yesterday with Daniele, we tried again to move the arm horizontally and vertically from the "zero position" to have a better understanding of the beahvior,
but it is roughly the same :
- a clear beam power reduction (=> cut of the beam) when one moves the horizontal axis
- a very small beam power reduction when one moves the vertical axis

as there are no end-position limits on the axis and as it is possible to "touch" the motors or different mecanichal pieces, we prefered to be conservative:
after several tries, one placed the arm at a position where the HOMs seem to be pretty well suppressed.

for the moment, we will consider this arm in a proper position.

redoing the PDH error signal scheme with discrete components is more flexible and it is easier to check the signal/noise ratio.
now, we are able to get a quite clean lock only with the PZT correction (w/o fast feedback correction using the EOM) even with both motors controllers ON (cavity and laser).

then, one can consider this part is over, even if one can still improve the lock with the EOM.
see this post for that part and some details on the new PDH signal scheme : https://elog.lal.in2p3.fr/FPC/THOMX+commissioning/240

this afternoon, I installed the frequency detection scheme in the bunker :

- I changed the large reflected photodiode DET100 by a fast small DET10.

- a 50ohms splitter is connected to the photodiode :

* one cable is going directly to the scope for monitoring
* one cable is going to the 500MHz sharp bandpass filter to select only this harmonic.
despite the small BW of the filter, one gets 3 harmonics : 500MHz + (0 +/- 33) MHz with a bit less power on sidebands.
power on 500MHz : ~ -30dBm if the laser amplifier is at 30%

- after the BPF, one goes directly to the RF amplifier which is 2x Minicircuit amp ZX60-33LN in cascade to get ~ +3dBm

- then one goes to the level 17 mixer on the RF port.
the LO port is feeded with the 500MHz coming from the Ring reference oscillator.

in attachement is a picture of the output signal : ~ 1.5Vpp
I think the width of the signal is coming from the 33MHz sidebands which are not perfectly removed by the 20MHz BW of the scope.
one finds a frequency shift of 250kHz.
as the laser frequency has been set at 500.25MHz (on the 15th harmonic), would it mean the Ring reference oscillator was at 500MHz sharp ? => to be checked with Kevin.

here is the picture of the L-shape arm and the inside of the vessel.

this image corresponds to the "zero" position on each positionner.

this morning we locked the laser and the FP-cavity with ~ 40kW => some HOM appeared "randomly" depending on the power.

we tried to play manually on the L-shape arm :
- first, we moved horizontally to put the L-shape in the beam axis
at some point, we saw the power level divided by ~2, we stopped and came back to restore the full power.
- then, we tried to move vertically.
whatever the direction, we were not able to see a clear cut of the beam.
BUT "strangely", at some point, the PZT drift followed the direction of the motion and was not really depending on the "cooling" or "heating" process when the lock stops or restart.
"strangely" again, at the endpoint, the power dropped by ~20-30% after a delock but it was impossible to restore the power when we put the arm at the initial position.
=> we had to adjust the FP-cavity alignment to restore the power !!!
=> we had the feeling that moving the arm could have misalign the cavity axis !!! :-(
=> we have to discuss with Yann and get the CAD files of the cavity to have a better understanding...

today, after the several unsuccessful attempt yesterday to get an improvement in the lock,
we decided to "rebuild" the error signal electronics block by block and step by step :
the scheme is basically :
- DET36 photodiode
- followed by a 10MHz low pass filter to remove the laser frequency harmonics and keep only the modulation frequency at 8.4MHz.
- connected to a FEMTO HPVA AC-coupled 40dB gain amplifier with 50ohms in parallel on its input (which is connected to the photodiode).
- connected to a Minicircuit mixer which is also demodulated by the generator at 8.4MHz
- followed by a 1.9MHz low pass filter to remove image frequencies

when this signal is sent to the Laselock box, the lock of the cavity is possible but very noisy.
we need to put a large D values in the PID to maintain the lock at the price of oscillations and gain loss ! :-(

when this signal is connected also to the M250 video EOM amplifier (which is 50 ohms), but this amplifier is not used,
we suddenly got a much better lock (see the attached pictures), certainly due to the 50 ohms connected to the input of the Laselock system => to be verified.
one could have some noise current at the Laselock input which produces less noise at the output when the input impedance is 50 ohms, instead of the several kohms when the input is unloaded...

