IR TRANS of X is shown in attached figure. I like it.
Great result in holidays !
The ALS DARM has been locked stably with current suspensions, and the residual DARM motion has been reduced down to ~2 Hz in RMS. This is almost same level with the PRFPMI CARM linewidth. This would help RSE lock a lot.
- UGF is now 20 Hz. This is much higher than privious ALS DARM loop UGF (<10Hz).
- The crossover frequency between MN loop and TM loop is 0.5 Hz.
- I tried to use the IM stage, but it was difficult due to the RM chain resonance at 2.1 Hz. Actually, RM chain resonance is our pain of the neck. I suffered from that resonance also for angular control. We need to damp it somehow...
Great result in holidays !
Temperature is going up
Because the temp in the corner station is getting cooler, I turned on several heaters in the corner station as
(Corner)
(Delonghi Heaters)
C-HT01:25C/L4 0 -> ~900W) near BS
C-HT02:26C/L4 (0 -> ~900W) near PR2
C-HT03:26C/L5 (~0W) near SR3
(SUNRISE Heaters)
C-HT04:26C/L3 (~600 -> ~900W) near SR-C
C-HT05:26C/L2 (0W) near large door btw B-lab and Pre-room
C-HT06:26C/L2 (0W) near AS table
C-HT07:26C/L2 (0W) near PR3-PR2
C-HT08:26C/L2 (0W) near IXC
Temperature is going up
[Miyoki, Nakano]
Since we tweaked the loop gain of ALS PDH, PLL, and FNC without careful investigation, they often got unstable these days. We went into the mine and check those loops. Also the ALS CARM has been modified. As the result, the ALS DARM signal has been improved by blue to red in the first attached figure.
- I summarized the OLG of each loop in https://gwdoc.icrr.u-tokyo.ac.jp/cgi-bin/private/DocDB/ShowDocument?docid=12016
- I made several templates to check the ALS performance. these are located in /kagra/Dropbox/Measurements/IFO/ALS
- PLL/PLL_ERR_FB.xml is the template to check the PLL performance. It shows the error signal and the feedback signal. I took the reference when the PLL is locked, and any other loop is not.
- PDH/PDH/ALS_PDH_ERR_FB.xml is for the ALS PDH loop. The reference of each signal was when the PDH and the FNC loop has been closed. No ALS_CARM, ALS_DARM, and so on. Ref4,5 is the feedback signal without FNC.
- FNC/FNC_ERR_FB.xml shows the error signal and feedback signals of the FNC loop. The reference is the ones when FNC and PDH loop is locked.
- ALS CARM loop OLG is shown in the attached figure. UGF is 2 kHz with 40degrees of phase margin.
I adjusted SR3 pitch from -4.7 to -3.0.
SR3 that is under ADS control seemed to slip or to be kicked in pitch sometimes (for example ~14:18, ~14:24) due to something.
I found GR Y trans power was low and the higher mode resonance.
Judging from the past oplev data, I found SR3 pitch jump in the past. I adjusted from -6.5 to -4.7 in DC setpoint.
After that, TEM00 was recovered and the alignment was also recovered automatically.
SR3 that is under ADS control seemed to slip or to be kicked in pitch sometimes (for example ~14:18, ~14:24) due to something.
I adjusted SR3 pitch from -4.7 to -3.0.
Nakano, Lenta
The follow figures are at /users/Measurements/IFO/ALS/archives. And the scripts i used for the plot is /users/Commissiong/scripts.ALS/ALSDARM_TF.py
What to do
- Design filter for ALS DARM Loop
First, in order to check the structures, measured TFs of MN and TM(fig 1). judging from this figure, design some filters to weak some peaks so that they would not be unity gain. Fig 2 is the result that measured OLTFs with created filters. These UGF are about 0.8 Hz
- Check Cross over
At the frequency of cross-over of these TFs, we checked if the phse was not 180 degree relative to each other. Fig 3 is the ration between MN and TM OLTF. Blue line means designed one, Orange means measured one.
You can see how the phase is when this line intersects with 0 dB. At cross over frequency = ~1.4 Hz, the phase is ~ 135 degrees. Also, Above the frequency, the gain seems to have enough margin. So this loop seems to be fine.
- Measure ALS DARM OLTF
It was confirmed that the loop was stabe, we closed the loop and measured the overall open loop TF. Fig 4 is the results. Bkue lines is designed OLTF, Orange is the TF of MN and TM , added together in python, Green is measured one.
There is not much difference between the model,and the actual measurements. The UGF is about 0.8 Hz. concern that 1.5Hz Length peak is close to 0dB.
Nakano, Kenta
We applied the DRIVEALIGN filters for TM L2P and L2Y as the same procedure of MN. Here are the results.
