Prior to the pending DAQ upgrade, here are the DAQ stats at the start of ER5.
Frame Look Backs
system |
science_frames |
commissioning_frames |
second_trends |
fw0 | 3 days 22.5 hrs | 19 days 8 hrs | 8 days 8.5 hrs |
fw1 | 1 day 21 hrs | 9 days 4 hrs | 3 days 23 hrs |
Size of frames and data rates:
Science frames are 32 seconds in length, 308MB size, 9.6MB/s data rate
Commissioning frames are 32 seconds in length, 508MB size, 15.9MB/s data rate
10:20 Richard working in H1-PSL enclosure. 11:00 Richard finished in H1-PSL enclosure.
09:00 Betsy in LVEA, gathering parts.
06:30 - 08:15 Sheila working in LVEA.
08:00 Towsen Electrical on LVEA roof working on lightning protection system. 09:30 John, Richard, and Towsen Ele on LVEA Roof
We have 0.77mW on the IR PD before the SHG, with 30dB of gain we get 0.21 Volts. With 2k transimpedance this means the responsivity is 0.0041A/W (this seems too low). We are getting 21mW onto ISCT1 from the PSL, meaning that the BS is 4% reflective, the same value we measured durring HIFO Y.
21 mW is not enough for us to get a decent amount of conversion to green. Right now we are getting about 2.5uW of green out of the SHG, not really enough to optimize the alignment.
Koji, Kiwamu, Stefan - We had some luck locking the xarm today. The best configuration was ITMX ISI State 1, ETMX ISI State 2. However we still had too much arm length motion - slightly more that the 15um that we can tolerate with the 1MHz VCO. - We tried feeding back to the SUS ETMX top stage, but the speed was too high (every now and then it reached ~30um/sec, which is too fast for relieving through the 0.44Hz pendulum resonance). - We haven't tried the HPI feed-back, which is currently not wired. - We tried switching off both ISIs - the fringe velocity was definitely improved, but the alignment was no longer stable enough. - We also did some alignment tweaking with the IAL system in dither-only mode, but we need better arm stability from the seismic isolation for further commissioning.
Koji, Kiwamu with help from Cheryl and JoeD
As a part of the high power preparation, we swapped the beam splitter, IO_MB_M2, from the high transmissive one (T10%) to the low transmissive one (T2%). As expected, this replacement had caused misalignment in both the main and ALS pick off paths. We realigned them. Now the IMC is back to operation and is locking. The ALS beam also successfully comes out to ISCT1.
Preparation of the mount and BS:
We started today from peeling the first contacts off from the BSs. We had two identical BSs, one of which was supposed to be a spare. We removed the first contacts from both BSs and found that one showed many residues on the HR surface. Therefore we decided to use the other one which had much less residues and asked Joe to clean a couple of tiny residue off of it. Also, we asked Joe to apply new first contacts on the one which had many residues so that someone can use it for useful purpose in some future.
Cheryl already had Tyler to modify a PSL-custom-made base mount to nicely accommodate a Ultima mirror holder. So we took it from her desk and mount a left-handed 1 inch Ultima equipped with three actuation knobs so that we can control a translation in addition to the usual two angular degrees of freedom. Note that the old BS which had been in place is a 2 inch BS and that's why we needed to swap not only optic itself but also its mount and base.
Alignment of the main path:
The alignment of the main path went very smooth.
Before doing anything, we placed an iris on IOT2L as a reference. It was put before the length diode. After installing the new BS, we realigned the main path using Cheryl's irises on the PSL table. The important and excellent point is that we touched only this particular BS and didn't touch anything else for this alignment. After centering the beam on the irises on the PSL table, we then took a look at the beam on the table. The beam was already landing on the table and in fact it hit the length diode. So we coarsely centered the beam on the diode by steering the BS. At this point, the alignment was good enough for us to lock IMC. A fine alignment was then done by maximizing the IMC power build up. We could get as high power as it used to be in the transmitted light of IMC and hence we were satisfied with this alignment. After turning the ASC loops on, we found that WFSs' DC centering were excellent. I would say this alignment recovery was very successful.
Alignment of the ALS path:
The alignment of the ALS path was not as easy as the main path.
Because the thickness of the BS changed, the beam spot on the first mirror in the ALS path was translated. The translation was so big that the beam was not making through the faraday any more. We tried to recover the alignment without shifting the position of the Faraday and by touching the first steering mirror, but it turned out that it was quite difficult to do it. So eventually, we decided to shift the Faraday. The first steering mirror was steered so that the beam hits the center of the next steering mirror. Then we moved the Faraday, which is between the first and second mirrors, so that the beam goes through it. The transmissivity of the Faraday was measured to be approximately 94% (= 24.50 mW / 26.03 mW) which is acceptable. The we tweaked the second and third mirrors to steer the beam on an iris on ISCT1. After the tweak was done, we assessed the risk of clipping by introducing a large angle misalignment in the third steering mirror and confirmed that the beam is in the middle of the clipping aperture.
Alexa, Keita, Sheila, Daniel,
Starting with yesterday's alingment alog 9230, I found some flashing this morning using the transmitted PD. This lead to a clearly visible 00 mode on the ETM camera. The final alingments we ended up with are in the attached screenshots.
