model restarts logged for Fri 19/Sep/2014
2014_09_19 01:41 h1fw0
2014_09_19 04:01 h1fw1
2014_09_19 09:02 h1fw1
2014_09_19 15:27 h1fw1
unexpected restarts of frame writers
J. Kissel The Message: I've cross-checked the calibration of all the ground sensors at the X End-Station, and used that knowledge to gain further confidence in their assessment of ground motion and ground tilt (second attachment). With these confirmed sensors, I tried to figure out why no one can find coherence between the ISI T240 X and either the GND BRS RY or GND T240 X (first attachment). My conclusion is that the ISI ST1 X DOF is limited by re-injected noise from ISI ST1 RY DOF between 20 and 200 [mHz], because we've copied and pasted our X T240 blend filter to RY without being conscious of this tilt-horizontal-coupling path (fourth attachment). I *think* this noise, is T240 sensor noise in this 20 to 200 [mHz] frequency band (see fifth attachment). This can be solved by sacrificing unneeded higher-frequency performance in RY (say between 1-10 [Hz]), and moving the RY blend up a bit, and making the T240 high-pass roll-off more aggressive, or "faster," as a function of frequency (third attachment is current X / RY blends). %%%%%%%% % The Deets % %%%%%%%% Calibration: ------------ In the second attachment, 2014-09-18_H1EXGND.pdf, I've calibrated everything into translational acceleration units. I summarize here, then explain the details after. Summary: (1) H1:ISI-GND_STS_ETMX_X_DQ 1e-9 [(m/s) / ct] --> Let DTT differentiate once to acceleration units (2) H1:ISI-ETMX_ST1_BLND_X_T240_CUR_IN1_DQ 1e-9 [(m/s) / ct] --> Let DTT differentiate once to acceleration units (3) H1:ISI-GND_BRS_ETMX_RY_OUT_DQ 1.568e-8 [(m/s^{2}) / ct] (4) H1:PEM-EX_SEIS_VEA_FLOOR_X_DQ 7.9e-9 [(m/s) / ct] --> Let DTT differentiate once to acceleration units (5) H1:PEM-EX_TILT_VEA_FLOOR_X_DQ 5.5e-8 [(m/s^{2}) / ct] (6) H1:PEM-EX_TILT_VEA_FLOOR_T_DQ 5.39e-7 [(m/s^{2}) / ct] For (1) and (2), myself and the SEI group have graciously calibrated these channels into 1 [(nm/s) / ct] in the front end, following the electronics chain as described in D1001575. So I merely have to convert to (m/s), and let DTT handle the differentiation by requesting m/s^{2} / Hz^{1/2} on the units menu. For (3), Krishna and I have installed a similarly dead-reckoned calibration that we believe is in 1 [nrad/ct]. However, converting to translational acceleration by multiplying by g = 9.8 [m/s^{2}/rad] and by 1e-9 [m/nm], leaves a discrepant factor of 1.6 between the GND T240 and the GND BRS (see pg 1 of 2014-09-18_H1EXGND.pdf), where there is great coherence, between 10 and 100 [mHz] and we expect the signals to be the same. Also notice the how the harmonics is the 8 [mHz] resonance pollute the spectrum (the BRS has been rung up to +/- 200 [ct] during this measurement period). That's when I invoked the PEM sensors, hoping they would be coherent enough between the sensors to cross-check, but alas, in the 10 to 100 [mHz] region, they're too noisy to really tell if the GND BRS or GND T240 are "right," so I added in the extra factor of 1.6 assuming the T240s were correct, hence 1.6 * 9.8 * 1e-9 = 1.568e-8 [(m/s^{2}) / ct] For (4), I used the pem.ligo.org prescribed 7.6e-9 [(m/s) / ct], it matched the GND T240 very well (within the 22% quoted precision) in the frequency region where we expect them both to be sensitive to translation, i.e. above 100 [mHz]. For (5) and (6), since the instrument has not yet been successfully calibrated (see LHO aLOG 13623) I assumed the that GND T240 and GND PEM Guralp were correct, and simply