From all of us at LHO, a hearty congratulations to Kiwamu and Rie Izumi, celebrating the 0th birthday of Hana Izumi (和泉はな)! * * * Birth date: Nov.1st Height: 19.5 inches Weight: 6lbs 5oz * * *
Wow, congratulations!!!
The TwinCAT/EPICS interface was updated to version 1.1 which includes an alias feature for symbols and structure items. This now makes it possible to export PCAL channels under the system CAL. See also here.
New PLC information has been added to describe the runtime and each configured online task. The update rate for the latency readbacks has been increased to 1Hz.
Previously, it was possible to circumvent the configuration control by making online changes and bypassing the installation scripts. The installation scripts will patch the svn revision number into an EPICS channel, if they were run from a fully updated source tree. A subsequent online change is now recorded and reported as a code error.
IP1 controller has indicated that pump B (1/2 of pump) has been inoperative for awhile now -> Swapping cables didn't change the indication at the controller that pump B still isn't working but has resulted in the "cold" 1/2 of the ion pump to be pumping now (i.e. problem is controller not HV cable or pump) -> CS pressure shows a "bump" reflecting this exercise -> IP1 controller is old style "MultiVac" type and cabling will need to be modified for, as yet to be acquired, new type replacement.
Aaron S., Richard M., Patrick T. I rescanned PLC1, PLC2 and PLC3 in the h1ecatx1 system manager. I then moved the link under 'End Link L0:End Link L3_4:ALS Laser Table Relay:Channel 2:Output' from 'PLC2:Standard:Outputs:AlsEndLaserHeadOut:NoiseEaterRelayOff' to 'PLC1:Standard:Outputs:IlluminatorOut:IlluminatorRelayOn'. I committed the change to svn. Richard and I switched it on and off in EPICS and Aaron verified that it turned on and off at end X.
model restarts logged for Mon 10/Nov/2014
2014_11_10 06:41 h1fw0
unexpected restart.
Okay, recall: WHAMs 2 & 3 HEPI had incorrect signs for the three rotational dofs on the Input matrices. The output matrix, converting the cartesian drives from the control loops to the local actuator basis also had incorrect signs on the rot dofs but there were also screwy magnitudes with errors from 25% to 8x so all over the map. While all the dofs run closed loop, having the output matrices correct will ease the burden on the loops and better enable open loop running. It may also reduce unneeded strain cycling. Remember, I have lots of opinions!
Anyway, unlike last week when the matlab script was using incorrect matrices for my controller development(obviously unbeknownst to me) and the switch I attempted was a spectacular failure, this time, turning the loops back on with correct matrices and new controllers was anticlimatic. No problem turning things on, no trips, and the ISI remained isolated. The most notable thing is that I learned
I'll suspect HAM3 is similar.
See the attached for 50 minutes of trends with chans 1-8 being the HEPI Local Sensors, chans 9-14 being the HEPI Cartesian Locations, and the last two are the OpLev Pitch & Yaw. Notice the right most vertical line I've drawn through the RZ steps as I'm trying to drive the OpLev 'Yaw' back to whense it came. Of course you'll see that the Yaw doesn't move and the Oplev Pitch does. Of course you say, "Hugh, why do you think your RZ channel is better than the OpLev channel?" Well, notice too the other vertical lines positioned at the same time through the HEPI Local Sensors. You should notice the at the RZ move is seen by the local horizontal channels and the vertical channels see nothing. Okay, I know that is circular...just take my word huh. We can do some physical sensor disconnects or something if you insist.
I did tweek the ultimate RZ Reference location to recover the OpLev actual Yaw of ~-8.2 units but the change was <200nrads. I tweeked in the Pitch too but after taking the system down and back up with Guardian, the Pitch and Yaw numbers on the OpLev are changed, slightly. We didn't have the mode Cleaner while I made this change but it relocked as soon as the LASER was back on. The Beams on the AS AIR camera view are up in the upper right though, I don't know but that may be different.
