40m QIL Cryo_Lab CTN SUS_Lab TCS_Lab OMC_Lab CRIME_Lab FEA ENG_Labs OptContFac Mariner WBEEShop
  40m Log, Page 115 of 341  Not logged in ELOG logo
ID Date Author Type Category Subjectdown
  10944   Tue Jan 27 15:45:26 2015 diegoUpdateLSCSmall tweaks to the locking

The UGF Servo medm page has been updated to reflect the last changes, namely the return of the sum of squares and the disappearance of Test3.

 

  5483   Tue Sep 20 16:31:24 2011 KeikoUpdateIOOSmall modulation depth

 Modulation resonator box is removed and the modulation depth is small right now.

I have broke the BNC connector on the modulation resonator box. The connector was attached by the screw inside very loosely and when we connect and disconnect the BNC cables from outside, extra force was applied to the cable inside and it was broke. It is being fix by Kiwamu and will be back in a bit.

 

 

 

 

  5484   Tue Sep 20 16:38:25 2011 KeikoUpdateIOOSmall modulation depth

Resonator box and the modulations are back now. But the modulation depth seems to be a bit smaller than yesterday, looking at the optical spectrum analyser.

 

Quote:

 Modulation resonator box is removed and the modulation depth is small right now.

I have broke the BNC connector on the modulation resonator box. The connector was attached by the screw inside very loosely and when we connect and disconnect the BNC cables from outside, extra force was applied to the cable inside and it was broke. It is being fix by Kiwamu and will be back in a bit

 

  4844   Mon Jun 20 18:12:20 2011 NicoleUpdateSUSSmall Table Cleaned and Levelled

P6220198.JPG

The small optical bench (next to the MC-2 Chamber and the tool box tower) has been cleared of the misc. object previously on it, cleaned, and leveled (after much calibration X___X).

PLEASE, PLEASE, PLEASE do NOT MOVE OR HIT THE TABLE! It was incredibly painful to level.

This is how leveling the table made me feel...

P6220199.JPG

VERY SAD...so do not move please!

The shaker has already been moved to the table and the amplifier for my shaking experiment is located behind the table (not on the table, as to prevent scratching).

 

 

  15159   Mon Jan 27 18:16:30 2020 gautamConfigurationComputersSluggish megatron?

I've also been noticing that the IMC Autolocker scripts are running rather sluggishly on Megatron recently. Some evidence - on Feb 11 2019, the time between the mcup script starting and finishing is ~10 seconds (I don't post the raw log output here to keep the elog short). However, post upgrade, the mean time is more like ~45-50 seconds. Rana mentioned he didn't install any of the modern LIGO software tools post upgrade, so maybe we are using some ancient EPICS binaries. I suspect the cron job for the burt snapshot is also just timing out due to the high latency in channel access. Rana is doing the software install on the new rossa, and once he verifies things are working, we will try implementing the same solution on megatron. The machine is an old Sun Microsystems one, but the system diagnostics don't signal any CPU timeouts or memory overflows, so I'm thinking the problem is software related...

Quote:

The burt snapshotting is still not so reliable - for whatever reason, the number of snapshot files that actually get written looks random. For example, the 14:19 backup today got all the snaps, but 15:19 did not. There are no obvious red flags in either the cron job logs or the autoburt log files. I also don't see any clues when I run the script in a shell. It'll be good if someone can take a look at this.

  15164   Tue Jan 28 15:39:04 2020 gautamConfigurationComputersSluggish megatron?

There were a bunch of medm processes stalled on megatron (connected with screenshot taking). To see if they were interfering with the other scripts, I killed all of the medm processes, and commented out the line in the crontab that runs the screenshots every 10 mins. Let's see if this improves stability.

  665   Mon Jul 14 00:36:19 2008 JohnSummaryPSLSlow sweep of laser temp - PMC PZT response
John, Rana

Follow up to # 663

Top trace: C1: PSL-PMC_PZT
Middle: C1: PSL-FSS_SLOWDC
Bottom: C1: PSL-PMC_PMCTRANSPD

The only calibration I could find for the PMC PZT (LLO e-log Sep 3 2003 - 23 MHz/V) predicts 31V for an FSR. I did a rough calibration and got our FSR to be around 210 V. I assumed 713 MHz for an FSR and applied this calibration (~3.4 MHz/ V) to the PZT data.

