The license file for Matlab (Ubuntu) has been updated on the DAQ test stand, Matlab should work properly.
I looked at past data from two weeks ago in which we were locking the Y arm without recycling.
Two weeks ago, the Y arm was as lossy as it is now, .
A good data set was found in Nov-21 08:44:55 - Nov-21 8:59:54. According to conlog, this is the last time when we changed the normalization factor for TRY to set it to 1. Here are the values for ASAIR_LF:
Compared with the data from yesterday (alog 15432), which gave a ratio of 0.82 at ASAIR_A, I don't see a big difference. Also, a plot from dataviwer is attached. Note that TRY dropped in the middle of the locking because we set the TRY digital gain such that it becomes 1 (although the one shown in the plot is without the PSL normalization which is 10.
Apart from this past data, I found another good data set from approximately two months ago in Oct-3 23:35:44 - Oct-4 02:34:44 when Sheila and Alexa were working on the Y arm IR WFS (alog 14296). However. I am having difficulty in getting the relevant data in second trend or full rate for some reason. I could get minute trend, but they were not helpful.
I managed to find the missing data frames on the Caltech cluster. The results for a lock on 3 Oct 2014 at around 23:18 UTC (1096413500) are very much the same:
H1:LSC-ASAIR_A_LF_OUT_DQ
locked = 1057.0
unlocked = 1226.4
ratio = 0.86
H1:LSC-ASAIR_B_LF_OUT_DQ
locked = 303.2
unlocked = 357.5
ratio = 0.85
A plot of the channels is attached (the x-axis is seconds). So, it seems the y-arm losses have not significantly changed in the past two months.
I briefly ran an Optickle simulation to see if we can explain the observed low recycling gain by the 750 ppm Y arm intracavity loss.
According to the result, I am concluding that the loss is the cause of the current low recycling situation.
(Some details)
I set loss of ETMY to 750 ppm such that it does not ruin the SB build up in the DRMI. For the X arm, I assumed a round trip loss of 60 ppm for now. Then I ran the same code as the one I did the day before yesterday (alog 15389).
According to the resultant plots (see the attached), both TRX and TRY went up to only 166 which is close to what we observed yesterday i.e. 140-ish. This corresponds to a recycling gain of roughly 5. Even with this amount of loss, REFLDC dropped to 35% which does not contradict what we observed i.e. 50-ish%. Due to the low TRX and TRY, the CARM signal synthesized by sqrt(TRY +TRY) can be a good linear signal only up to an offset of 15 or so at which point sqrt(TRX+TRY) starts loosing the sensitivity because we are running into the plateau region of TRX and TRY. Indeed we needed to hand the CARM from the sqrt signal to REFL9 at an offset of -15 yesterday.
Interestingly, ASDC does not increase so much between off-resonance and on-resonance points even though we, at the beginning, speculated that differential losses would show up at the dark port as a large contrast defect. So it means that the light at the dark port is still dominated by the 45 MHz SBs which qualitatively agrees with the observed ASDC from yesterday. I am guessing that the reason why the carrier defect is so small is that the power recycling gain is already so low in the first place that the whatever carrier components at the dark also result in a low power.
Sorry, Evan. Fig files again.
Can you please post PDFs of the plots as well?
DCS Room progress;
Here is a photo from today of the interior wall going up. Trenching work is almost complete - backfill is underway.
After talking with Sheila and Kiwamu this afternoon, I've started trying to make the seismic sensor correction a little easier for non-seismic people. I've made 2 scripts, one for on, one for off, called HEPI_Senscor_On.py and, imaginatively, HEPI_Senscor_Off.py. They live in the same folder as Hugo's Test_Configuration scripts: /ligo/svncommon/SeiSVN/seismic/BSC-ISI/H1/Common/Misc/Test_Configuration_Scripts. I will work on a wiki page to make this clearer for the operators, but I think currently the operators should run the off script in the morning if there is going to activity in the VEA's or if we have contractors banging on the ground outside the OSB.
To run, go to the folder and type: ./HEPI_Senscor_On.py or ./HEPI_Senscor_Off.py, as appropriate.
As proposed, WFSA (hard sensor) is fed back to the ITMX and WFSB (soft sensor) to the TMSX.
In the attached, you can see that even when I made a terrible kick such that the transmission dropped below 50%, it eventually came back on its own.