The commissioners shoudl feel free to steer HEPI as they wish.
Finally, I may suggest a guardian change that isolates and steers the DOFs we want to restore and then get the steered free hanging position and isolate the remaining dofs. Certainly you can see how that wold reduce the strains and possibly provide more repeatable position recovery.
A new safe .snap and the updated foton file has been commited for the HEPI.
WP 4939
The epics software for Mac OS X has been recompiled to eliminate the need for mounting /ligo via /Network/Servers/cdsfs0/ligo. There are a couple of advantages to doing this, but it's for the benefit of your cheerful CDS administrators for the most part.
The epics version is the same one that has been running, just recompiled. Hopefully OS X users won't notice any changes, but running epics programs like medm and StripTool may stop running at some point and need to be restarted.
Ubuntu workstations are not affected by this change.
The state of the PSL upon my arrival this morning was that it showed NO output power.
The power watchdog tripped. With the tutelage of Rick savage i was able to restart the frontend. Everything came back as normal except for the ISS loop that was acting erratically. It was discovered by Gabriele that the second ISS loop was still engaged and that was the cause for the unstableness. The output power monitor PD is currently reading 29.4W. ISS is reading ~6.8% dif power @ -2.11 refsignal.
The original log and first comment was entered by me. I used a browser window that was left by Corey from the day before.
In preparation for tomorrow's maintenance on HAM2 and HAM3, I have set the following gains and whitening stages on HAM2 and HAM3 OPLEVs.
Whitening and Gain settings of | Gain (dB) | Whitening stages |
HAM2 | 18 | 2 |
HAM3 | 6 | 2 |
I added the dewhitening filters in the input flter banks of the oplevs and switched on two stages to mirror the analog state. The analog switch states are shown in the attached pics. The label on the cable tells us the name of the oplev that is addressed by the dip-switch board.
The oplevs are, as yet, not calibrated. The HAM3 shows an uncompensated whitening though I have switched on the dewhitening filters stages. Will investigate further.
Ed writes such nice alogs about oplevs and whitening filters and diode lasers and stuff!!
Imagine how nice it would be if I knew what i was talking about!
There's a huge 364Hz line in ALS TRY.
I'm not sure if this was already there for a while. Though we probably changed the loop gain of the ALSY PDH loop when we realigned the green path on ISCTEX today.
alexa, nicolas
In order to determine whether the line we were seeing on ALS TRY was a result of aliasing of something higher frequency in the ADC, we checked the analog spectrum of the ALS TRY photodiode.
Attached photo shows the line at 364 Hz in analog as in digital, so no evidence for aliasing.
Sheila, Alexa, Nic, Evan
Sheila and I went down to EY to see if we could track this down.
This 365 Hz line appears to be caused by oscillation in the 2" PZT mirror on ISCTEY. If the green refl beam clips on the RFPD, this produces amplitude modulation that leaks into the PDH error signal.
Line is obvious on ALSY QPDs on the transmon.
It seems the oscillation is mostly yaw (364Hz), though there is a pitch line about 10x smaller (420Hz), as well as lines in ALSX QPDs (pit: 389Hz yaw: 363Hz).
