patrick.thomas@LIGO.ORG - posted 09:21, Wednesday 05 March 2014 (10524)
Jax electronics testing at end Y
(Jax, Keita)
Today Keita and I marked out the rough position of the back corners of ISCTEY at BSC10 to facilitate its move to its final location next to the BSC. The nominal position of the table is 18" away from the ALS viewport in the -X direction, with the left edge of the table 22" in the +Y direction from the ALS viewport (lower left viewport on BSC10). The numbers are given in inches because none of the measuring tapes we found had metric markings. Here is the information and calculations used to determine this position.
The beam angles from the TMS were taken from sheet 7 of D0902163. Since this is the layout for EX, we need to consider the mirror image of this layout. Therefore, the horizontal angle is not in the -X direction, but in the -Y direction. The horizontal angle phi is determined from the alignment angles as outlined in the attached pdf.
For the ALS beam angle of 12.4 degrees, the horizontal shift per meter is 22cm. The table is placed approximately half a meter from the viewport, corresponding to a shift of 11 cm/~4.25". The center of the table opening for the ALS beam is 27.5" from the edge of the table enclosure, so add this displacement to 27.5" to determine the shift in position. This gives our layout position of 22" in the +Y direction from the ALS viewport.
The only question we have remaining is the position of the table in the X direction. We have laid out and calculated a position consistent with the table installed at EX, but there is some possibility the mode-matching on the table was set with a distance of 1 m in mind. If the table has to come back half a meter, this would add 11 cm to the horizontal displacement. The clearance on either side of the ALS table after moving the E-module will be 9" (ref. Bubba), so if the position has to be modified to address this, the additional 11 cm/4.25" will be attainable with our available space. Alternatively, the mode-matching telescope on the table can be adjusted to be optimal for this position, which is more consistent with the table as installed at EX.
(Edited because the equations were exported as PDFs instead of PNGs and didn't show up... and AGAIN because despite showing up just fine in the previews, the equations refuse to show up. Attached the work as a pdf.)
Sometime between leaving the cartridge test stand and yesterday inside of BSC10, the optics in the suspension got really dirty. The barrels of the ETM test mass, ERM optic, and PUM have accumulated alot of particulate. I wiped these clean just prior to welding in Jan and recall them looking much better than they do in the pictures below when they were hanging on the test stand. So, between the last bit of cartridge work, the cartridge flight and the in-chamber work over the last week lots of particulate has swirled around the in-vac hardware (about a 2 week duration). Here's the sequence of cleaning in case Calum wants to know:
On test stand, just prior to flight:
- I wiped and vacuumed SUS, bagged both SUS and TMS
- Hugh wiped all of the ISI above, then bagged ISI
Cartridge flight Feb 24th
~2 days after in-chamber I wiped entire floor and viewport surfaces of BSC10 chamber, I did NOT inspect optics
- cabling, TMS payloading, SEI float/balance
- yesterday, I see this gross particulate accumulation, some on the suspension structure too, harder to see since it's metal surfaces though
- I wipe the entire floor of chamber again
More cleaning to follow, obviously.
Note, the particle counter which was hanging around in the test stand area was worked on early last week and then parked under the tube in the chamber cleanroom. This wasn't a particularly useful place to count particles so I moved the counter to in front of the door yesterday. Long story short, no particle data is useful before yesterday. Par.
Done
Done
Alexa, Daniel, Sheila
RED TEAM: We are exciting ETMX, ITMY Pit and Yaw over night.
I have excited ETMX, ITMX Pit and Yaw, L2 test filters at various frequencies. The oplevs move from around .2urad to .5urad. In this configuration the green lock transmission ranges from 760 counts to 400 counts. Increasing the drive caused the arm to lose lock (and we assumed Keita wanted to make his measurement with the arm locked), and caused ITMX suspension to trip. Hopefully, this set up is enough for Keita's measurement. The start time was around: 8:34:52 UTC time. The excitations are being run by four awggui's on opsws7. I have attached the awggui configuration and a picture of the oplev motion on dataviewer.
(Stefan - not Keita)
The excitations were turned off at 5:46 am (13:46 UTC).