Short lock losses don't kick the WFS out of range, and as of now it is maintaining green transmission between 1.15 and 1.18 for 30 min.
ITMX beam position control (fed back to ETMX) is not implemented yet, but should be easier than WFSs themselves.
Eventually we might be able to change the scheme such that hard is fed back to hard, soft to soft, and then ITM beam position to TMS, but I'm doing it this way as it is easier.
Before doing camera, however, I'll move to Y arm.
The X arm green WFS are running again this afternoon, and they did bring the cavity to a high build up. I reduced the gain from 0.5 to 0.25 because it oscillated when I first turned it on. I've edited the guardian to turn this on and off.
This morning, Travis and I removed the Q7 lower structure from the test stand and parked it in it's storage container. We added the new ground plugs to the 2 extension cables in the box and put the door on it for (hopefully) the final time for a few years.
The upper structure still remains mounted to the test stand, awaiting a storage box to shuffle it over to the near by ISI-storage operation.
We also opened the Q6 Lower Structure box and added the plugs to it's cables. This box should also be good to go for long term storage as soon as we finish torquing the door on this afternoon.
To fix a security issue, SSLv2 and SSLv3 protocols have been disabled for the DAQ test stand. The web server was restarted to implement the change. Various vulnerability test methods show the issue to be fixed. This should affect nobody. The Jenkins page still works.
Pressure at which IP4 steps from 7000V to 5000V now the same for both halves of pump and comparable to neighboring pumps
A comparison between the ISC front-end processors at LLO and LHO.
FE models | LLO typ. | LLO max. | LHO typ. | LHO max. |
---|---|---|---|---|
lsc | 33 | 35 | 33 | 40 |
omc | 11 | 17 | 8 | 11 |
lscaux | 21 | 27 | 17 | 20 |
oaf | 28 | 32 | 25 | 29 |
calcs | 11 | 15 | — | — |
iscex | 27 | 30 | 28 | 31 |
iscey | 24 | 35 | 28 | 31 |
asc | 161 | 163 | 154 | 165 |
ascimc | 34 | 95 | 28 | 33 |
LLO numbers taken from alog 15918.
Daniel beat me to the entry. Here is the data in a file with SUS added.
The contractors for the new DCS room will be working outside causing ground vibration and inside the VPW all day today.
The well pump and water system is back up and running normally.
model restarts logged for Wed 03/Dec/2014
2014_12_03 08:33 h1nds0
2014_12_03 09:38 h1dc0
2014_12_03 09:40 h1fw0
2014_12_03 09:40 h1fw1
2014_12_03 09:40 h1iscex
2014_12_03 09:40 h1nds0
2014_12_03 09:40 h1nds1
2014_12_03 09:41 h1broadcast0
2014_12_03 12:43 h1fw0
2014_12_03 13:32 h1iscex
2014_12_03 13:34 h1broadcast0
2014_12_03 13:34 h1dc0
2014_12_03 13:34 h1fw0
2014_12_03 13:34 h1fw1
2014_12_03 13:34 h1iscey
2014_12_03 13:34 h1nds0
2014_12_03 13:34 h1nds1
no unexpected restarts. Completion of minute trend offloading on fw0/nds0. Completion of code changes for end station ISC models (with associated daq restarts). Conlog frequently changing channels report attached.
Dave O. Elli, Daniel, Kiwamu,
We briefly checked how lossy our arm cavities are by locking the individual arm without recycling.
For the X arm:
ASAIR_A_LF = 1180 cnts when unlocked.
ASAIR_A_LF = 1155 cnts when locked.
For the Y arm:
ASAIR_A_LF = 1180 cnts when unlocked.
ASAIR_A_LF = 970 cnts when locked.
We made a corase estimation of intra cavity loss (or a.k.a round trip loss) for the y arm, which is estimated to be about 750 ppm (!). In the calculation, we did not take a mode-mismatch or RF sidebands into accout. We need a closer look at this arm cavity to see why it is so lossy.
If beam mis-centering on the TMs is the cause of all the extra loss, it will have to be quite a big mis-centering. I did a quick calculation, and to get 750ppm loss at the ETM, the beam has to be offset from the center by about 6cm (roughly one beam size). The attached plots show the intensity spill over for a 62mm radius beam on a 163mm radius coating, with no offset and with a 6cm offset. The proportions of beam power outside the coating, with no offset and 6cm offset, are 0.991ppm and 762ppm respectively.