scaled the PEM TILT X channel to match them above 100 [mHz], ending up with 5.5e-8 [(m/s) / ct] (and let DTT do the differentiation). I then blindly assumed that the Tilt channel uses the same calibration value, but for rotational displacement, i.e. 5.5e-8 [rad/ct]. Scaling by g = 9.8 [m/s^{2} / rad] that yeilds the above 5.39e-7 [(m/s^{2}) / ct]. It seems to match up reasonably well, and it's not hard to imagine that the electronics chain is the same for both channels, but the sensor appears to be limited by some noise incoherent with the GND BRS in the 10 to 100 [mHz] region. The Tilt-Horizontal Coupling Model: ----------------------------------- On the final page of 2014-09-18_H1EXGND.pdf, I plotted the ISI performance against all of the ground sensors, and noted the hump between 10 and 100 [Hz] that looked suspiciously like a blend filter bump. Going on a hunch I've had for a while Similar to what I did in my thesis, knowing that we've thus far only copied and pasted our X blend filters to the RY DOF, I used the same 1-stage, MISO model (e.q. 5.3 on pg 80) to predict how much the residual platform tilt (RY) motion is coupling into the X DOF, G_x g x x = ------- * ----- * F * res RY from res RY 1 + G_x w^2 T240 HP Thankfully, at these low frequencies (f < 1 [Hz]), the ISIs have loop gain, G_x, much much greater than 1, the closed loop gain (the first term) is well-approximated by unity, and I only have to know the blend filter F^{X}_{T240 HP}. This model shows varying degrees of success. (1) Between 50 and 300 [mHz], this predicts the X ST1 motion exactly. ISI T240 RY doesn't show coherence, however, but I'm confident that's because it's incoherent sensor noise of the RY loop in this band -- at least up until 100 [mHz]. I'm still confused why the double-peaked microseism (100-200 [mHz]) in ISI X is very coherent with GND X, but (a) doesn't show up in nor is it coherent with the GND RY spectrum, and (b) perfectly matches the shape of the ISI RY motion projected into ISI X. (2) The model WAY over predicts the X displacement between 10 and 60 [mHz]. I've triple-checked my blend-filter-multiplication-via-DTT-calibration, and I'm confident I'm doing it right -- see third attachment. Steps: - Grab a matlab version of the blend filter from ${SeiSVN}/seismic/BSC-ISI/Common/Complementary_Filters_BSC-ISI/aLIGO/TSheila.mat - Ask matlab for its poles and zeros via [Z,P,K]=zpkdata(High_Pass_Filters(1)), where X is the first DOF - Turn them from [rad/s] into [Hz], by dividing by -2*pi - Copy them into foton, bode plot, and find the correct normalization factor such that the filter asymptotes to 1 at high-frequency (-160.202 [dB]) - Copy the normalization gain, poles and zeros into DTT and multiply the correct displacement calibration, gain: 1/(2*pi) [rad / (rad/s)] * 1e-9 [rad/nrad] * 9.8 [(m/s^2) / rad] * 1 / (2*pi)^2 [m / (m/s^2)] poles: 0, 0, 0 zeros: [none] (3) Independent of the model's confusion, at least it shows lots of room for improvement with GND X to ISI X in this region (100 - 700 [mHz]).
J. Kissel, S. Karki Typo in the above entry -- I used 7.6e-9 [(m/s) / ct] for the H1:PEM-EX_SEIS_VEA_FLOOR_X_DQ (PEM guralp), which is much closer to the new pem.ligo.org value of 7.39e-9 [(m/s) / ct], which Sudarshan has recently updated. My value of 7.6e-9 [(m/s) / ct] was from the previous value reported on pem.ligo.org, which I naively assumed hadn't changed. It's a 3% discrepancy; well within the quoted 22% uncertainty reported for the PEM Guralp. One more example of use wanting to be able to edit aLOGs more the 24-hours in the past...