In terms of volts per metre our PZT gives 2.54 nm/ V whereas the LLO PZT is 17.16 nm/ V.
Attachment 1: SlowTsweep.png
SlowTsweep.png
  4201   Tue Jan 25 20:42:46 2011 OsamuUpdateGreen LockingSlow servo for green laser

I implemented a slow servo for green laser thermal control on c1scx.mdl. Ch6,7 of ADC and ch6 of DAC are assigned for this servo as below;

 

Ch6 of ADC: PDH error signal

CH7 of ADC: PZT feedback signal

CH6 of DAC: feedback signal to thermal of green laser

 

Note that old EPICS themal control cable is not hooked anymore.

I made a simple MEDM screen(...medm/c1scx/master/C1SCX_BCX_SLOW.adl) linked from GREEN medm screen (C1GCV.adl) on sitemap.

During this work, I noticed that some of the epics switch is not recovered by autoburt. What I noticed is filter switch of SUSPOS, SUSPIT, SUSYAW, SDSEN, and all coil output for ETMX.

I had no idea to fix them, probably Joe knows. I guess other suspensitons has the same problems.

  4202   Tue Jan 25 21:57:59 2011 KojiUpdateGreen LockingSlow servo for green laser

1. The dewhitening filter CH6 had no output. I disconnected the cable and put it to the monitor out of the AI filter.
So the dewhitening is not in the loop.

2. I have made a thermal control filter

BANK1: pole 0Hz, zero 1mHz / LF boost stage
BANK2: pole 1mHz, zero 30mHz / LPF stage
BANK3: pole 1Hz, zero 0.1Hz / phase compensation stage
Gain: 0.05

It seems working with the gain of 0.05. As the thermal is very strong, the output has less than 10.
This means the we are effectively only using ~4bit. We need external filter.

Note that output of 30000counts were about 3V at  CH6.

3. Measured End PZT feedback with and without the thermal control. The UGF seems to be 0.2Hz.
The suppression at 10mHz is ~100. This is so far OK.

Quote:

I implemented a slow servo for green laser thermal control on c1scx.mdl. Ch6,7 of ADC and ch6 of DAC are assigned for this servo as below;

 

Ch6 of ADC: PDH error signal

CH7 of ADC: PZT feedback signal

CH6 of DAC: feedback signal to thermal of green laser

 

Note that old EPICS themal control cable is not hooked anymore.

I made a simple MEDM screen(...medm/c1scx/master/C1SCX_BCX_SLOW.adl) linked from GREEN medm screen (C1GCV.adl) on sitemap.

During this work, I noticed that some of the epics switch is not recovered by autoburt. What I noticed is filter switch of SUSPOS, SUSPIT, SUSYAW, SDSEN, and all coil output for ETMX.

I had no idea to fix them, probably Joe knows. I guess other suspensitons has the same problems.

 

Attachment 1: 110125_Xend_thermal.pdf
110125_Xend_thermal.pdf
  11754   Wed Nov 11 22:50:39 2015 KojiConfigurationCDSSlow machine time&date

I was gazing at the log file for Autolocker script (/opt/rtcds/caltech/c1/scripts/MC/logs/AutoLocker.log )
and found quite old time stamps. e.g.

Old : C1:IOO-MC_VCO_GAIN             1991-08-08 14:36:28.889032 -3
New : C1:IOO-MC_VCO_GAIN             1991-08-08 14:36:36.705699 18
Old : C1:PSL-FSS_FASTGAIN            1991-08-09 19:05:39.972376 14
New : C1:PSL-FSS_FASTGAIN            1991-08-09 19:05:44.939043 18

It was found that the date/time setting of some of the slow machines (at least c1psl and c1iool0) is not correct.
I could not figure out how to fix it.

Question: Is this anything critical?

