I've attached a flowmeter to the vent from the CP1 dewar. I'll be using this to monitor the exhaust flow for the next couple of weeks. One concern is this may ice up and block the flow, I'll be keeping an eye on it to make sure this doesn't happen. There is an overpressure vent upstream of the flowmeter as well.
[by everyone involved in IMC commissioning]
This post explains how the two IMC-X and IMC-F channels are generated and calibrated. As of the date of this post, this has been made uniform between the two sites.
NOTES:
- here the term "IMC length" is used to indicate the round trip length (~32 m). Because the IMC is a very squeezed triangle, there is sometimes a tendency to refer to half of this value (~16 m) as "the length of the IMC" (assimilating it to a linear cavity); this has been a source of misunderstanding in the past. Here, the change in half-round trip length is called IMC-X and is equivalent to the displacement of the MC2_M3 optic.
- there may or may not be other filters present in the filter banks cited below. This additional filters are temporary, for testing purpose, or leftovers from old tests, and are supposed to be disabled. Only the filters explicitly mentioned in this post should be engaged for the calibration to work.
********************************************************************************************************
IMC-X (DAQ channel IMC-X_DQ)
********************************************************************************************************
The displacement of MC2_M3 is read by the OSEMs and recorded as MC2_M3_L_OUT_DQ; however, the sensitivity of the OSEMs is not sufficient to measure the motion of the IMC when locked, so IMC_X needs to be estimated from the lock feedback signal sent to the various MC2 stages. Note that this is independent from how the lock signal is filtered and distributed to the various stages of the suspension.
M1:
M1_LOCK_L goes into the IMC_X_M1 filter bank, where the following filters are enabled:
"dc_cal": converts counts into displacement at DC (thus including counts->force calibration and the dc value of the force->displacement TF for M1). LLO measured value: 8.2e-13. LHO is using the same value.
"white": whitening filter to compensate for rounding errors of the digital system. Double zero at 0.2 Hz, double pole at 1 kHz, dc gain = 1.
"wresg1" and "wresg2": transfer function from "force on M1" to "displacement on M3", assuming the resonant gain damping filter is enabled (as it should be), and normalized to 1 at dc (the dc gain is absorbed in "dc_cal"). Two filters are necessary because the expression is too complex to be represented by a single foton filter.
"to_um": converts the channel into um for convenience.
M2:
Same as M1 (appropriately replacing M1 with M2), except:
"dc_cal" value is 1.75e-11 (measured at LLO, copied at LHO)
"wresg1" and "wresg2" are replaced by a single "wresg" filter, as the expression is simple enough that it doesn't need to be split.
M3:
Same as M2 (appropriately replacing M2 with M3), except:
"dc_cal" value is 1.75e-9 (measured at LLO, copied at LHO)
IMC-X is obtained by summing the outputs of these three filter banks (IMC_X_M1/M2/M3).
NOTES:
- it is calibrated in um
- remember that it needs to be "de-whitened"
********************************************************************************************************
IMC-F (DAQ channel IMC-F_OUT_DQ)
********************************************************************************************************
The "fast output" of the common mode board used for IMC locking is digitized and sent into the IMC-F filter bank, where the following filters are enabled:
"cts2V": to convert ADC counts into volts. Both sites use the value of 0.000610016 V/cts.
"InvGenFilt": invert common mode board "generic filter". It is a double pole at 10 Hz, double zero at 100 Hz, 0.5 DC gain (to account for differential output fo the board).
"VCO": control signal --> frequency shift calibration; it accounts for the VCO gain, an internal filter and the double pass into the AOM. Zero at 40 Hz, pole at 1.6 Hz. DC gain (2*measured VCO gain) is 496900 V/Hz at LLO, and 536604 V/Hz at LHO.
"tokHz": converts the output in kHz for convenience.
To be "calibrated" in meters and plotted together with IMC-X (MC2 displacement = delta(IMC length)/2), IMC-F needs to be multiplied by 5.85e-14 m/Hz, according to the formula:
delta(IMC-X) = delta(IMC-F)*lambda/(2*FSR)
It turned out that the picomotor controller box #3 (S1107569) was healthy.
We thought this wasn't correctly functioning [1], but it actually works good.