Rana, Alexa
We repeated a similar procedure as to alog 14022 but this time for L2P. At low frequency, i.e. 0.01Hz we found the optimal gain to be 0.0106. As Stefan mentioned in his alog, the L2P coupling seems to fall off faster than 1/f^4 at high frequnecies (ie about 3Hz). Therefore at high frequencies we want our filter to be essentially zero. The attached screen shot shows the original filter (FM1 red trace) and two new filters (FM2 blue trace, FM3 green trace). FM2 has a gain at low frequency to match our measurement. The phase is flipped so that the gain in the filter module can remain -1. Then the filter falls off as 1/f^2. FM3 is intended to provide the user an option to include a notch at the resonance at 0.7Hz as seen in Stefan's L2P TF. The other two resonances at 1.1Hz and 0.5Hz seem to cancel with the P2P TF, so we did not include these.
As a reference, the FM2 filter: zpk([],[0.766044+i*0.642788;0.766044-i*0.642788],-0.0106,"n") and FM3 filter: notch(0.749,10,30)
After we turned on this filter and locked the michelson on the dark fringe, we quickly noticed that the PIT motion was actually worse. This was due to two things. First, I had not saved and properly loaded the filter, so the sign was off. Second, my phase at mircoseism was falling off from 180 too quickly. You can see this in the first screenshot by comparing it to FM1. With these corrections, the PIT motion appears to be better.
For reference FM2: zpk([],[0.098242+i*1.12291;0.098242-i*1.12291],-0.0106,"n")
Tonight I will run L2P transfer functions to examine the residual motion. These measurements will start at 1am (and require that the BS oplevs are turned off).
I have to apologized that I accidentally had the PRMI locked in the first a couple of minutes of Alexa's scheduled measurement. As soon as I noticed it, I switched the LSC and oplev damping servos off, but this must have screwed up some data points around 10 Hz.
See the first plot for the HAM6 HEPI sensors spectra collected today; this is with the ISI Damped and the HEPI unstopped and I believe the position loops closed. With the L4Cs looking pretty good, it doesn't look like an interference issue. The character of the V2 IPS suggests it has a problem,... although it does show some character around 25hz like the other sensors. Plotted with HAM4 from April for reference. There is no earlier HAM4 Spectra in the files. Could it be something with the control loops?
The second attachment compares the HAM6 Spectra 24 hours ago before I unlocked HEPI. Maybe I screwed up some cabling this morning. I checked the local basis IPS and they didn't shift enough to go out of range but I should check that physically.
Recall I have not run new TFs on this platform. There are TFs for HAM6 from October 2013, these look very good and very similar below 10hz to say HAM4--see comparason attachment #3.
Don't understand the foton file generic loops. Jim and I looked at the Isolation fotons for HAM6 5 & 4. The boosts are the same but the controller filters are not. HAMs 4 & 5 are near identical for all dofs. At HAM6 the phases and shapes are all the same but very different in magnitude for every dof. Here is the ratio
Dof | Ham4/6 |
X | 1 |
Y | 4 |
Z | 2 |
RX | 1.5 |
RY | 3 |
RZ | 2 |
Probably no surprise the boost for X will work but the loop goes unstable with all other dofs.
Loops are closed but with no boosts. I haven't started guardian for HAM6, maybe it exists. I'll try when I won't interfere.
Dan, Koji (from a distance), Rana
We used an OMC mode scan to measure the modulation depth of the 9 and 45MHz sidebands, to close the loop on the recent changes to the amplification path before the EOM. Koji had done this previously, and measured Gamma1 (9MHz) = 0.198, Gamma2 (45MHz) = 0.305.