As you can see, these spectra got lower and these don't have coherence especially in low frequency region. On the other hands, there are some peaks of Pitch or Yaw. But These are damped when the suspencions are ALIGN state. we don't have to take care them.
Drive align filter has been implemented. The procedure is as follows:
- Measure the L2P/L2Y coupling without DRIVEALIGN. I measured the TF from the MNL to OLP and OLY.
- Roughly fit the TF as shown in the attachd figure. I did not care about Q factor finely. It is difficult to know how much it is, and it changes by the condition of the control, so it is not really usefull to know the exact Q.
- We know which is the resonance of the Y2Y. So remove it from L2P/L2Y coupling TF and use it as the DRIVEALIGN filter.
The result of the decoupling is shown in the attached figure. Red is the L2P./L2Y with the frequency dependent DRIVEALIGN, blue is the one with the frequency independent DRIVEALIGN, and the brown is the one without DRIVEALIGN.
- At whole frequency region, coupling has been improved by DRIVEALIGN. Especially around 1 Hz, the frequency dependent one reduce the L2P around 1 Hz.
Nakano, Kenta
We applied the DRIVEALIGN filters for TM L2P and L2Y as the same procedure of MN. Here are the results.
As you can see, these spectra got lower and these don't have coherence especially in low frequency region. On the other hands, there are some peaks of Pitch or Yaw. But These are damped when the suspencions are ALIGN state. we don't have to take care them.
Several point of this loop is:
- The residual motion is < 20 nrad for all mirrors.
- The low UGF is due to the low S/N ratio of GRQPD signals. This ASC loop coexists with the oplev loop, and easy to switch. Once we got good ASC signal e.g. WFS, we can turn the oplev loop of.
- ADS loop is just for the initial alignment. It is not urgent task.
- Oplev signal (attached in Kenta's report) does not seem to be suppressed. However, the TRANS power stability has been improved by this loop as shown in the attached plot. The red and blue curve is the GR TRANS power spectra with and without QPD loop respectively. This indicates that the oplev DC signal doesn't show purely the mirror motion, but the ground motion.
Nakano, Kenta
We modified and closed GRQPD DC control loops for all mirrors of both arms in PIT and YAW directions. The UGF of PIT is about 0.3 Hz, The one of YAW is about 0.2 Hz.
Todo
Now, Gurdian can't turn on ADS for ETM . So I modified Guardian to do it. And then, We can request Gurdian to ARM ASC ON
Several point of this loop is:
- The residual motion is < 20 nrad for all mirrors.
- The low UGF is due to the low S/N ratio of GRQPD signals. This ASC loop coexists with the oplev loop, and easy to switch. Once we got good ASC signal e.g. WFS, we can turn the oplev loop of.
- ADS loop is just for the initial alignment. It is not urgent task.
- Oplev signal (attached in Kenta's report) does not seem to be suppressed. However, the TRANS power stability has been improved by this loop as shown in the attached plot. The red and blue curve is the GR TRANS power spectra with and without QPD loop respectively. This indicates that the oplev DC signal doesn't show purely the mirror motion, but the ground motion.
[Shimode, Yokozawa]
We check the potential difference at SRM rack and PR2 rack.
The result will be appeared soon
After Miyoki-san's swift action in reducing the temperature (enrtry 15043), the amount of feedback signal in GAS filters has been incresing steadily.
Unless the actuation grows too rapidly, I'll wait until it becomes stable to try to offload with fishing rods again.
The temp in the corner station decreased by about 0.6C due to SR-C cooler operation.
BTW, the only SR-C operation near the OMC chamber could affect enough on the temp around MCE far away from OMC. FFU operation seem to contribute a good air circulator.
Nakano, Kenta
By rebooting Oplev control loop, we remade GR QPD DC control loop of ITMY and ETMY in Pitch direction. The U.G.Fs of both loops are about 0.1 Hz.
Tomorrow, remake the same loop of X-arm mirrors.
[Kenta, Masayuki]
Now, all of the suspension has strong OL loop, and the ALS DARM does not have any difference from the privious oplev control.
As shown in the second spectra, the RMS is below 10 nrad. The YAW motion is still large below 1 Hz. This is because the TM oplev is not used for the YAW control, but only MN oplev. We should use the sensor which monitoring TM motion to reduce the yaw motion below 1 Hz, so we are trying to use GR QPD to reduce these frequencies.
BTW, above 3 Hz, the ALS DARM signal seems to be limitted by the GR RFPD dark noise used for GR-PDH loop. If we can increase the modulation index, the ALS performance would be improved.
Maybe did you also check the Yarm pitch treatment would not degrade the ALS DARM as you did in 15036, right??
> several nrad RMS
Majide?