J. Kissel Continuing to compare L1 vs. H1 damping loop designs (see ), I've made a comparison of the QUAD's design, and also with the original low-noise design from which they're derived (see LHO aLOG 6760). Attached are the H1 and L1 designs compared against each other, dampingfilters_comparison_2014-11-07_LHOvs2014-11-10_LLO.pdf, as well as both against the original low-noise design dampingfilters_comparison_2014-11-07_LHOvs2013-06-14vs2014-11-10_LLO.pdf. As Sheila's quick comparison pointed out (see LHO aLOG 14923), the differences aren't that large (and detailed below). In summary, both LHO and LLO have modified the low-noise design, but only slightly -- both apparently in order to get a reduction in test-mass impulse response. LLO took the "more sophisticated" in longitudinal by adding a boost instead of just increasing the overall gain, hence their modification preserves some phase margin. Because we're giving the LLO ALS DIFF control loops a try, however, we have copied over and installed the LLO design -- assuming that any little bit of difference will matter for these complicated global control transfer functions. Stay tuned... - T, V, R loops are identical to the original design The differences are the following modicifations: - In L - LLO "resg1" boost ( resgain(f=1.07 [Hz], Q=4, height=13 [dB]) * gain(6, "dB") ) - LHO overall gain increased by 2 (from -1 to -2). - In P - LLO "notch60" 60 [Hz] notch ( notch(f=60 [Hz], Q=50, depth=40 [dB]) ) - LHO overall gain increased by 3 (from -1 to -3). - In Y - LLO "+12db" gain of 3.9811 ( gain(12, "dB") ) Comments: In L, - The extra "resg1" at LLO, significantly reduces the Q of the *second* (~1 [Hz]) L mode, and therefore significantly modifies the L to L and L to P global control transfer functions -- not only the resonant feature at 1 [Hz], but also the zero at 0.75 [Hz]. Other features are moved around a little, but no such significant change in shape. - As such, the displacement from residual ground motion at 1 [Hz] has been reduced by a factor of 3, though (at least from the modeled input seismic motion) it doesn't contribute much to the over all RMS. - The impulse responses of the two designs are comparable. In V, and R, - As expected from the HSTS design studies, the impulse response of these DOFs reveals that without other local damping (say from optical levers), the highest V and R modes (at 9.7 and 13.8 [Hz]) get excited when the PUM and TST are kicked, and continue undamped at their original extremely large Q. - If needed, both the fundamental V (at 0.55 [Hz]) and R (at 0.86 [Hz]) could use an increase in their resonant gains to decrease the 1/e time for those modes, which seem to dominate the amplitude of the time series for the initial 10-20 seconds. V would be much easier than R, as the LUGF phase margin for R is already pretty small. In P, - As with the HSTS, the global control transfer functions are barely affected by the differences in design, though -- to be fair -- the differences in design are much less dramatic than in the HSTS. - Test mass impulse times are comparable. - The first "pitch" mode at 0.51 [Hz] is more damped (by about a factor of ~2) in the LHO model because of the increase in overall gain. However, since the second mode at ~1 [Hz], is more L motion at the test mass than P, increasing the gain at this frequency by a factor of 5 with the "resg1" in the L design seems to have reduced this "P" mode by the same factor at the test mass. Interesting! - The notch at 60 [Hz] has little affect on the loop design -- I suspect this was installed to kill a ground-loop feature in a particularly bad OSEM. In Y, - This is pretty straight forward -- the yaw loop is not particularly limited in phase margin, only in noise re-injection. If we're willing to sacrifice a factor of 4 in noise, cranking up the gain reduces the impulse response by 4, and the Qs of the global transfer functions.
In Y, LLO also has a 60 Hz notch that we have now copied over.
One thing to note: we have left our drive align matrices as they were installed. They are rather different than what LLO has installed; we have not copied any of these over.
I could easily enage the ISS second loop, following the procedure described in 14291. It worked without problems.
However, the loop performances were very poor, as shown in the first plot. The dashed lines corresponds to the second loop open, the solid lines to the second loop closed. The inloop signal is squeezed down to a reasonable level. However, we're getting only a factor 10 of out-of-loop noise reduction. This is much worse than in the past.
I could improve a bit the situation re-centering the beam on the ISS QPD: it was quite far away. The second plot shows that the loop performance is improved, but still worse than what we got in the past. The thrid plot compares in-loop and out-of-loop signals when PD1-4 or PD5-8 are in loop. There is no significant difference, so it's not only one of the two PD sets to be misaligned.
Tomorrow I'll need to play again the picomotr game to recenter the photodiodes. The last centering was performed more than one month ago, on October 3rd, and then there has been no activity around the ISS, as far as I know. Therefore, it's maybe not surprising at all that the alignment is no more optimal. The IMC alignment might have changed.