Today, we measured Gamma1 = 0.208, Gamma2 = 0.240 +/- 0.01. The error in Gamma2 is due to an asymmetry between the upper and lower sidebands; using the ratio of the 45MHz USB to the carrier returns Gamma2 = 0.251, while the LSB returns 0.233. This 10% discrepancy between the 45MHz LSB and USB is consistent across several sweeps of the cavity. The analysis code used today was thrown together a little quickly and doesn't do a sophisticated job of integrating the curve around the peaks (in fact, doesn't do any integration, just compares the peak heights to the carrier), but when applied to the same data that Koji used earlier this month it finds values similar to his to within a few percent.
The attached plot has an overlap of five cavity sweeps, with the peaks of the 9 and 45MHz sidebands used in the calculation marked with a cross. The data are here. The mode scan was performed off a single bounce from ITMX. This weekend we'll do a more careful scan for both ITMX and ITMY to calculate (more?) accurate numbers of the mode-matching for each path.
Daniel, Kiwamu, Rana
The modulation depth mystery for 45 MHz is still unsolved. Educated guesses are welcome.
In December of 2013, the people have measured that the input power to the EOM was 11 dBm and the modulation depth was 0.07 radians.
Today, we checked at the CDS electronics room, the field rack, and all around the PSL table and near the EOM, for the presence of any more un-documented amplifiers. We found none.
We measured the power at the field rack to be +12 dBm. So we think that based on the EOM calibration from December that we should be getting < 1 radian.
However, the OMC scan seems to show a modulation depth of ~0.24 radians for the 45 MHz sideband. Unless the carrier peak is saturated in the DC readout PDs with a single bounce Michelson beam, this seemsto be impossible to square with the EOM calibrations.
In December of 2012, Volker measured the modulation coefficient to be (0.33 rad / 10 Vpp) = 0.066 rad/V. 10 Vpp ~ 24 dBm (for 50 Ohms). This agrees well with Kiwamu's entry.
We have the further evidence of high modulation that the REFL and AS demod signals didn't change much (or at all) when the ZHL-1A was pulled out of the modulation chain.
What's going on here?
Measured the gain of ZHL-1A amplifier that we removed at 45.5 MHz through a 20 dB attenuator into a 50 Ohm loaded scope; gain is nominal over a wide range of input power levels.
Richard called and informed me that a circuit breaker has failed while Bubba was restarting supply fan 5 in AHU3. We cannot repair this today so those in the control room may be a little warm this weekend. There is still air flow, perhaps 1/2 of normal and currently the control room seems to be the most affected.
The attached plot shows 4 rooms in the area and clearly the control room is the worst.
I would suggest turning off as much equipment and lighting as possible for the weekend. You may prop open the exterior doors (near the control room) which may help. Please do not let coyotes into the building.
PS. The projectors are likely big heat sources.
I've installed another Mac Mini to drive the left hand side projector in the control room. It is accessible as projector1, in the same way as projector0 which is running the seismic DMT plots - that is to say, using Screen Sharing. Note, if you want to run DMT on this machine, you will need to talk to Jim/Dave to get it set up - any of the other tools should work as expected.
With the looming weekend I bit off a mouthfull this morning when I suggested I unlock the HEPI.
I did so but while doing that I noticed much of the Actuator ranges were very consumed--talk to me if you want more info. Range of motion tests failed on several Dofs but only because the offsets were so large that the test criteria was satisfied at the start--obviously I need to work on the test script.
Back in October 2013, TFs were run and controllers developed--I don't know who at this point. Regardless, these loops are too aggressive and the Boosts ring up. I've got the controllers on manually now without the boosts controlling to the cartesian position before I unlocked the platform. However, some of the loops aren't quite getting to the Set Point. I don't know if it is the controller without the boost or if it is interference with something in the platform. Given that I see the Postion approaching the Set Point but not quite getting there has me leaning toward the interference issue.
To do:
* I should install the Generic 2hz loops already on HAMs 4 & 5 and see if the loops can be turned on with the boosts and if the error point will go to zero.
* Run more range of motion/linearity tests to assess interference more thoroughly.
* Collect Spectra to confirm intereference
* If motion range can be confirmed, collect Transfer Functions; if interference is a problem, see below.
* Redo filters with good TFs and get under control.
