one possible source for this acoustic noise was the vibrations coming from the water cooling of the dipoles which is normally never turned off even when the machine is off.

this afternoon, Kevin turned off the main valve of the water cooling of half of the dipoles (the half on the FPC side), and we recorded the peak-peak value of accelerometer signal on a long trend (~20mn) but we didn't see a clear difference before and after.

I keep recording the accelerometer pk-pk signal during the night in case of one could see something different...

today, we were able to store more than 93kW in the FP cavity, always with the same parameter : 33% laser amplifier ratio

new plots, with the accelerometer placed directly on the optical cavity table (on the marble part), close to the oscillator.
the correlation seems a bit better....

we would need to put the accelerometer on one foot of the table to check if the table has not some micro-movement or at the IP to check if the water in the dipole are doing some vibrations.

this morning, I did some new plots with the accelerometer placed on the top of the case of the OneFive laser.
I previously observed with the accelerometer placed on the top of one cavity vessel, the correlation between the accelerometer signal and the PZT noise is pretty good but not 100%.

I will move the accelerometer on the optical cavity table.

at the end of the day, I moved the accelerometer from the top of the cavity vessel to the top case of the Onefive laser.

I quickly saw again quite strong correlations between the vibrations compensated by the PZT and the accelerometer signal.
when one does a laser motor step which makes a systematic lock loss, one doesn't see any signal on the accelerometer.
=> more measurements have to be done.

I would like also to put also the accelerometer on the mechanics attached to the pipe at the IP.
=> to be discussed to find the best place as a lot of cables, the 2 dipoles and the aluminium sheets around the vacuum stufs don't let a lot of place to put the accelerometer....

then, I moved the accelerometer on top of the "X-hutch side" vessel, placed directly on the metal top case, inside the housing.

- the "no noise same range" image shows a standard situation during 10s when there no noise either on PZT or Accelerometer.
on can compare the noise level on the accelerometer with the previous post when it was outside on top of the housing.
its noise is much much lower... which means the housing is properly dumping the acoustic noise at this frequency around 20-30Hz.

so, we increase the accelerometer measurement sensitivity to better measure its noise.
- the "no noise new range" image shows a standard situation during 10s when there no noise either on PZT or Accelerometer but with a smaller range.

- the "noise 1,2,3" images show the situation when the PZT start to compensate large noise with good correlation with accelerometer placed on top of the optical vessel.

- the "noise 1,2,3 not clear" images show the situation when the PZT start to compensate large noise with correlation with accelerometer but the signal level is not the same as before.
this make me think the origin of the noise is maybe not coming from the inside of the optical vessel.

=> conclusion : we see for the first time a correlation between the PZT noise and some vibration/acoustic noise.
now, we have to investigate the precise origin of this noise (or the different sources).

this morning, I did some measurement with nobody interfering with the tests.

- the "no noise" image shows a standard situation during 10s when there is no noise either on PZT or Accelerometer.
most of the time, we are in this situation.

- the "slaping door" image shows the case where the large igloo door is opened and slaping when it closes.
the accelerometer and the PZT exhibit correlated noise when the door is slaping.
we can see a PZT "recovery" time longer than the perturbation.
but these events are rare and are not the source of the problematic perturbations.

- "noise 1,2,3" images show the typical situation when the PZT start to compensate large noise without any correlation with accelerometer placed on top of the housing.

=> conclusion, some external noise (to the housing) should not be the source of the perturbations on the PZT.

today I connected a copy of the laser PZT signal to the 2nd scope CH2 (with AC coupling to remove the DC offset) to be able to monitor synchronously the Accelerometer and laser PZT signals.
the accelerometer is still connected to the 2nd scope CH4 and placed on top of the housing.

I filtered both signals in the Labview Signal Express software with a low-pass filter at 30Hz to focus on low frequencies noise (~20Hz).

now, I need to wait to work with the bunker closed to compare with normal operation (if some people work in the same time in the bunker, obviously, we will get some correlation between the accelerometer and laser PZT signals.....)

I installed an accelerometer setup in the bunker.
presently, the accelerometer is placed on top of the housing and its signal is connected to the 2nd scope (33MHz and 500MHz RF beating) on channel 4.
the FPC is locked to ~90kW.

the accelerometer noise is filtered on the Labview Signal Express software in order to focus on the 20Hz noise.
one applied a RII elliptic 5th order low pass filter at 30Hz.

the 20Hz noise can be seen on the PZT which always compensate for CFP frequency drifts.

figure 1 : example of typical accelerometer filtered noise (yellow curve) when the PZT compensation (green curve) is quite (measurement on 4 seconds)

figure 2 : example of accelerometer filtered noise (yellow curve) when the PZT compensation (green curve) exhibits some 20Hz noise (measurement on 10 seconds)

conclusion : there is no clear evidence of a correlation between accoustic noise outside of the housing (measured by the accelerometer) and the 20Hz noise in the laser PZT compensation.
=> putting the 2 signals on 2 different scopes doesn't help because the slow acquisition done is not synchronous.