then, we were able to get a stable lock at ~40kW with 30% coupling and 30% of amplifier :
see this post with the same values : https://elog.lal.in2p3.fr/FPC/THOMX+commissioning/227
but now, the lock has been done WITH ALL the motors controllers ON !!! :-)))
now, we can try to improve the lock with the fast feedback loop.

and the only real problem is the mode degeneracy we need to block with the L-shape.

today, during an attempt in improving the CEP with the Smaract controller CH2, we lost the laser modelock.

at this point, we decided to reference the Smaract CH2 for CEP :
now, 0 mm is roughly the middle of the motor range.
the endpoints of the motor are : +/- 5.9 mm
once we found back the laser modelock, we searched for the region where the modelock was still effective : +/- 1.8 mm

previously, we were using a spare Smaract MCS controller to drive the Onefive linear stages (without -LV -low vibration- option and using Ethernet),
during the repair of the initial MCS controller with -LV option.

this morning, it has been remplaced by the initial OEM Smaract  MCS controller integrated by Onefive (with -LV option and using USB).

it worked fine with the Precision Tool Commander 2 software !

today, we tried to lock the FP cavity with the Smaract motors ON (with option -LV).
we know the Smaract controllers produce some noise and the lock is very bad or impossible when the controller in ON (whatever the displacement mode is : closed or open loop, or piezo scan).
then we need to do a fast feedback loop on the EOM inside of the Onefive laser.
the problem is we cannot fill directly the error signal (~ 300mVpp) as the signal level is too low to produce some effect.

1- we tried to use the AC-coupled homemade amplifier alone but the output range is too low (+/- 3V)

2- we tried to use the M250 video amplifier for EOM with AC-coupling but the output range is still to low (+/-30V ? => to be confirmed)
but we saw an improvement in the locking.

3- we tried to combine AC-coupled homemade amplifier + M250 video amplifier for EOM with AC-coupling.
we are able to lock (~ 30% coupling) but the lock quality is very poor : we clearly see that we oscillate around the maximum of the Airy peak.
we tried several combination of the global gain (fast feedback + Laselock) using the diffuser, of the Laselock PID parameters but it seems we are not able to lock properly.

we measured a global delay of this double amplification stage of 80-90ns for the homemade amplifier and 150ns for the homemade+M250 amplifiers.
this delay is compatible with ~ 500kHz BW for the feedback => it seems it is not the reason...

we measured also the linearity of the homemade amp => there is ~ 30dB between a frequency and its 1st harmonic even at low signal...
the amp scheme is not very linear.
in comparison, the HV M250 amplifier exhibit > 80 dB of linearity !
we will try to remplace the homemade amp by a commercial FEMTO amplifier to imrpove the linearity and see if it improves also the lock.

we measured also a 100Hz AM modulation on the output signal of the homemade amp+M250.
=> we can try to work in differential (HV+ - HV-) to see if it helps to remove this modulation.

today, to match the ring frequency at 500.25MHz (15th laser harmonic), we did a frequency shift on the laser using the Smaract motors (not too fast, ~2µm/s, to prevent laser modelock loss)
and we "followed" this shift on the FP-cavity using the FP-cavity motors.

we did several steps during the operation, to control the alignment, coupling and transmission.

finally, we locked again the laser and FP-cavity at 17kW for 30% of input power.
the coupling was quite low ~10%

on the August 30rd and 31th, a global power shutdown was forseen for ThomX.

to prevent damage on equipments, I switched off all of them (and disconnected power cables from the wall plugs).

yesterday, after the week-end, all the equipments have been restarted and the cavity as been locked properly.

the power coupled to the input fiber of the amplifier has decreased a little bit from 3.8mW to 3mW during summer.
=> we need perharps to do some alignment on the Schaftner-Krischoff mount.

Yesterday, I checked the 2 output HV channels of the LaseLock : they work properly but the noise is bigger than the low voltage outputs
~ 1mV rms on 0-10V outputs
~ 8 mv rms on 0-100V output

presently, the voltage on the laser PZT is 0-10V, but during the several runs we did, this voltage is too low to compensate low frequency fluctuations:
we need to change often the motors position implying a cavity lock loss.

we have to check if running the laser PZT with the HV output of the Laselock is possible without too much additionnal noise.
=> it can be done simultaneously with the fast feedback loop implementation, which has to be done.

several times already, when the electrons ring injection kicker is running @10Hz, we observed peaks at the same frequency on reflected and transmitted photodiodes,
at the exact moment when the kicker is activated. we have in addition, USB communication issues with the Alphanov amplifier... the connection is lost several times.
sometime, the connection is retrieved, sometimes not... one has to stop the Alphanov application and restart it => the laser power get down to 0 % !!!

one way to get rid to this problem could be to add a "metallic sock" around the USB cable, to connected to the ground to remove the external charges.
we could ask also for a kicker shielding as this noise could perturbate several systems in the bunker.