**** Do more with HAM6 ISI, currently only damping!
Interference problem solutions:
1)--May be able to remove Bellows shields and modify to give more room--not easy but not too bad. It seems to me that building the Actuator with lots of welding distorts things enough that the ideal Actuator floating/relaxed/center position isn't always the center position.
2)--Disconnect Actuator from Foot and rezero. May be the better solution but will take a day at least. We are pretty good at connecting the Actuators without shifting the platform too much but that is the main risk--Disconnecting the actuator and rezeroing will lose the reference IPS position.
The daqd executable on x1fw1 has been changed from daqd-trunk-r3626 to daqd-r3827 to test changes to trend.cc to properly record integer trend files.
Test is successful, trend data is now correct for integers. I will leave this installed and check in changes to code.
Rana, Alexa
Since we saw a lot of YAW motion from the BS last night, we decided to look at the L2Y coupling again. The original filter FM1 installed in the drive align was created via alog9394.
We injected a sinusiodal excitation in BS-M2_LOCK_L and examined the M3 oplevs. First we excited at 0.01Hz, and measured peak-to-peaks via StripTool at several gains. Assuming a linear relationship, we then determined the best coupling gain to be 0.0025 (at 0.01Hz). Then we excited at 4.57Hz, and found the mag, phase at two gains. This allowed us to determine the coupling gain to be 0.002203 (at 4.57Hz). We then created a filter accordingly to match the respective gain couplings at low and high frequency. The filter is installed in FM2 (see first attached image). The gain in the filter bank should be set to -1 for either FM1 or FM2. The second image shows both FM1 (blue) and FM2 (red).
For refernce FM2: zpk([0.694211+i*0.827328;0.694211-i*0.827328],[0.642788+i*0.766044;0.642788-i*0.766044],-0.0025,"n")
(Borja)
please find attached a short document on how to identify which ESD quadrant we are driving from the slope of the 'oplev deflection vs VBIAS' plots we obtain during the ESD charge measurements. We know that the quadrant labelling on the Epics channels and MEDM windows do not correspond with the actual ESD quadrants being driven and the discrepancy is different between ETMX and ETMY.
ALIGO Install
Commissioning
3ifo
no restarts reported
Kiwamu, Sheila, Rana
We started by locking PRMI and taking a long spectrum of the BS op Levs, it seems that Alexa's new length 2 yaw filter (alog 14022) has reduced the yaw at around 0.05 Hz by a factor of about 2 compared to alog 13997. The first screenshot attached is a spectrum of the BS oplev with PRMI locked.
We then moved on the DRMI locking. After debugging the guardian a little bit, we were able to catch lock infrequently. Out first lock ended at Sept 19 4:55:19 UTC, and was about 10 minutes long. It locked again at 5:05:36 UTC, at 5:24 UTC we left everything alone to get a clean stretch of data until 5:42:40 UTC. At 5:33 UTC there were a few mode hopping glitches. In general the mode hopping events are not as frequenct tonight as last night, and they are reduced when we have better alignments. The attached video is from a DRMI lock.
We measured the loop gains. We found that with the gain settings used last night (alog 1402 ) gave us a prcl ugf of 130 Hz, and a MICH ugf well below 10 Hz. We now are using a PRCL gain of 6.4 (to get a UGF around 80 Hz) and a MICH gain of -50 (ugf of 10 Hz). Measurements attached.
Kiwamu also noticed that SR3 pitch motion at 0.8 Hz was large, kiwamu implemented a SR3 oplev damping loop with an upper UGF a few Hz and a lower ugf of 0.3Hz is, copied from the PR3 oplev damping servo.
PRMI is now locked.
Here's I'm plotting some signals from a long PRMI lock to look at the low frequency content.
In the first plot you see the spectra. The RMS of the UL coil is ~30k cts (DAC range is 131k cts peak).