next try : use the same scope and put the accelerometer inside the housing, for example on top of one of the FPC vessel.

this morning, I restarted the lock of the CFP after almost 2 weeks without operation.
after quick alignment and CEP tuning, I got 88kW for 33% ratio for the laser amplifier.

the FPC seems pretty far from the RF frequency (2.8kHz @ 33MHz) but it's possible the present RF frequency has been tuned for 61.5 or 70MeV (the present FPC length is tuned for 50MeV).

Jean-Noel had to reinject some SF6 gas in the section pipe => it immediately produces some lock losses during the whole filling process and even several minutes after he finishes.
this is an interesting correlation with the vibrations from the beam pipe => one needs to install the accelerometer to check the behavior.
but the lock losses were not related to "high frequency" or "20Hz oscillations" noises which are the 2 main processes for lock losses, it was just like some "cuts" in the signals.

I finally got 90kW in the FPC for 33% ratio for the laser amplifier after walking alignment procedure.

today we did x-rays => we got 41 000 pA at maximum and 90kW in the FPC for 33% amplifier ratio.

finally, I changed the strategy for the feeback on RF.
i removed the integration and derivative parameters and reduced the gain parameter :

P=0.25 / I=D=0 => it seems to be more stable => less low frequency oscillations becoming larger and larger during a perturbation.

the machine people saw a correlation between BPM jumps and vacuum gauges peaks related to breakdowns in the beam pipe.

these breakdowns could produce some accoustic noise which could be related to our lock losses.

=> we tried to do a corrrelation between our lock losses and the vacuum gauge peaks => WE DON'T SEE any correlation !!!

This afternoon, we started almost all the parts of the machine with Vincent and Nicolas, and we didn't see any change in the Loop noise signal.
we saw only a peak comb at 10Hz on the Loop noise signal when the septum is ON. the amplitude of the peaks is then related to the voltage on the septum.
but we still don't see any noise correlation on the PZT signal.

the global conclusion about CEM noise is it is not related to laser PZT noise.

if the 20Hz oscillation is coming from a mechanical unstability in the CFP, we should be able to trigger it by moving the longitudinal motors of the CFP.
we tried also to move the MOT.03 and MOT.06 motors and we didn't see any clear correlation with the 20Hz oscillation.

the 20Hz oscillation could come from the CTA pressure variations on the housing => we can try to trigger the oscillation on the housing.

The goal of these measurements is to check if one can find some correlation between the CFP lock losses observed during a run and some CEM noise in the bunker.
the main issue is coming from a 20Hz noise apearing/disapearing in the laser cavity PZT signal.
is it a real cavity length noise that must be compensated or is it a pickup noise in the error signal which produce some unwanted compensation ?
the second issue (apearing much less often) is a higher frequency noise visible mainly in the transmission signal of the CFP which becomes wider and noisy during 1-2 seconds.

This morning, I installed a simple wire loop all around the table (the machine is OFF).
this wire is connected to the wire of a BNC connector at one end and to the ground of the same BNC connector at the other end to form a loop.
this BNC connector is connected to a 1kHz low pass filter to remove high frequency CEM noises and connected to a 10-1000 variable gain amplifier, plugged to the CH3 of the CFP scope (instead of the PDH/PZT-CAV signal).

I locked the CFP above 80kW after tuning the CEP.

I took several pictures to illustrate what I observed :
- on the right, the scope signal in time domain (blue = TRANS / green = PZT laser / cyan = REFLECT / pink = LOOP noise)
- on the left, the spectral analysis in frequency domain (red = PZT laser / white = LOOP noise)

in these 3 measurements, I don't do anything to the CFP, I just wait for the signals.

1) normal condition : see picture "Without Noise @ 20Hz"
the PZT signal in time domain is relatively flat and the 20Hz noise is barely visible in frequency domain.
we don't see any correlation with the LOOP noise for which one can see clear 50Hz and 100Hz peaks.

2) 20Hz noise condition : see picture "With Noise @ 20Hz"
the PZT signal in time and frequency domain exhibit a clear 20Hz oscillation.
the REFLECT and TRANS signals are also correlated to this 20Hz oscillation.
but still no correlation with the LOOP noise in time or frequency domain.

3) 20Hz noise lock loss condition : see picture "Delock @ 20Hz"
because of the lock loss the spectral analysis of the PZT signal is meaningless,
but in time domain, one can clearly see the growth of the 20Hz signal, inducing large noise on TRANS and REFLECT signal but still no correlation with the LOOP noise.