This means that the slow digital system which has been controlling the box is more suspicious at this point.
We knew that this controller box was not correctly working when it was controlled from the slow digital system (see the details in [1]). This can mean that either/both the box is not functioning or/and the slow digital system is not. Today I removed the box from the top of IOT2L and tested the box itself without the slow digital system by controlling it with the tester board. With this setup I drove a picomotor which was hooked up to the front panel in order to see if the box can drive it with no problems. I tested all eight channels by sequentially connecting the picomotor to each of them. All of them worked without any problems : the drive speed seems correct and both CW and CCW rotation can be performed as designed. Both x and y rotation made sense. The test was done at 500 Hz most of the times. Also I tested channel 1 in a frequency range from 100 Hz to 500 Hz by changing the drive frequency and the rotational speed still seemed (in fact even sounded) healthy to me. I concluded that this box was healthy and put it back on the top of the IOT2L enclosure.
We will check both hardware and software of the slow digital system at the H1 electronics room.
[1] LHO alog 5938 Comments on " HAM3 ISC IR QPDs are likely broken, in-vac cabling error, IO QPD should be OK"
Attached are plots of dust counts requested from 5 PM March 31 to 5 PM April 1. A big thank you to Jim B. for his help with this!
Following Keita's aLog 5843 describing the issue we have with our EPICS channel names -iop model vs user model DAC output-, I started modifying the macros defining those channels (following the method he used for TMSy)
The text file attached describes the values asssigned to the arguments of the sitemap for each suspension of SUSHAM. Since there is a 256 character limit in the medm editor, we will have to wait until that limit is raised to 2048, as it has been done at LLO, to implement the changes.
[...] to be continued
Kyle, Gerardo In preparation for the pending power outage we have been experimenting with ways to maintain minimal flow in the LN2 transfer piping so as to prevent the piping from warming up. Manually filling the 80K pumps is much quicker when the transfer piping is kept at nominal temperature. Our first approach of cracking open the LLCV bypass valves proved to be to "coarse" and unrepeatable. Today we tried a different method by using shims to block open the LLCV valves to near their nominal % open positions and then valving-out the instrument air which prevents CDS control. Each LLCV requires a unique shim combination and we tried to get the best approximation as we could today. ~1600 Valved-in the instrument air and reestablished CDS control until Wednesday morning. Pump levels will be all over the place this week but are being monitored regularly.
0815 Hertz rental delivering generators
0900 Sheila transitions End Y to laser SAFE
0911 Corey and Kiwamu working on QPDs in LVEA
0920 Michael R de-energizing PSL and computers in H1 PSL enclosure
1000 Pablo working on PCAL periscope in H2 enclosure
1052 Apollo putting doors on HAM2
1110 Kyle and Gerardo shimming cryopump valves (see WP#3794)
1130 Gregorio picking up visitor at airport in Pasco
1342 Dave, Cyrus and Jim making pilgrimage to mid stations--new work stations installed
1420 Cryopump alarms related to Kyle and Gerardo
1550 MidY airhandler alarms---Ski notified.
Keita, Kiwamu, Sheila Keita wanted to check this afternoon that the TMS alignment offsets are small enough that we will have enough range left after pumping down to find the beam on the ITM. We found the alignment offsets that Chris, Kiwamu and I had for retro-reflecting the beam after Kiwamu rebalanced the table on March 15th. It seems as though these offsets were overwritten, maybe when something got restarted, but I just made a new safe.snap for h1sustmsy with the values found on the 15th: H1:SUS-TMSY_M1_OPTICALIGN_P_OFFSET=-14000 H1:SUS-TMSY_M1_OPTICALIGN_Y_OFFSET=5500 There is also gain in the EUL2OSEM matrix, a factor of 4.17 for yaw and 2.8 for pitch. Since it is an 18 bit dac, we are using up 30% of the range in pitch, and 17% in yaw.
BSC6 ISI unlocked, per Vincent's request.
[Corey, Kiwamu]
We replaced the two broken infrared QPDs that were in the POP QPD sled at HAM3 with the new ones.
After the replacement we confirmed that the new QPDs works fine. From the ISC point of view HAM3 is ready for pitting the doors.