Today, I connected the BNC-DB9 female prolongator cable to the FP-cavity PZT cable (DB9-DB9) and to the channel B of the Laselock.

I additionnaly connected the reflected laser beam (normally connected to scope ch2) to the spectrum analyzer @ 1GHz to observe the 30th harmonic of the laser comb.
on the 2 attached pictures, the 1st one is @0V on FP-cavity PZT and the 2nd one is @10V on the FP-cavity PZT.
as the laser is locked on the FP-cavity, its frequency follows the FP-cavity length and its frequency changes.

as expected, applying 10V on the FP-cavity PZT, increases the laser harmonic frequency @ 1GHz by about 10Hz => the relative change is 10^-9  V^-1
(as the FP-cavity length is always moving due to temperature or low frequency vibrations, there are always some FP-cavity length fluctuations, and the measurement has to be quick to get a correct evaluation).

Today, I installed a DB9-DB9 female-male cable on the PZT connector of the FP-cavity.
the PZT is connected between pins 1 and 2, with a capacitance around 70nF.
I need to make a prolongator cable BNC-DB9 female to connect it to the feedback system.

with a (dLpzt / dV) of 5nm/V, one should be able to see 37mHz/V on Frep which is equivalent to ~10Hz/10V @ 1GHz (30th Frep harmonics)
I connected a photodiode on the spectrum analyzer to measure this variation => to be done tomorrow.

I don't have any information about the polarity of the PZT on the DB9 connector but I know that the PZT length should increase with positivite voltage in normal operation.
from the Yann documentation about the PZT mount (in attached file), it should mean that the cavity length should decrease when the PZT length is increasing, and then the FP-cavity FSR should increase.
=> to be tested tomorrow.

about the PZT mount :
I understand that the HR face of the P4 mirror is the face placed on the only part with a chamfer (chanfrein), on the FP-cavity side.
in the documentation, the PZT connector orientation is misleading as it is oriented to the FP-cavity side instead of to the "outside" as one can see it in the cavity picture in attached file.

• régler le problème 1:
  Kevin m'a apporté un déphaseur Minicircuits JSPHS-661 (400-660MHz / 180° de phase) qui permet de déphaser le 500MHz de ~ 1ns avec une tension DC 0-10V.
  on peut alors changer le signe du lock pour scanner les 2ns d'une période complète de 500MHz.
   
• régler le problème 2:
  la synchro anneau se fait sur la RF du synthé 500MHz avec une signal de trig fabriqué à partir d'un 16MHz, issu d'une division de ce 500MHz.
  en cas de perte de synchro de la cavité FP, on va relocker sur le 500MHz mais avec une phase aléatoire par rapport au 16MHz.
  on peut donc remplacer ce 16MHz par le signal 33MHz issu du laser de telle façon que l'injection des électrons dans l'anneau se fera toujours avec la même phase par rapport à ce signal à 33MHz.
  il faudra donc envoyer ce signal issu du laser cavité FP au système de synchro anneau, de cette façon la phase d'injection des électrons dans l'anneau par rapport au laser sera toujours la même.
  mais il n'y a aucune raison que les électrons tombent exactement sur le pulse laser (avec la bonne phase).
  il faudra donc scanner la phase du signal de trig pour décaler l'injection machine par rapport au signal 33MHz avec des steps ~ 1ns.
  pour cela, on peut utiliser les générateurs de delais Greenfield Technology GFT1020 actuellement utilisés pour la synchro (résolution 100ps).

voir schéma attaché en pdf

pour préparer le lock cavité-anneau, j'ai un setup de lock en salle optique entre le laser OneFive 133MHz et un synthé à 533MHz (133MHz x4).
ce matin, j'ai pu locker les 2 ensembles avec le laselock avec une stabilité RMS, je pense inférieure à la ps.
ma limite de mesure du jitter temporel au scope est de ~ 2.5ps.

