113th ATF2 weekly meeting

International phone/webex meeting

Status and plans of the work on the stripline BPMs and the progress with orbit reconstruction, Yves Renier (LAL)

First, Y.Renier reported recent kicker noise studies on MQF1X, MQD2X and MDF9X. Former two BPMs are short with large aperture and the third one was long with small aperture but previously mismatched to the clipping module's polarity. Pulsing the kicker, signals were observed with and without the 25MHz high pass (HA) filters by oscilloscope. The signals were also analyzed by FFT. The kicker noises have three major frequencies at 1.6MHz, 3.9MHz and 6.7MHz . The electronics consists of head amp and clipping module. Typical kicker noise was a few mV at input of head amp and amplified to 20mV by it. The HP filter cut off all the noise. The most effective place is at the output of the head amp, where no noise remain. I would like to monitor the amplification stability for one week. I will adjust the ADC gates and the calibration.

Q : Why were the kicker noises so small ?

A : The noise is generated by switching. While the switching noise is very high, it is dumped after transformation in 30m long cable. But, the lower frequency part must remain.

Q : What is the switching frequency ?

A : The switch time is about 40ns.

Q : How many do we need the HP filters ?

A : The total number should be 48 since there are 12 stripline BPMs.

C : We already purchased 30.

Q : Does the filter improve the DR BPMs near the septum magnets ?

A : Yes, in principle. We had better check the frequency range of the DR BPMs since the filter has a limit of highest frequency too.

C : Since there are many power lines, they could create potentials for variation of ground line.

A : The HP filter can cut off DC level, too. However, noise could come from the ground line.

C : The calibration should be done during check run (i.e. 2nd week in October) after the ADC gate tuning.

Next, he reported progress/updates in the reconstruction of electronic noise and beam fluctuations using strip-line and cavity BPMs. Seven parameters of x(IEX), x'(IEX), y(IEX), y'(IEX), ΔE/E(IEX), x'(KEX2) and y'(KEX2) were fitted (reconstructed) by least-square minimization of BPM readings with the current lattice, where R_i,j is a transfer matrix element from the parameter of j to BPM of i and j=1 to 7, i=1 to total number of BPMs except for sextupoles BPMs, MQF4X, MQF9X, MQM15FF and MQF17X for strange behavior. IEX and KEX2 are the starting point of extraction and the second kicker position, respectively. The data consist of 500 orbit measurements after BBA with corrections of dispersion and coupling during 14th May,2009. The beam energy was changed at 10 times by the DR RF-cavity to calibrate the energy dependence. The energy changes were well reconstructed.

Distribution of reconstructed parameters without the energy changes were shown with RMS's which are listed in a table below.

After the fitting, scale factors were introduced for each BPM for linear relation between calculated and measured BPM readings. Several BPMs have large scale factors of greater than 3 and some has negative scales, i.e. wrong sign. Variation of BPM readings with respect to calculated ones is considered as the BPM error subtracted beam jitter. Taking account of scale factor in addition, we could estimate "correct" BPM error. These errors were shown in x and y for each BPM. The corrected error, i.e. BPM resolution are estimated to be 2 to 10 μm and 0.5 to 1.5μm in x and y, respectively.

In future, we would like to include ring BPMs and to have one pass determination of scale factors and parameters.

C : Since QBPMs have been calibrated on 12th May, i.e. 2 days before the above data taking, we can not understand the large scale factors. The calibration constants were stable as described in next talk.

Status of the Cavity BPM system, Stewart Boogert (Royal Holloway, University of London)

S.Boogert reported a short summary of calibration run in this spring and plans for his stay in this September and an operation methodology in coming beam study.

There has been 14 shifts for checkout and calibration in February through May, 2009. The phase jumping problem due to the trigger race has been fixed by using a system trigger via locking box 100MHz signal. After this trigger fix, the phase ( of IQ rotation) becomes very stable except for a small jump. We will monitor for about one month before next beam operation. 31 c-band BPMs have been calibrated in May, where 22 of them have movers. Non-mover calibration still has a problem.

The calibration consists of following 4 steps; (1) move BPM for ±200μm, (2) recode I and Q signals, (3) determine I-Q rotation so that all variation is the rotated-I, i.e. I' and (4) determine scale for position information. I and Q are calculated by the digital down conversion (DDC). They depend on the start trigger, the beam arrival and the DC sample point which is adjusted for saturation levels. Therefore, calibration constants could be changed by these conditions. The software should be aware the changes and compensate in the code.

The c-band BPM system performance was summarized. The calibration of all 22 QBPMs was done with 5 steps for 400μm range on movers . The position resolution was 5 μm together with beam jitter to be subtracted. Also, long range calibration was tested for ±1mm with 10 steps. BPM response was linear, while movers have difficulty occasionally such as residual roll after x or y movement.

The calibration parameters of QM16FF-y in 3 runs ( 3 weeks ) are listed in a following table to see the stability.

There were changes in IQ rotation and scale. Some differences are due to sample point, while other differences are as yet inexplicable. We will look at more BPMs.

He is setting up new computing environment to improve data quality and operations. Also debugging environment is prepared for offline debug. Welcome new contribution since present manpower (SB and AA) is limited. Especially, manpower is needed for operator/code interface for non experts users.

Finally, the operation methodology was shown. Monday: stop tone calibration system; Tuesday (to be removed once the system is stable): calibrations w and w/o movers; Friday: start tone calibration system; Sunday: Automatic copy, archive data, calibration data, weekly data to RHUL/SLAC/where-ever .

Q : What is the problem in the non-mover calibration?

A : ???

Q : You concluded that the BPM resolution was a few μm. It is difficult to see the stability of calibration constants. We experienced to need calibrations every time in check runs in this June for the fast kicker studies.

A : As I(SB) explained, the calibration depends on the trigger start, beam arrival or saturation levels . If we change such parameters, the constants are also changed. However, they have been very stable for one week at least in May runs as you can see the plot after trigger fix in page 8.

Q : At present, all the electronics have 20dB attenuators in front. Removing them, how much is the dynamic range ?

A : It is still ±1mm, since the (exponential) tail part can be used even if signals are saturated.

Q : Do you understand why IQ rotation phase changed in the above table ? It seems to be large.

A : I(SB) checked consistency between scales with corrections by variations of phase and sample points. So, these changes are correlated. However, we do not understand the small jump observed in the beam run, May. I will check the data in details, then I will report results in next meeting.

Q : Are S-band BPMs available for next run in October ?

A : Yes, probably. We need manpower. Welcome your contribution.