We had a lot of trouble early in the evening isolating ETMX, we kept tripping the watchdog on stage 2. We tried reseting the CPS offsets, and eventually were able to get it to isolate with Tcrappy and level 2 controllers, but the motion was rather large (~ten fringes per second). We left stage 2 with damping only.
Were you sending any particularly large locking transients to the ETM, to the top mass in particular? What did the input motion look like? Did it trip on actuators, or some sensor?
We haven't been using feedback to HEPI, and turned off the feedback to the top mass when we started having trouble. there were acutator trips, GS13 trips, and T240 trips (I frogot to disable the T240 watchdog once).
Also, at one point last night I turned on the sensor correction with the blend filters still on Start, this was obviously bad although nothing tripped. Maybe this is only intended to be used with T240s, or only with Tcrappy blends.
Alexa, Daniel, Sheila
We normalized the refl PDH error signal, and saw that we can stay in the linear range of this signal with only ALS COMM locked.
We first normalized LSC-X_TR_A_LF by adding a gain of 0.000154 so that the tranmission peak is at roughly 1. We set LSC-REFLBIAS to REFL_A_RF9_I over the transmitted power (LSC-X_TR_A_LF). The first screen shot is the PDH, the transmitted power, and the normalized error signal as we move the COMM VCO set frequency. In the red trace we moved the frequency by 2kHz; meanwhile, in the blue trace we moved half as fast over 1kHz. We used Alexa's equation from alog 7054 to roughly calibrate the signal: 80Hz cavity pole/150 counts pp of the PDH signal=0.533Hz/count.
We put this gain into the REFL_BIAS filter, we will do a more carefull calculation in the morning. With this calibrated error signal we were able to measure an out of loop spectrum of the noise between the arm cavity and the PSL. Attached are some of the spectra that we took. In the top panel the green trace is a reference with the cleanroom over HAM1 on, in the red trace we turned the cleanroom off and Alexa tried to tune up the arm cavity alignment by maximizing the green transmitted power. This realingment clearly helped at low frequecies, and reduced the coherence with the ETMX PIT oplev.
The RMS we measured is completely dominated by low frequency noise that changed, probably because the alignment was drifting. The RMS down to 0.1 Hz varied from 30Hz-70Hz, we also measured an RMS of 58Hz down to 0.01 Hz.
We saw coherence with the end station PDH reflected signal above 1kHz, coherence with the PIT oplevs (mostly ETMX, which currently has level 2 controlers on Stage 1 and damping only on stage 2) around the PIT resonances (0.4-0.5 Hz), and coherence with YAW OpLevs from 0.04-0.5 Hz. We also looked for coherence with some of Robert's PEM sensors, we have coherence from the periscope on ISCT1 from about 50-100Hz, as we kind of expected.
Tonight the noise is much worse than it was on saturday, the transmitted power was not stable at all over time. This might be because we don't have as good of an alignment, or be because the arm cavity motion is larger because of the problems with ETMX ISI. We also measured the spectrum with higher and lower gains in the COMM board, higher gain did not seem to help, which is different from the situation on saturday alog 10439 .
Did you have the ALS WFS servos running for these measurements?
Note, this same measurement at LLO looks strikingly similar: see LLO aLOG 11265!
No WFS, dither alignment, or longitudnal feedback to the test mass was on for any of these measurements.
When comparing this spectrum to the LLO one has to undo the red cavity pole in the above spectra. It is a little bit more complicated, since we are dividing by the red transmitted power which itself can carry noise. With this correction both spectra are referenced to the input laser frequency and look more or less identical.
When comparing this spectrum to the HIFO-Y spectrum in T1300688, the story at mid frequencies is more or less the same:
We completed the changes in the lsc model to include the ALS-C_REFL_DC_BIAS path. It is split off the input matrix and now has power normalization, an extra LSC filter module (LSC-REFL_BIAS), and triggers. It bypasses the output matrix and directly links to the ALS filter module which is outputed to the DAC. The DAC channel is available at the DB9 breakout panel at the ISC-R4 rack. It is intended to be wired into the CM summing node. Currently, it is wired to the second input of the common mode board. Medm screens have been updated to add the extra matrix row where needed. The new filter modules are only available from the ALS_CORNER overview screen. Model and medm screens have been submitted to svn.