QPD swapping
The broken QPDs (SN#39 in QPD1 and SN#53 in QPD2) [1] were replaced with the spares. The list below shows which QPD mount has which diode after the replacement.
QPD1 = SN#20 (the one close to the west door)
QPD2 = SN#24 (the one close to the north door)
We accessed the HAM3 table from the west and north side. We followed the same in-situ replacement technique as we did at EY [2] so that we don't have to remove the whole thing but we just swap only their QPDs. The replacement operation went quite smooth and was much easier than the last time when we performed it at EY where the optical table was suspended and freely swung. After the replacement we tested the QPDs by three ways : photo-current check with a hand-held QPD tester, diode voltage drop test and signal chain test with dataviewer. The QPDs signal made sense when we used the hand-held QPD while the QPD was illuminated with a visible red laser pointer. The voltage drop across each segment was approximately 0.36 V which seemed healthy. We were able to see signals at the IOP of the ASC front end with dataviewer although the signals were spiky since there was no anti-whitening filter for them. (In fact there was even no ASC front end model which takes care of these signals). From these tests we are confident that the new QPDs work fine and conclude that this mission is done.
QPD inspection
After the in-vac work we inspected the broken QPDs at the EE shop by inspecting the appearance and electrical characteristic. Filiberto looked at the QPDs with a magnifier and noticed that some of the wires were physically cut. The picture below shows a front view of one of the two broken QPDs. You can see that two wires, which are pointed by blue arrows in the picture, are cut. This is probably due to too much voltage (current) going through them when we were accidentally sending the picomotor driving signal which was more than 100 V due to the cabling error [1].
According to the electrical tests with a multi-meter, segment 2 and 4 of both QPDs are completely open as expected from the magnifier inspection. The rest of two segments are fine : they show 0.36 V voltage drop and the correct polarity. Note that this broken characteristic is different from that of the one we extracted from EY which was shorted [3].
Some notes
We fully removed the ISC-related tools from the HAM3 table so that we don't have to return to this chamber any more. However there was a thin metal plate left on the table which looked some kind of guard for the suspended mirrors. As we had no idea of what it was we left it in the chamber.
[1] LHO alog 5937 "HAM3 ISC IR QPDs are likely broken, in-vac cabling error, IO QPD should be OK"
[2] LHO alog 5718 "Comments on 'A new PZT mirror installed at ISCTEY and red in-vac QPD swapped on TMSY' "
[3] LHO alog 5942 " Broken infrared QPD of TMSY inspected "
| pin # | signal assingment in QPD cable (D1101624-v3) | signal assignment in picomotor cable (D1101516-v4) |
| 8 | QPD2 S | PM3 Y SIGNAL |
| 9 | QPD2 A2 | PM3 X SIGNAL |
| 10 | QPD2 A4 | PM2 Y SIGNAL |
| 11 | QPD1 S | PM2 X SIGNAL |
| 12 | QPD1 A2 | PM1 Y SIGNAL |
| 13 | QPD1 A4 | PM1 X SIGNAL |
| 20 | QPD2 C | PM3 Y RETURN |
| 21 | QPD2 A1 | PM2 X RETURN |
| 22 | QPD2 A3 | PM2 Y RETURN |
| 23 | QPD1 C | PM2 X RETURN |
| 24 | QPD1 A1 | PM1 Y RETURN |
| 25 | QPD1 A3 | PM1 X RETURN |
It looks that the reason why we had only segment 2 and 4 broken is due to the pin assignment.
In the above table you can see that the first half of the pins (of the DB25 cable) are assigned for segment 2 and 4 while all of them is populated by the picomotor SIGNALs in the case of the picomotor cable. On the other hand the 2nd half is shared by segment 1 and 3 where there is only RETURNs in the picomotor. When we looked at the picomotor signals from the control box using a breakout board the RETURNs didn't show any signals and hence I believe they behave more like ground in this setup. I guess we sent too much current on only these SIGNAL lines, resulting in melt of the wire in segment 2 and 4 while segment 1 and 3 survived because they are connected to RETURNs.
The test of the fifth BSC-ISI assembly is now complete. The testing report can be found at E1100298-v1.