Another thing: While I was in c1iool0 I frequently saw the message like

c1iool0 > 0xc461f0 (CA event): Events lost, discard count was 514

Is this anything related to EPICS Freeze?

controls@nodus|~ > telnet c1psl.martian
Trying 192.168.113.53...
Connected to c1psl.martian.
Escape character is '^]'.
c1psl > date
Aug 09, 1991 19:13:26.439024274
value = 32 = 0x20 = ' '
c1psl >
telnet> q
Connection closed.

controls@nodus|~ > telnet c1iool0.martian
Trying 192.168.113.57...
Connected to c1iool0.martian.
Escape character is '^]'.

c1iool0 > date
Aug 08, 1991 14:44:39.755679528
value = 32 = 0x20 = ' '
c1iool0 > 0xc461f0 (CA event): Events lost, discard count was 514
Change MC VCO gain to -3.
0xc461f0 (CA event): Events lost, discard count was 423
Change MC VCO gain to 18.
Change boost gain to 1.
Change boost gain to 2.

 

  12651   Wed Nov 30 14:54:01 2016 JohannesUpdateCDSSlow machine replacement

I was talking with Larry yesterday, and he suggested the rack-mounted supermicro machines SYS-5017A-EP (~$400) or SYS-5018A-FTN4 (~$600) that he uses for moving data around in LIGO. They have 2 gigabit ethernet ports and can thus function as modbus gateways, conveniently placed in the rack close to the slow DAQ/DIO chassis and running some local ubuntu or other distro (I think Aidan uses CentOS in the PSL lab). These only have atom processors, which would be sufficient for the slow machine replacement, but there are many more powerful models with sometimes subtle differences. If we motion towards a more complete GigECam coverage in the lab it could be better to kill two birds with one stone and get something a little faster that can do the video capture/processing, since these machines will be distributed more or less strategically around the lab. Just a thought, as I have currently no clear idea what resources are required for this or how much we're throwing at this GigECam upgrade.

 

Quote:

I've attached a schematic for how we will connect the Acromag mosules to the slow channel I/O curently going to c1auxex. The following changes are made:

  • We are getting rid of the slow readbacks from the Anti-Image and Oplev boards, as Rana says they are unnnecessary.
  • The whitening switching for the QPD is currently done by a Contec "fast" binary I/O module, but can be managed by acromag instead. This alllows CAB_1Y9_34 to  be fed directly into the Acromag box since all of its connections can now be managed slow. 
  • There's no need to change the PD whitening scheme around (since the signals never get huge), so we can set those to always be on and then lose those Contec channels. This means all of the necessary pins on CAB_1Y9_10 can go to Acromag. 
  • All the other backplane cables go the the fast machines only. 

 

 

  8957   Thu Aug 1 21:28:09 2013 gautamUpdateCDSSlow channels set-up in ALS

The following slow channels have been added and are now being recorded by FB.

 

C1:ALS-X_OVEN_TEMP

C1:ALS-Y_OVEN_TEMP

C1:ALS-BEATX_FREQ

C1:ALS-BEATY_FREQ


Details:

In order to integrate the data collected by the Raspberry-Pi from the Y-end doubling oven temperature controller and also the data from the frequency counter which will be hooked up to monitor the beat frequency, Koji helped me set up some slow EPICS record channels (in ALS as we felt this was most appropriate). The procedure for setting up slow channels was as follows (virtually identical to what is detailed in this elog:

  1. Add the channel names to the file C0EDCU.ini (path = /opt/rtcds/caltech/c1/chans/daq/C0EDCU.ini).
  2. Make a database (.db) file so that these channels are actually recorded (path = /cvs/cds/caltech/target/c1aux/als.db).
  3. Restart framebuilder. 
  4. Verify that the channels indeed exist and can be read and written to using ezcaread and ezcawrite.

I will now integrate these channels into my scripts, and make some simple MEDM screens.

 

  13185   Thu Aug 10 14:25:52 2017 gautamUpdateCDSSlow EPICS channels -> Frames re-enabled

I went into /opt/rtcds/caltech/c1/target/daqd, opened the master file, and uncommented the line with C0EDCU.ini (this is the file in which all the slow machine channels are defined). So now I am able to access, for example, the c1vac1 channels.