The second plot shows the time series. The peak signals are a little over 100k. In the old configuration, the PRM was kept from saturating by driving the upper stages at DC and also by splitting the drive to PR2. By increasing the drive range of the M3 drivers by 10x, we've made it possible to lock the PRC with a single actuator (which is simple). After locking we can now smooth on the M2 driver and reduce the M3 votlage below 1 Hz by ~20-30x. (done by hand, needs to be added into the guard).
The third plot shows the BS optical lever signals in and out of lock. You can see that the noise from 10-100 mHz is dominated by the MICH control signal and its consequent cross-coupling through the imbalance of the BS suspension. After the nice work by the Seismic Gang over this past week, the yaw is no longer a problem and we should be able to reduce the remaining pitch motion in the BS OL by tuning the L2P filtration later this morning.
9/19 update: corrupted BSOL.pdf file replaced with real one
I started preparing the online calibration stuff in h1oaf. Tonight I got the PRCL loop calibrated assuming the UGF is set at 80 Hz.
According to my model, the readout gain of the PRCL loop (which is REFL_A_RF9_I) when the DRMI is locked was estimated to be 3x1012 cnts/m. So I put the inverse of this number (which is 3.3 x 10-7 um/cnts) into the calibration filter to get the error signal converted into um.
Here is the open-loop equilvalent (a.k.a. unsuppressed- ) PRCL noise, calibrated in um/sqrtHz, but in the PRMI locking state ( and therefore the sensing gain maybe slightly different from that of the DRMI, which can result in inaccurate noise floor at high frequencies.)
The model open loop transfer function looks like this:
Note that FM2, 3, 4, 9 and 10 of the LSC PRCL loop are assumed to be engaged. Also FM1, 3, 4, 6 and 10 of PRM PR2 are assumed to be engaged. There is no digital filters in the PRM M3 stage. PR2 is not included in the model as is in reality.
Based on the sus models and the lateset digital filter settings, I confirmed that the cross over frequency of the PRM M2 and M3 stages in the model is at 4.5-ish Hz which agrees with Sheila's adjustment (alog 14019). The model takes the latest factor-of-10 increase in the coil driver strength of the PRM M3 stage into account.
I am summarizing here recent progress on SEI units, and providing guidelines regarding what needs to be done next week.
A lot of different configurations have been tried in the past 10 days.
Two main conclusions:
- We don't want to put low blends on Stage 2 to avoid low frequency noise reinjection ( see logs here and here)
- We have a lot of pick up between Z and RZ. Relaxing the Z actuation (higher blend) helps in Yaw (see logs here and here)
The best BSC-ISI configuration so far is:
Stage 1
- X, Y, RX, RY, RZ: TSheila blend
- Z: T750mHz blend
Stage 2
- T750mHz blends on all degrees of freedom
We've successfully implemented some sensor correction on HAM2-ISI in X (see log here).
It helps the absolute motion by a factor of ~4 in Yaw, but we have to check what the motion of the cavity looks like (see the to-do list below)
A bunch of improvements have been made on the commissioning scripts (see log here). Jim is finishing up the work started on Routine_6 and should commit that pretty soon.
We also created some simple scripts to make the commissioning/debugging process faster (see log here).
BSC-ISI Sensor Correction
By doing some comparison with the LLO configuration (see log here), it seems that the next big step would be to implement sensor correction on the BSC-ISIs. It would help us winning a factor of a few around 0.5Hz.
We already started to look into that and we should have some exciting results soon.
Cavity Motion Study/Common Control
Even if we had some promising results with sensor correction on HAM2, we need to check the motion in the cavity: reducing the optical lever motion (absolute motion) doesn't mean that the cavity is quieter.
One of the idea will be to implement sensor correction using the SAME ground instrument for HAM2 and HAM3. It will be interesting to see what the cavity is doing when both of the chambers are control by the same input signal.
We need to:
- Implement sensor correction on HAM3
- Compare PRCL when the sensor correction is on and off on both chambers
see above
DRMI locks have glitches, so best to use the long PRMI stretches. Typically, no one is touching it in PRMI mode. We are leaving it in PRMI_SB all night; you can use POPAIR 18 to find the on/off segments to check the ISI motion sensors versus the PRCL length control signals.