MitchR, JimW, w/ Apollo Chris and Zach
The BSC2 ISI was locked and covered this morning in preparation for the Wed-Thurs power outage. The cover consists of one new BSC Dome Tall cover, a layer of ameristat taped down to the walking plates and a final BSC Dome Cover. Travis is in putting a cover on the locked SUS now. We might be going in again later, to lock HEPI, but otherwise the chamber is ready to be closed up.
The beamsplitter now has a C3 cover. From a SUS perspective, the chamber is ready to be closed.
In preparation for the power outage Wednesday, the PSL and its relevant components have been turned off. Here is a full list.
For the LVEA racks:
For the diode room:
The HEPAs and AC units are still running in the laser and ante room, but will probably have to be turned off before the outage. The reason for this is that the AC units often need a manual restart, and if the software comes up after the power outage with the fans on, the room may overheat.
I transitioned the LVEA to laser hazard and looked at the beam in the IMC, working to understand the yaw needed on MC1 and MC3. I reinstalled an iris (a 4th one I had made that was "extra") in front of MC1. The iris I moved to MC1 is the one that IO had used in front of IM2, and then on REFL, and was not installed on any beam when I started. The 4 irises we now have in HAM2 are: - in front of MC1 - the extra one that has been floating around the table, that I moved to MC1 today - in front of MC3 - installed after the last vent per the dog clamps I left on the table marking it's position - behind MC3 (on MC trans) and before IM1 - installed after the last vent per the dog clamp I left on the table marking it's position - downstream of PRM - installed last week by IO on the PRM trans beam The only iris we have in HAM3 is at the edge of the table in front of PR2. This was installed last week by IO on the PRM trans beam. I transitioned the LVEA back to laser safe.
Richard cleaned the fiber this morning at the Y end patch panel, and after that we had 18-19uW at the table feed-through and I measured 14uW with the power meter in air on the table. After realigning for the swapped beamsplitter, (and re-aligning the power monitor PD) the beatnote was -38dBm, around -20 dBm after the amplifier. A plot of the OLG will follow, but the main message is that the ALS PLL is working. It has been locked for almost hours now, based on looking at the striptool of the beat frequency and the control signals. The settings are: cg -15, fg 22, common compensation on, slow ugf set to -0.015Hz and slow pole set to 0.0001 Hz. The beatnote frequency has glitches that are not present in the channel for the VCO frequency, where the signal level is much higher. It might be easier to write an autolocker if we could get rid of those glitches, perhaps by amplifying the signal.
Chris, Max, Sheila- Before Richard cleaned the fibers, Chris and I measured the power in the fibers at a few different places: Input to fiber distribution box in R4: 1.2mW Y end output of fiber distribution box: 75 uW Max's polarization box: 60uW Output at Y end patch panel: 10uW At ALS table feedthrough:10uW on ALS table at fiber output-approx 10 uW Also the OLG of the PLL with the settings above is attached.
I changed the ALSLaser library, so that the polarity between the slow loop and fast loop is correct when the ugf setting for the slow loop is positive. Also, when I came in this morning the PLL was still locked and looked like it had been for the last 10 hours.
[Kyohei W. and Kiwamu I.]
There was a concern that the actuation coefficient of the VCO --- which actuates the PSL laser frequency via the refcav for the IMC locking --- were calibrated wrong and hence we might have kept plotting more-or-less inaccurate noise budget. Indeed the coefficient was underestimated by approximately 34% according to a measurement we did today. Giacomo helped us out by confirming that the coefficient had been indeed miscalibrated [1]. Of course this brings our past noise budget of IMC_F slightly higher, but certainly this is not a major issue.
The measured VCO coefficient is 268302 Hz/V (see the attached figure made by Kyohei). Please use this number hereafter.
A good news is that our measured value agrees with that of Livingston, which is about 246 kHz [2], within 9%. Our measurement was done by applying various DC offset at the input of the VCO box (2-pin lemo) and by looking at the frequency of one of the two outputs with a frequency counter. The input of the VCO box is designed to be a differential input and hence the x-axis of the plot is Vpos-Vneg.
[1] LHO alog #5534 Comments on "IMC noise"
[2] LLO alog #4645 "low range VCO installed"
Here is the data which I forgot to attach.