The location of the master file is no longer in /opt/rtcds/caltech/c1/target/fb, but is in the above mentioned directory instead. This is part of the new daqd paradigm in which separate processes are handling the data transfer between FEs and FB, and the actual frame-writing. Jamie will explain this more when he summarizes the CDS revamp.

It looks like trend data is also available for these newly enabled channels, but thus far, I've only checked second trends. I will update with a more exhaustive check later in the evening.

So, the two major pending problems (that I can think of) are:

  1. Inability to unload models cleanly
  2. Inability of dataviewer (and cdsutils) to open testpoints.

Apart from this, dataviewer frequently hangs on Donatella at startup. I used ipcs -a | grep 0x | awk '{printf( "-Q %s ", $1 )}' | xargs ipcrm to remove all the extra messages in the dataviewer queue.


Restarting the daqd processes on fb1 using Jamie's instructions from earlier in this thread works - but the mx_stream processes do not seem to come back automatically on c1lsc, c1sus and c1ioo (reasons unknown). I've made a copy of the mxstreamrestart.sh script with the new mxstream restart commands, called mxstreamrestart_debian.sh, which lives in /opt/rtcds/caltech/c1/scripts/cds. I've also modified the CDS overview MEDM screen such that the "mxstream restart" calls this modified script. For now, this requires you to enter the controls password for each machine. I don't know what is a secure way to do it otherwise, but I recall not having to do this in the past with the old mxstreamrestart.sh script.

  13189   Fri Aug 11 00:10:03 2017 gautamUpdateCDSSlow EPICS channels -> Frames re-enabled

Seems like something has failed after I did this - full frames are no longer on Aug 10 being written since ~2.30pm PDT. I found out when I tried to download some of the free-swinging MC1 data.

To clarify, I logged into fb1, and ran sudo systemctl restart daqd_*. The only change I made was to uncomment the line quoted below in the master file.

Looking at the log using systemctl, I see the following (I just tried restarting the daqd processes again):

Aug 11 00:00:31 fb1 daqd_fw[16149]: LDASUnexpected::unexpected: Caught unexpected exception      "This is a bug. Please log an LDAS problem report including this message.
Aug 11 00:00:31 fb1 daqd_fw[16149]: daqd_fw: LDASUnexpected.cc:131: static void LDASTools::Error::LDASUnexpected::unexpected(): Assertion `false' failed.
Aug 11 00:00:32 fb1 systemd[1]: daqd_fw.service: main process exited, code=killed, status=6/ABRT
Aug 11 00:00:32 fb1 systemd[1]: Unit daqd_fw.service entered failed state.
Aug 11 00:00:32 fb1 systemd[1]: daqd_fw.service holdoff time over, scheduling restart.
Aug 11 00:00:32 fb1 systemd[1]: Stopping Advanced LIGO RTS daqd frame writer...
Aug 11 00:00:32 fb1 systemd[1]: Starting Advanced LIGO RTS daqd frame writer...
Aug 11 00:00:32 fb1 systemd[1]: daqd_fw.service start request repeated too quickly, refusing to start.
Aug 11 00:00:32 fb1 systemd[1]: Failed to start Advanced LIGO RTS daqd frame writer.
Aug 11 00:00:32 fb1 systemd[1]: Unit daqd_fw.service entered failed state.

Oddly, I am able to access second trends for the same channels from the past which will be useful for the MC1 debugging). Not sure whats going on.


The live data grabbing using cdsutils still seems to be working though - so I've kicked MC1 again, and am grabbing 2 hours of data live on Pianosa.

Quote:

I went into /opt/rtcds/caltech/c1/target/daqd, opened the master file, and uncommented the line with C0EDCU.ini (this is the file in which all the slow machine channels are defined). So now I am able to access, for example, the c1vac1 channels.

The location of the master file is no longer in /opt/rtcds/caltech/c1/target/fb, but is in the above mentioned directory instead. This is part of the new daqd paradigm in which separate processes are handling the data transfer between FEs and FB, and the actual frame-writing. Jamie will explain this more when he summarizes the CDS revamp.

It looks like trend data is also available for these newly enabled channels, but thus far, I've only checked second trends. I will update with a more exhaustive check later in the evening.