• une fois locké avec le laselock, je peux facilement décaler légèrement en phase les 2 signaux de façon très précise (<1ps) en jouant sur l'offset de lock,
  mais je ai une plage assez petite (+/- 250ps) qui correspond grosso modo aux plages linéaires du sinus (1/4 de période) soit 500ps (F ~ 533MHz => T ~ 2ns)
  en changeant le signe du lock, je peux faire des sauts du lock d'une 1/2 période, soit 1ns...
  mais cela ne suffit pas à couvrir l'intégralité de la période du signal de référence.

=> 1er problème : je n'ai accès qu'aux plages "linéaires" du signal de référence.
il faudrait un petit déphaseur programmable piloté en remote pour faire des steps de 100ps environ, sur une plage de 1 ou 2ns afin d'être sur de scanner tout la période du 500MHz.

• en coupant le lock, les fréquences driftent l'une par rapport à l'autre.
  et en raccrochant le lock, on peut scanner toute la période entre 2 pulses d'électrons par steps de 2ns.
  puis en changeant le signe du lock, par steps de 2ns mais décalé de 1ns.
  on peut donc facilement scanner la période des électrons avec des steps de 1ns et une plage de 500ps autour de chacun de ces steps.

=> 2e problème : lorsque l'on perd le lock de la cavité FP/laser involontairement, on perd l'info de la longueur de la cavité FP.
et lorsqu'on retrouvera le lock, il va se raccrocher sur une autre oscillation du 500MHz.
et donc on va perdre la phase avec les électrons à 16MHz.
=> on peut éventuellement afficher ce signal à 16MHz, en même temps que le 500MHz pour rechercher l'oscillation correspondante à la bonne phase sur le 16MHz.
mais dans tous les cas, il faudra rechercher à nouveau la phase é-/laser à chaque délock.

• Autre possibilité, faire la synchro sur le 16MHz au lieu du 500MHz.
  la tentative aujourd'hui n'a rien donné car le signal est 30x moins intense => beaucoup plus de bruit.
  => le lock n'arrive pas du tout à accrocher même en filtrant énormément le signal IF avec 10kHz de BW.

Today, we locked the cavity with input power @ 30% for the amp => we got 40kW with only 30% of coupling (and a bad lock => we could have more power inside cavity).

P@30% = 16W
Coupling = 30% => 4.8W of input power => Measured Gain = 8300

today, we observed on the photodiode used in reflexion of the cavity, that its voltage level stops increasing after 50% of power for the laser amplifier when the cavity is not locked (the FP-cavity is just a reflective mirror).
the photodiode itself is not saturated (low power sent after a wedge and an absorbing filter) and the reading is done on 50 ohms.
so we are investigating the reason of this "saturation".

1) we did a beam power measurement just after the 2 mirrors, right after the amplifier, with the "big" powermeter:

power ratio        Power (W)
10%        =>  0.9
20%        =>  7.7
30%        =>  15.5
40%        =>  24.2
50%        =>  33
60%        =>  41
70%        =>  48
80%        =>  55.5
90%        =>  62.2
100%      =>  68

plot in attached file

2) we did the same measurement after the periscope:

power ratio        Power (W)
10%        =>  0.82
20%        =>  7.7
30%        =>  15.5
40%        =>  24.2
50%        =>  32.6
60%        =>  39
70%        =>  44
80%        =>  47.4

we clearly see a power reduction from 50% and beyond.
with the viewer we saw 2 effects : a slight clipping in the telescope and some speckles on the periscope mirrors mainly.
the laser is also slightly shifted on the 2nd mirror after the amp.
it is possible the laser beam moved due to thermal effect => we will need to realign properly the whole injection line and be aware about mirrors and lens cleaning !

3) we did some measurements using 2 wedges with the "small' powermeter in reflection of the cavity.
we remove the small aperture half-waveplate to see the effect and we used a lens to focalize the beam:
the half-waveplate has an effect on the reflectivity of the wedges, this is the reason of the difference in the measurement.

                                                               Power (µW)
power ratio        with lambda/2           w/o lambda/2             w/o lambda/2+lens
10%        =>           12.4                           4.8                               5.1
20%        =>            132                            48                                51
30%        =>            262                            96                               101  
40%        =>            370                            137                             147   
50%        =>            420                            158                             175 
60%        =>            430                            165                             181

we observe a clear "saturation" after 50%-60% which is very similar to what we observed with the photodiode in reflection when the cavity is not locked.
the small iris used in front of the injection window of the cavity is a good "candidate" for clipping the beam at high power (we observed it was still quite hot after stopping the beam).
=> we have to redo the measurements after removing it.