Today I set down again to improve the working point of the IMC angular controls, ina way similar to waht I did in the past (see here)
The first step was to histogram the band-limited RMS noise of the ISS second loop photodiodes, to measure the dependency of the RIN on the residual angular motion of the IMC. The first attached plot shows the BLRMS between 200 and 1000 Hz, histogrammed against the six IMC degrees of freedom. It is apparent that there are offsets in almost all signals. The offset is particularly evident in DOF_1_Y. Fo future reference, the period considered is between 10.21 and 10.42 local time.
Using these histograms, I finely tuned the following offsets (which are different from what was running before):
DOF 1 | DOF 2 | DOF 3 | |
---|---|---|---|
Pitch | -15 | 15 | 1e-3 |
Yaw | -9 | -5 | 0 |
The second attached figure shows the improved histograms after this tuning. The third plot shows the reduction in the noise as seen by the ISS second loop (sorry, the plot is not calibrated in units of RIN). For future reference, the period considered is between 11.10 and 11.29 local time.
The intensity noise is however still clearly non stationary and the fluctuations seem loosely related to the residual motion of the IMC degrees of freedom.
After some other activities, reported below, I had the impression that the alignment was not more optimal (higher noise in the ISS diodes). I checked again the offsets and it turned out that setting DOF_3 to zero did improve the noise. Therefore I suspect that some of the offsets are not very stable over time.
To track more accurately those possible drifts, I added two lines on the PZT: pitch at 155 Hz (amplitude 3) and yaw at 247 Hz (amplitude 3). My goal is to leave these lines on all night and then use them to reconstruct any variation of the jitter to RIN coupling.
FInally, I injected some broad band noise in the PZT, to measure the transfer functions to the IMC signals. Here are the times
PZT pitch amplitude 30, elliptic band pass 10-550 Hz: from 12.20 to 12.27 local time
PST yaw amplitude 30, elliptic band pass 10-550 Hz: from 12:28 to 12:34 local time
The histograms hinted at the fact that the fluctuations in the RIN were correlated to DOF_1_Y. I measured its open lop gain and the unity gain frequency was about 60 mHz. I increased the gain by a factor 4, thus moving the UGF to a bit more than 200 mHz. See fourth plot. However, this did not produce any significant improvement in the intensity noise. This is consistent with the idea that most of the coupling is due to "fast" motion of the input beam.
Nic, Evan
We went down to EY to see what was up with the green WFS on ISCTEY (D1400241). To start with, we decided to resteer the beams so that they are not clipping on any optics.
We also tightened down PD4, which had some issues with loose mounting.
Somehow, aligning the beam on the Green LSC PD gave us a factor of 10 more power.
Attached, one may see the change in power on the green refl PD at end Y. The first shows just a few hours from today, showing the change in power.
Also attached is 2 weeks of trends, to see if the PD was well aligned in the past, and it was just bad this morning. As far as I can tell, the power from this morning was a common number measured by this diode. There are also times with the power larger, but the value is not stable. I'm not sure if this is when the arm is aligned. I will check once we again align the arm.
The WFS PZT centering servos were measured and tuned up. All DsOF have UGFs that equal to roughly 10Hz.
Upon taking a look at all large suspensions and small suspensions aux channels only ETMY stage L1 was found to have a noise aux channel which is showing some possible issue. ie ~ -4300 cts. This is showing on "noise" (4), volt (UL) and fast_I (LR). I have begun setting up DTT templates for all aux sus channels to further examine questionable readings.
E. Merilh, J. Kissel This investigation concludes that the issue identified in LHO aLOG 14922 is not as pervasive as originally thought. *phew*! The DTT templates to which Ed refers will assess the noise performance of these channels. We'll work on resurrecting Koji's automatic SUS AUX channel characterization suite, found here: /ligo/svncommon/SusSVN/sus/trunk/QUAD/Common/MatlabTools/ezQuadDewhitening This will assess whether each of the monitor boards show the expect driver's driven transfer function after install.