So, the two major pending problems (that I can think of) are:

  1. Inability to unload models cleanly
  2. Inability of dataviewer (and cdsutils) to open testpoints.

Apart from this, dataviewer frequently hangs on Donatella at startup. I used ipcs -a | grep 0x | awk '{printf( "-Q %s ", $1 )}' | xargs ipcrm to remove all the extra messages in the dataviewer queue.


Restarting the daqd processes on fb1 using Jamie's instructions from earlier in this thread works - but the mx_stream processes do not seem to come back automatically on c1lsc, c1sus and c1ioo (reasons unknown). I've made a copy of the mxstreamrestart.sh script with the new mxstream restart commands, called mxstreamrestart_debian.sh, which lives in /opt/rtcds/caltech/c1/scripts/cds. I've also modified the CDS overview MEDM screen such that the "mxstream restart" calls this modified script. For now, this requires you to enter the controls password for each machine. I don't know what is a secure way to do it otherwise, but I recall not having to do this in the past with the old mxstreamrestart.sh script.

 

  13192   Fri Aug 11 11:14:24 2017 gautamUpdateCDSSlow EPICS channels -> Frames re-enabled

I commented out the line pertaining to C0EDCU again, now full frames are being written again.

But we no longer have access to the slow EPICS records.

I am not sure what the failure mode is here - In the master file, there is a line that says the EDCU list "*MUST* COME *AFTER* ALL OTHER FAST INI DEFINITIONS" which it does. But there are a bunch of lines that are testpoint lists after this EDCU line. I wonder if that is the problem?

Quote:

Seems like something has failed after I did this - full frames are no longer on Aug 10 being written since ~2.30pm PDT. I found out when I tried to download some of the free-swinging MC1 data.

 

  13197   Fri Aug 11 18:53:35 2017 gautamUpdateCDSSlow EPICS channels -> Frames re-enabled
Quote:

Seems like something has failed after I did this - full frames are no longer on Aug 10 being written since ~2.30pm PDT. I found out when I tried to download some of the free-swinging MC1 data.

To clarify, I logged into fb1, and ran sudo systemctl restart daqd_*. The only change I made was to uncomment the line quoted below in the master file.

Looking at the log using systemctl, I see the following (I just tried restarting the daqd processes again):

Aug 11 00:00:31 fb1 daqd_fw[16149]: LDASUnexpected::unexpected: Caught unexpected exception      "This is a bug. Please log an LDAS problem report including this message.
Aug 11 00:00:31 fb1 daqd_fw[16149]: daqd_fw: LDASUnexpected.cc:131: static void LDASTools::Error::LDASUnexpected::unexpected(): Assertion `false' failed.
Aug 11 00:00:32 fb1 systemd[1]: daqd_fw.service: main process exited, code=killed, status=6/ABRT
Aug 11 00:00:32 fb1 systemd[1]: Unit daqd_fw.service entered failed state.
Aug 11 00:00:32 fb1 systemd[1]: daqd_fw.service holdoff time over, scheduling restart.
Aug 11 00:00:32 fb1 systemd[1]: Stopping Advanced LIGO RTS daqd frame writer...
Aug 11 00:00:32 fb1 systemd[1]: Starting Advanced LIGO RTS daqd frame writer...
Aug 11 00:00:32 fb1 systemd[1]: daqd_fw.service start request repeated too quickly, refusing to start.
Aug 11 00:00:32 fb1 systemd[1]: Failed to start Advanced LIGO RTS daqd frame writer.
Aug 11 00:00:32 fb1 systemd[1]: Unit daqd_fw.service entered failed state.

Oddly, I am able to access second trends for the same channels from the past which will be useful for the MC1 debugging). Not sure whats going on.


The live data grabbing using cdsutils still seems to be working though - so I've kicked MC1 again, and am grabbing 2 hours of data live on Pianosa.

So we tried this again with a fresh build of daqd_fw, and it still fails.  The error message is pointing to an underlying bug in the framecpp library ("LDASTools"), which may be tricky to solve.  I'm rustling the appropriate bushes...