Just in case someone wants a plot ...
Dave has a calculation which makes some assumptions about mode-mismatch and sideband power for the X arm.
Suppose the power in the sidebands is 6% of the incident beam, and 15% of the incident carrier doesn't enter the cavity because of mode matching issues; i.e., about 20% of the light is nonresonant. Then the equivalent power reflection is (1155 - 236) / (1180 - 236) = 0.974. This gives a loss of 108 ppm in the X arm.
Kiwamu, Sheila, Evan, Lisa, Daniel, Elli, Dave O., Dan, Alexa
On December 3rd 2014, 3:01 UTC we had the IR resonanting in all the cavities. MICH, PRCL, and SRCL were still on the 3f signals; DARM was on AS 45Q and CARM was on REFL 9 I at 0 pm. This is what we did:
Here are the REFL and arm powers for a few values of the CARM offset (controlled via TR_REFL9, the digitized REFL9I error signal). These were taken with tdsavg
and a 5 s average.
It appears that we are indeed sitting at a minimum of REFL LF and a maximum for the arm transmissions.
TR_REFL9 offset (ct) | TRX QPD sum (ct) | TRY QPD sum (ct) | REFL LF (ct) |
---|---|---|---|
−1.5 | 78 | 87 | 53 |
−1.0 | 100 | 112 | 45 |
−0.5 | 118 | 134 | 39 |
0 | 129 | 146 | 36 |
+0.5 | 129 | 146 | 36 |
+1.0 | 119 | 134 | 39 |
+1.5 | 101 | 115 | 45 |
+2 | 80 | 90 | 51 |
+2.5 | 56 | 63 | 60 |
This is a trend of relevant signals during the locking sequence. We are running ASC loops with low bandwidth on BS, PRM and DHARD. The transition to REFL_9I happens at time = 600 sec. The relative increase in the power build up is only 15%. Dan will post more precise numbers, but his first estimate is that we have a total of 280 mW at the AS port, 75% due to sideband power, 25% carrier. POP_DC only increases by a factor of 3 when the carrier is resonant..it should be more like 10-20.
Congratulations! Way to go. Now let's chase that buildup..
Excellent!
Here are the OLTFs taken at 0 pm CARM offset. The CARM loop has turned out kind of weird as it crosses unity, but it seems the phase bubble can support it. We might need to do some retuning.
We rung up the BS roll mode at 25.7 Hz
And we beheld, once again, the stars.
Here are some power calculations for the AS port and the TMS IR QPDs:
==== AS_C sum output (the channel that is used for the shutter logic) ====
quantum efficiency = 0.8, QPD transimpedance = 1000 ohms, sum output gets a factor of 1/4 from R23 (see D1001974) --> 200 volts / watt on the QPD
OM1 transmission = 5%, M6 is 50/50 splitter (see T1200410)--> AS_C gets 2.5% of the light entering HAM6
AS_C sum output is 1.4V when locked at zero CARM offset: 1.4 V * (1/200) * (1/0.025) = 0.280 W into HAM6
==== Transmitted power through the arms ====
Using the IR QPDs on the transmission monitor tables, H1:ASC-X_TR_A_SUM_OUTPUT and similar, and assuming the same QPD transimpedance electronics. (I am not sure of the calibration of the other PDs, like LSC-Y_TR_A_LF.)
These four channels currently have one stage of analog whitening and 18dB of gain --> 7.9x analog gain
Quantum efficiency = 0.8, QPD transimpedance = 1000 ohms, 2x from differential output, 32768 counts per 20V --> 2621440 counts per watt
On the TMS (see T0900385), M4 is 100% reflector for IR (assumed), M12 transmission is 5%, each QPD gets half of this signal --> TR QPDs get 2.5% of light leaving arm
The QPD signals were about 6000 counts (see attached). 6000 counts * (1/2621440) * (1/0.025) * (1/7.9) = 11.5mW leaving the ETMs
From the galaxy optics page the H1 ETMs have 3.6ppm transmission, this implies about 3.2kW in the arms.
The first figure attached is a striptool of the second lock of the evening; the second figure has trends of the TMS QPDs.
That's great, congratulations. LLO had high arm losses (maybe not this high) at one point from poorly centered beams on the test masses; that would be something to check if you haven't already.
Wonderful to see! and well-earned. Congratulations!
Great job, congratulations!