The trans PD (ALS COMM) has flashes up to 400 counts, which means that we have about 35 uW (a factor of 10 more than the single shot beam). The refl PD drops from about 19000 counts to about 16000 when the cavity is locked to the 00 mode. There is also a 02 mode that has about 200 counts in transmission, so about half the power of the 00 mode.
It seems as though all the previous alingments where we thought we had fringing were red herrings. I moved the camera on ISCT1 (now using a lens and letting the beam hit a baffle) so that the single shot beam is no longer visible but you can clearly see the cavity modes.
Since the locking was so unstable we thought of trying to feedback to HEPI as was done in HIFO-Y. However, the IPC from the SUS to HEPI has been deleted. We either need to add it back or add an IPC directly from the LSC model to HEPI. Hugo is working on it. However, the situation that we have seems worse than in HIFO Y, there we would run out of range after tens of minutes, today we are running out of VCO range every few seconds. Also, we get about 5 fringes per second, which if it is due to motion of the optic would mean 1um per second motion. We need to investigate if the optics are really moving this much.
Alexa and I went to the end station to look at what was saturating (we are exceeding the VCO input range it seems). While we were there we also had a look at the IQ scatter plot on the scope. One thing is that there is approximately the same amount of signal in I as Q, so this is not a usefull way to try to tune the demod phase. Keita came out and we tried to measure the transfer function from the laser frequency to the demodeulated signals, however we could not because the lock was not stable enough. We also tried to adjust the demod phase while watching the open loop TF, this also wasn't sucessfull because the loop did not stay locked long enough for us to get repeatable measurements.
Alexa and I also looked at the residual AM by misaligning the ITM and looking at the output of the PD. With the ITM misalinged, we have -33dBm of residal AM at the input to the demod, with it aligned we get some thing around -17dBm.
Stefan is currently investigating the ISI performance.
We started turning off ISI isolation to see how good or bad the length isolation is, and attached is the comparison of ISI isolation for both ITMX and ETMX ON (dashed lines) VS OFF (solid lines).
On: 2014/01/15 01:03:37 UTC
Off: 2014/01/15 02:31:55 UTC
We haven't looked at the seismometer, but just by looking at the fringe moving around it was obvious that ISI isolation makes the low frequency length motion much, much larger, and both SUS BOSEMS (left bottom) as well as the ISI CPS (left top) confirm this.
This looks too bad to be true, I guess ISI might have been in some bogus state when they were supposed to be in level 1 isolation.
OTOH, when ISI isolation was on the angular motion was smaller, and the oplevs (right, top=ETM, bottom=ITM) show that the problem was mostly in EX at about 0.4-0.5Hz. Again the same caveat (we're not looking at seismometer) applies.
Stefan is still playing with all available ISI isolation levels to in a hope to find a good compromise.
We left the LVEA around 3:30 after chasing GS13 signals. EE/CDS drove our sensor cables at the AA & Interface chasses and confirmed we have signals coming out. We find now that the HI ISI HAM4 model is incorrect according to the wiring (D1000298.) Except for the expected difference regarding CPS on HAM2 & 3, the models match. However, the GS13 & L4C wiring is different. So we have to change the model or change the wiring. At the moment we have just H2, V2 & H3 sensor on the ISI plugged in. We'll continue when the wiring/model gets sorted.
Richard, Filiburto, Hugh, Hugo, Dave
In order to track ISI HAM4 signal issues we injected signals into the AA chassis for h1seih45 for GS13 V1,H1,V3,H3 (first two ADC cards of this IOChassis). Signals were tracked through the IOP and h1isiham4 models. Hugo is checking the model against the wiring diagram.
Keita, Alexa and I have returned from end X
Added an offset zeroing script for QPDs to the svn: asc/common/scripts/setQPDoffsets (SVN revision 6861).
It uses tdsavg to set the four segments to zero, and sets an offset of 4 counts into the sum filter module to avoid a division by zero.
Preconditions: no light on diodes, segment gains have to be -1
Tumbleweed bailing on X arm 10:07 - 10:28 Apollo moving contents of garbing room from West bay to HAM4 (by hand) (WP 4393) 10:54 Site water sample collecting 11:30 Lightning rod work 13:29 Sheila working at end X
12:51 DAQ restart, new channel list for H1EDCU_ECATX1PLC3.ini for Thomas' ring heater changes.
I modified the mod_ini scripts to arrange channels in alphabetical order so we actually got new H1EDCU_ECATXXPLCX.ini files for all systems (new channel order) but only H1EDCU_ECATX1PLC3.ini had new/removed channels.
New ini files committed to svn, version 6859.
Breaking for lunch
(Alexa, Daniel, Sheila)
PLL Beatnote:
PLL Open Loop Transfer Function:
PLL Noise Spectrum -- via IMON of PFD (S1000758)
PLL Servo Board Transfer Function:
Shot Noise of PLL BBPD:
The missing SR785 measurement:
PLL Open Loop Transfer Function:
Input 1 pol: NEG, Common Compensation: ON, Generic Filter: ON, Fast Option: ON, Fast Gain: -8dB, Input 1 Gain: 0db, Boost 1:ON rest OFF
Before:
After:
Some more pictures are available on ResourceSpace: https://ligoimages.mit.edu/?c=1433