  13443   Wed Nov 22 00:54:18 2017 johannesOmnistructureComputersSlow DAQ replacement computer progress

I got the the SuperMicro 1U server box from Larry W on Monday and set it up in the CryoLab for initial testing.

The specs: https://www.supermicro.com/products/system/1U/5015/SYS-5015A-EHF-D525.cfm

The processor is an Intel D525 dual core atom processor with 1.8 GHz (i386 architecture, no 64-bit support). The unit has a 250GB SSD and 4GB RAM.

I installed Debian Jessie on it without any problems and compiled the most recent stable versions of EPICS base (3.15.5), asyn drivers (4-32), and modbus module (2-10-1). EPICS and asyn each took about 10 minutes, and modbus about 1 minute.

I copied the database files and port driver definitions for the cryolab from cryoaux, whose modbus services I suspended, and initialized the EPICS modbus IOC on the SuperMicro machine instead. It's working flawlessly so far, but admittedly the box is not under heavy load in the cryolab, as the framebuilder there is logging only the 16 analog channels.

I have recently worked out some kinks in the port driver and channel definitions, most importantly:

  • mosbus IOC initialization is performed automatically by systemd on reboot
  • If the IOC crashes or a system reboot is required the Acromag units freeze in their last current state. When the IOC is started a single read operation of all A/D registers is performed and the result taken as the initial value of the corresponding channel, causing no discontinuity in generated voltage EVER (except of course for the rare case when the Acromags themselves have to be restarted)

Aaron and I set 12/4 as a tentative date when we will be ready to attempt a swap. Until then the cabling needs to be finished and a channel database file needs to be prepared.

  13458   Wed Nov 29 21:40:30 2017 johannesOmnistructureComputersSlow DAQ replacement computer progress

[Aaron, Johannes]

We configured the AtomServer for the Martian network today. Hostname is c1auxex2, IP is 192.168.113.49. Remote access over SSH is enabled.

There will be 6 acromag units served by c1auxex2.

Hostname Type IP Address
c1auxex-xt1221a 1221 192.168.113.130
c1auxex-xt1221b 1221 192.168.113.131
c1auxex-xt1221c 1221 192.168.113.132
c1auxex-xt1541a 1541 192.168.113.133
c1auxex-xt1541b 1541 192.168.113.134
c1auxex-xt1111a 1111 192.168.113.135

Some hardware to assemble the Acromag box and adapter PCBs are still missing, and the wiring and channel definitions have to be finalized. The port driver initialization instructions and channel definitions are currently locally stored in /home/controls/modbusIOC/ but will eventually be migrated to a shared location, but we need to decide how exactly we want to set up this infrastructure.

  • Should the new machines have the same hostnames as the ones they're replacing? For the transition we simply named it c1auxex2.
  • Because the communication of the server machine with the DAQ modules is happening over TCP/IP and not some VME backplane bus we could consolidate machines, particularly in the vertex area.
  • It would be good to use the fact that these SuperMicro servers have 2+ ethernet ports to separate CDS EPICS traffic from the modbus traffic. That would also keep the 30+ IPs for the Acromag thingies off the Martian host tables.
  12290   Tue Jul 12 09:18:12 2016 ericqUpdateGeneralSlippery substance mystery

I found a note on Steve's desk that R. Abbott left yesterday afternoon about an unidentified slippery substance being present on the floor by cabinet S12, along the X arm. (Steve is away this week)

Just now, I found no trace of the substance in the vicinity of that cabinent (which is one of the cabinets for clean objects). Maybe the janitor cleaned it already?

  12291   Tue Jul 12 09:35:51 2016 JohannesUpdateGeneralSlippery substance mystery

I've noticed the spot that Rich means before, too. I think you only notice this when you're wearing the shoe covers, not sneakers or crocs. I didn't see any 'substance', it seems more like the floor finish (wax?) seems to be more slippery in that area than others.

Quote:

I found a note on Steve's desk that R. Abbott left yesterday afternoon about an unidentified slippery substance being present on the floor by cabinet S12, along the X arm. (Steve is away this week)

Just now, I found no trace of the substance in the vicinity of that cabinent (which is one of the cabinets for clean objects). Maybe the janitor cleaned it already?

