the best match. This method presently uses a 'brute force' algorithm and limited to 5 beating periods due to available computing power. This method could be improved in the future. However, at long beating times of 10 ms as assumed in the technical concept, such tricks are not required.

Trapping Efficiency with B2B

Unfortunately, we were unable to record long (seconds) signals. Fortunately, one of the analog instruments displayed the signal strength as a function of time for the optimum and worst phase. A screenshot is given in the figure below. First, it can be observed that the bunch remains stored for many seconds in both cases. For the 'worst case', the measured signal (in Volts) is 12 db less than with the 'optimum case'. Thus the signal strength (in Volts) is a factor of four less. This should correspond to the number of stored particles. Thus, when comparing 'worst'/'optimum' phase, the number of particles stored in rf-buckets differs by a factor of four. Remark: There could be more particles in the ring, but these would be outside the RF-potential and lost upon further manipulation such as acceleration or decelaration.

Figure: Signal strength of stored beam as a function of time. Shown are two traces. Optimum phase difference of 289 degree (white) and 'most destructive' phase difference of 109 degree (white). When comparing the signal strength for long storage time of about 1s (middle) the signal of the 'worst' phase difference is attenuated by ~ 12db compared to the case, when the bunch-to-bucket transfer is done with correct phase.

Conclusion

The bunch-to-bucket transfer system has been used in standard operation for about one week without issues. To continue using beam instrumentation devices connected to MIL timing, the bunch-to-bucket system should be triggered to an event different than EVT_KICK_START1/2.

Real bunch-to-bucket transfer has been demonstrated from SIS18 into ESR for the first time with the new system. This first attempt was successful. All measurements have been done within one hour only and no quantitative measurements have been performed. The recording time of the bunch signals was only about 3.5ms which should be prolonged for future measurements.

-- DietrichBeck - 9 June 2021