Additional Chores
TSheila Blends which are TBetter on all dof except Tcrappy on Rz, need to be added to ITMY & ETMY.
Unlock HAM6 HEPI and get it on position Isolation Loops--need to run HEPI TFs
HAM6 ISI TFs with HEPI Loops closed, develop/implement Isolationcontrollers.
HAMs 4 & 5 higher Level ISI controllers--on going.
Additional accomplishments with lots of help from Seb & JeffK:
Fixed HAM Isolation plot saving--many previous (HAMs 2 & 3) don't have plot s of the controllers.
Added Chamber/Date/ControlLevel info and Gain Peaking to controller plots.
There was some suspicion that the REFL beam is somehow clipped on LSC REFL_A diode. Yesterday I made a long scan of RM1 and RM2, separately and at the same time.
LSC-REFL_A_LF is good, ASC-REFL_B too.
I had to move RM1 and/or RM2 by a large amount to make the LSC-REFL_A_LF drop.
ASC-REFL_B is also good, though the safe range is smaller than LSC (not surprising considering the short focal length lens in front of LSC diode).
ASC_REFL_A is unhealthy.
Something is wrong with SEG4.
It seems as if SEG4 has a lower trans impedance, in that when I steer the beam to SEG4 (all power falls on that segment) we get 18000 counts, but when I do the same thing to SEG1/2/3 they rail at 32k counts.
In the attached, the beam was in SEG3 at first, and I walked the beam to SEG4, then SEG1, then SEG2, and then back to SEG3 by turning RM1.
The problem seems to be in the WFS head, not in the WFS interface board. Connected the WFS output cable from the REFL_A head to REFL_B input and the problem is still there.
For the moment I adjusted the digital gain of SEG4 such that there's no coupling from PIT and YAW motion to the SUM (new H1:ASC-REFL_A_DC_SEG4_GAIN=1.9), but this needs to be investigated further.
The second attachment shows the REFLA SUM spectrum of before (green) and after (red) of the gain adjustment when I was modulating PIT of RM1 at 3.3Hz and YAW at 5Hz. For comparison, Blue/Brown are the REFLB SUM data which show no alignment coupling.
Could be the electronics in the head or the cable (one end of the differential signal is dead?). I hope that the diode is not damaged.
Though not impossible, I don't think it's clipping because SEG4 never went larger than 19000 counts during RM1/RM2 full scan.
I checked the analog signal coming to the WFS interface chassis using a DB15 breakout board, and sure enough, there was no signal coming from one of the SEG4 differential drivers (don't remember if it was pin4 or pin12).
To check if the problem is in the cable between the rack and the chamber, we swapped the cable for WFSA and WFSB at the chamber.
Right after they were swapped, for a few minutes both WFSA and WFSB looked good and we were confused, but eventually WFSA SEG4 (which showed up in WFSB SEG4 channel because of the cable swap) got back to crappy half-signal state.
We swapped the cables back, WFSA SEG4 got back to good state for some minutes, and again after a while it went back to the bad state.
It seems to me that a crappy feedthrough is doing its own thing. The problem feedthrough is D6-1C1 on HAM1.
(However, it's not totally impossible that the feedthrough is OK but the seg4 in-vac circuit works only for a few minutes after it is powered on because of slowly developing oscillation or thermal problem or whatever.)
We know from the cable swap excersize that it's not the WFSA in-air cable.
I wiggled PRM at 4.13 Hz in PIT (attached, left) and 6Hz in YAW (right).
The DC level during the measurement was: about 80 for LSC-REFL_A_LF, about 40000 for WFS DC SUM.
The OSEM witness is supposed to be calibrated in micro-radians.
Using the above information, the PRM angle to the RIN coupling for these sensors are:
About 1 RIN/urad for LSC-REFL_A_LF, about 10 for WFSs.