 

  15290   Wed Apr 1 00:51:41 2020 gautamUpdateWienerSlightly improved MCL FF

Summary:

Retraining the MCL filters resulted in a slight improvement in the performance. Compared to no FF, the RMS in the 0.5-5 Hz range is reduced by approximately a factor of 3

Details:

Attachment #1 shows my re-measurement of the MC2 position drive to MCL transfer function.

  • The measurement was made using DTT swept sine, with the amplitude enveloped appropriately to avoid knocking the IMC out of lock.
  • Coherence was >0.97 for all datapoints.
  • Fitting was done using Lee's IIRrational, with the weighting being the coherence. I think there are some features of the fitting I don't fully understand, but I wanted to try and do everything in python and for this simple fit, it came out nicely I think. 

Attachment #2 shows the IIR fits to the FIR filters calculated here

  • Again, IIRrational was used. 
  • In the frequency band where subtraction is possible, the fit is good.
  • But there is definitely room for improvement in the way this is done, for now, I did quite a bit "by eye" and tweaked the order of the filter and the minimum number of excess poles relative to zeros to get the AC coupling, but it'd be nice to make all of this iterative and quantitative (e.g. by minimizing a cost function).
  • One nice feature of IIRrational is that it directly gives me a formatted string I can paste into foton. The order of these fits were 22, so I split them into two 19+3 order filters to be compatible with the realtime system before loading the coefficients (the overall gain was allocated to a single filter arbitrarily, with the other filter in the pair set to have unity gain in the zpk representation).

Attachment #3 shows several MCL spectra.

  • Blue trace is the unsubtracted test dataset.
  • Red is the performance of the calculated FIR filter, but the filtering is done offline.
  • Gold is the performance of the IIR fit to the FIR filter, as shown in Attachment #2, applied offline to the test dataset.
  • Green is the calculated ASD of MCL from a ~1 hour stretch from earlier tonight, when I left the feedforward loop on. So this is an actual measurement of the online performacne of the filter.
  • Grey is the performance of the old filter loaded in the CDS system - the filtering is done using scipy, and the sos coefficients from the C1OAF.txt file.

Conclusions + next steps

  1. Retraining the filters has resulted in a slight improvement, especially at ~3 Hz.
  2. More tests need to be done to confirm that noise isn't being reinjected in the frequency bands where subtraction isn't possible (e.g. using arm cavities as OOL sensors).
  3. The online filter isn't quite as good as what we would expect from calculations (green trace is noisier than gold). Need to think about why this is.
  4. Why can't we get more subtraction at 1 Hz?
  5. Now that I have the infrastructure ready, I will attempt to revive the PRC angular FF loops, which was the whole point of this exercise. 
Attachment 1: MC2_act_calib.pdf
MC2_act_calib.pdf
Attachment 2: IIR_fit_to_FIR.pdf
IIR_fit_to_FIR.pdf
Attachment 3: FIRvIIR.pdf
FIRvIIR.pdf
  15611   Mon Oct 5 00:37:19 2020 gautamUpdateBHDSingle bounce interferometer locked

Summary:

The simple interferometer, composed of a single bounce reflection from ITMY and the LO beam deilvered via fiber to the AS table, can be locked - i.e. the phase of the LO beam can be controlled such that the DC light level on the DCPDs after the two beams are interfered can be stabilized. This test allows us to confirm that various parts of the sensing and actuation chain (e.g. PI PZT for homodyne phase control, Trek amplifier etc etc) are working.

I will post more quantitative analysis tomorrow.

Optical configuration:

  • LO beam is a pickoff of the main PSL beam from just before it goes into the vacuum. The optical power arriving on each DCPD after the various beamsplitters, coupling loss etc is ~200 uW.
  • IFO beam is the single bounce reflection from ITMY. For this test, ETMY, ITMX and ETMY are misaligned. Optical power arriving on each DCPD is ~80uW.
  • The two beams are interfered on a 50-50 beamsplitter. The mode-matching efficiency was estimated to be ~50% which isn't stellar, but should be fine for this test.
  • So, at half-fringe, we expect the signal on each DCPD to be linearly proportional to the phase difference between the two fields, and so we can use that as an error signal.