Conclusion : the 20Hz noise seems not related to any CEM noisy signal but more probably to a real mechanical noise in the CFP (not in the laser cavity because it would be seen in the phase noise measurements).

in this measurement, I tried to produce a tone around 400Hz with my voice, close to the FP cavity.
it is difficult to right volume as the cavity can easily lose the lock.

4) Voice noise condition : see picture "Voice noise @ 400Hz"
here, I changed the spectral span to 500Hz.
one can see a small bump in frequency domain around 400Hz due to the accoustic noise in the PZT signal.
the general shape of the PZT spectral signal is maybe related to the PID parameters : one can see more signal at lower frequencies.
in time domain, one can see TRANS and REFLECT signals are more noisy than before and these signals shapes seems identical to what has been observed when one has lock loss due to high frequency noise.
maybe we could put a mic in the bunker, connected to one scope to check if sometime one doesn't have much more accoustic noise...
 

this morning, I restarted the cavity after the Christmas shutdown.
everything went fine.

I got ~ 86kW for 33% laser amplifier ratio after optimization of the CEP and alignment with walking procedure.

IcePap controllers are OK and the displacements (MOT.03 and MOT.06) let the CFP locked.

I locked also on the RF frequency (I tuned the laser and CFP cavity length) => +4.1ns (C2-C4) between the 10Hz trig (C2) and the 33MHz laser signal (C4)
I observed that the search & relock range on the regulator B plays an important role on the RF locking stability.
so, I increased the previous range +/-0.25V to +/-0.5V.
we have always the 20Hz noise which can be increasing some time but the lock seems more robust.

this morning, Kevin reduced the steering current in the IcePap controllers of the FP cavity motors.

the motor MOT.06 was producing a pattern in the Transmission signal when it was moved and doing a lock loss very often.
so, we changed its current from 0.8A to 0.4A and it fixes the problem => no more systematic lock loss.

so, we changed also the MOT.03 steering current from 0.8A to 0.7A.
on this motor, we have also a false warning about the Low limit switch which seems to be activated (strange because, we are always using it in the positive direction)
Kevin reverted the logic to remove the message.

we obtained 91kW in the FPC for 33% amplifier ratio after CEP and alignment tuning.

we did synchronized xray production with a relative delay between laser 33MHz and trigger (CH2-CH4) of +4 ns

major result of the day: X-ray vertical scan by moving the hexapod

abscise : hexapod position
ordinate : xray flux in asynchronous condition

red curve : continuous injection at 10Hz (the scan lasts for ~5 minutes)
green curve : one single injection (the beam is not extracted).
blue curve : one single injection, ~ 10-15 minutes later

This morning, I added an amplifier on the 33MHz beating signal in between the mixer+LPF and the scope/Laselock.

it seems to improve the robustness of the RF/FPC lock.

because of this gain, I increased the upper and lower thresholds on the search criterion of the RF/FPC lock from +/-50mV to +250mV/-200mV

=> see the picture of the Laselock parameters.

the optimum phase for X-ray production is roughly +3.6ns between C2 (machine trigger) and C4 (33MHz laser signal).

we have to use the machine at 70MeV with a new frequency at 500.0325MHz / 33.3355MHz.

=> we need to move the FPC tomorrow by roughly 60kHz @ 500MHz.

last thing we tried :

we removed the threshold on the FPC/RF error signal.
=> we cannot choose automatically the RF bucket anymore (we need to manually let the phase drift slowly and start the lock at the right moment).
=> but the FPC/RF lock seems more robust.

in that case, we have 2 different sources of lock losses:
- the ones dues to the laser or CFP motors move.
even at low speed or in "piezo scan" mode, one observes too fast mouvement that are not properly compensated and involving some phase shift.
- the ones not related to any action.
=> the 20Hz signal seems to increase until it makes the system losing the lock

we observed that :

- the MOT.03 motor always exhibits some perturbations on the transmitted, reflected and PZT signals (see "peaks" in the picture) in contrary to the MOT.06 motor.
does the differences come from the motor relative positions (-900 000 steps for MOT.06 and -100 000 steps for MOT.03) or from the controller configuration ?

- the stability limits (oscillations arise) of the PID for the RF/CFP locks are P = 1 / I = 0.0001 / D = 5
then we put the new PID parameters : P = 0.25 / I = 0.000025 / D = 1

- the 20Hz oscillations are stil arising from time to time

- the lock laser/CFP is pretty robust, one observes more RF/CFP lock losses.

- we removed the 250Hz filter on the RF/CFP error signal to increase the feedback BW but we didn't see a any improvement

- at 5pm, the laser/CFP lock seems as stable as in the morning, then we don't see any change in stability during time.

we loggued CFP power measurement and signals from the cavity (~ 1GB of data)