PLC3 on h1ecatx1 can't be activated and run using the install srcipts. If you open PLC control and try to login it asks you which run time system to use, if you select run time 3 it can login and seems to be running OK.
[Ed, Yuta, Stefan, Arnaud, Sebastien, Jeff, Evan]
Temperature loop
On Sunday, Stefan helped me get a stable temperature loop going for the auxiliary laser PLL. The lasers now stay frequency-locked for about 5 minutes. We take the fast control signal, feed it into an SR560 for gain/rolloff control, then attenuate a bunch, and feed it into another SR560 which sums this signal with a trimpot-controlled DC offset signal. This signal then goes into the slow control of the laser. I suspect the short lock time is due to the fact that the SR560 has no integrating feature; right now I've just set it to have a DC gain; the pole of the temperature loop is set by the thermal pole of the laser. A diagram of the loop topology will be uploaded soon.
PRMI FSR sensing: no success
On Monday, the EE shop made me a 60-ft BNC cable with LMR-195. I used this to take the raw RF signal from REFLAIR_B and bring it over to the IOT2R setup. This signal is demodulated using the PLL offset frequency as the LO, and the resulting DC trace is monitored on a scope.
This afternoon, Yuta and I stole time from the green team in order to lock PRMI and try to see if the demodulated REFLAIR_B signal would show any kind of error signal in response sto the PLL offset being swept across an FSR of the PRC. We swept from 58 MHz to 62 MHz, but did not see a clear DC response from this error signal; the DC was between -1 mV and 0 mV, with an rms noise of a few hundred millivolts. If the offset frequency was set to be near a harmonic of 9.1 MHz, the error signal would become dominated by the beat of the harmonic against the offset frequency (as one would expect).
Next, we tried looking at POPAIR_B_LF to see if we could see a DC power buildup as a function of PLL offset. We didn't see anything; the DC power fluctuated between 70 and 90 counts at all times, and showed no change in response to the PLL offset.
Addendum on PRMI Iocking
Yuta tried for some time this afternoon to get PRMI locked. The biggest stumbling block was that PRY showed no fringes. Eventually, with the help of Arnaud, Sebastien, and Jeff, it was realized that
A schematic of the PLL loop and a diagram of the table are attached.
Jim, Sebastien
We've been seeing some weird issues all day on BSC-ISI ETMX.
After restarting the models on all the BSC-ISI platforms this morning (see post here), we realized that the horizontal (X, Y, RZ) lvl3 controllers on ST1 were all equal to an empty zpk with a gain of zero
zpk([],[],0.000000000000,"n")
First mystery.
We successfully reinstalled those filters using the seismic scripts to do so (autoquack function).
Second mystery: when we turned the controllers lvl3 on, we noticed a huge drive on the horizontal DOFs, especially on ST1-H3 (~15000 counts according to the overview MEDM screen).
After some investigation, we can see some high frequency features in the closed cloop power spectra, especially at 718.875Hz (see plot attached).
This features show up on all the axis, especially in X (which explains the huge drive on H3), and only with the lvl3 controllers (looking at these controllers in foton, they actually look fine...).
My guess so far is that there is something wrong with our foton file. Arnaud is taking some measurement right now, so I'll continue the investigation later.
Lvl2 controllers are actually engaged on ST1 and ST2 (with Tcrappy blend filters)
The external epics gateway and the vacuum alarm emailer programs were restarted after OS patches were applied and required server reboots.
Because finding the statistical uncertainty in a transfer function data point, as a function of its coherence and how many averages were taken, comes up just about every few years (e.g. the last time I saw it was in the LLO eLOG), here it is quoted in the aLOG for the next generation: Relative uncertainty in transfer function magnitude: d|TF| 1 - C ----- = sqrt ( ------- ) |TF| 2 C N where C is the coherence, and N is the number of averages. Absolute uncertainty in transfer function phase: 1 - C d <(TF) = sqrt ( ------- ) [rad] 2 C N where C is the coherence, and N is the number of averages, and is in units of radians. Ref: Bendat and Piersol, "Random Data" 2nd Ed, p317.
I exited the LVEA about 1635 pst. I've got the two Actuators on Corner3 (SE) connected. Three corners to go. HEPI is unlocked.