Servo topology:

Attachment #2 shows the servo topology.

  • For a first attempt to close the feedback loop, we can consider the two blocks labelled "Sensing Chain" and "Actuation chain" to have a flat frequency response. While this isn't true, for a taget loop with ~100 Hz UGF, I think the approximation is reasonable.
  • From the peak-to-peak value (160 cts) of the DCPD signals when the homodyne phase is uncontrolled, I estimate a sensing response (at half-fringe) of approximately 0.3 ct/nm, since this corresponds to 532nm of relative phase between the two beams. 
  • An inverting summing amplifier is used to map the +/- 2^15 ct DAC range to 0-125V on the PI PZT. Assuming the full stroke of the PZT is 10um per the datasheet, and that this voltage range drives half of the full stroke (this is just a guess since all the old PI PZT circuits were designed to work at 0-250 V), we get an actuation coefficient of 0.075 nm/ct.
  • Using these two numbers, we can then design a digital feedback loop that gives an open loop transfer function with ~100 Hz UGF, and sufficient stability margin.
  • From the earlier measurements, we have an estimate for the amount of phase fluctuations caused by (i) seismic disturbances and (ii) fiber phase noise. This is the quantity we wish to suppress, and the suppression factor will be 1/(1+L), where L is the open loop gain.
  • I didn't do this in any systematic way, but the loop in Attachment #3 seemed like a reasonable shape that would suppress the error signal RMS by ~10x, as shown in Attachment #4. So I decided to try this out.

Other notes:

  1. The idea of offloading the DC control voltage to the ITMY suspension seemed to work fine.
  2. It also seems like the relative phase between the two beams doesn't drift by so large an amount in short time scales, at least at night/quiet seismic conditions. So it is possible to maintain the lock for several seconds without having to offload the DC signal to the suspensions.
  3. I didn't bother adapting the FSS Slow PID script to do this offloading in an automated way, seemed like more trouble than was just doing it by hand. But we may want to automate this in the future.
  4. I couldn't make a clean measurement of the loop transfer function using the usual IN1/IN2 method. Introducing a step offset at the error point, the servo is able to track it (I didn't fit the step response time, but it's not as if the loop bandwidth is <1 Hz or something). I have to compare the measured in-loop error signal ASD to the free-running one to get a feel for what the UGF is, I guess, to rule out a weird loop.
  5. Update 1100 Oct 6 2020: I have now added measured, in-loop, error point spectra to Attachment #4. Looks like there might be significant sensing noise re-injection.
    • Initially, I forgot to turn the HEPA on the PSL down for the measurement. So I have the two traces to compare. Looks like with the HEPA turned up to full, there is more noise in the 50-200 Hz range.
    • The trace marked "highGain" was taken with an overall loop gain that was 3dB higher than the nominal value - I could see some oscillations start to appear, and in the spectrum, maybe the feature at ~150 Hz is evidence of some gain peaking?

Conclusions:

  1. The PI PZT seems to work just fine.
  2. Need to look into the loop shape. I guess it's not reasonable to expect a UGF much higher than 100-200 Hz, because of the various delays in the system, but maybe the low frequency suppression can be made better.
  3. What are the next steps?? What does this mean for the RF44 sensing scheme?
Attachment 1: simpleHomodyne.png
simpleHomodyne.png
Attachment 2: singleBounceIFO.pdf
singleBounceIFO.pdf
Attachment 3: proposedController.pdf
proposedController.pdf
Attachment 4: freeRunningSuppressed.pdf
freeRunningSuppressed.pdf
  15612   Mon Oct 5 00:53:16 2020 KojiUpdateBHDSingle bounce interferometer locked

🤘🤘🤘

 

  11061   Tue Feb 24 18:54:26 2015 ericqUpdateASCSingle arm QPD ASC stability

I've lowered the UGFs for the transmission QPD servos to ~1-2Hz, and made it just an integrator. I left the arms locked with the QPD servos on for a few hours during the daytime today, and they succesfully prevented the Y arm from losing power from alignment drift for ~4 hours. Turning the servo off caused TRY to drop to ~0.6 or so. 

The X arm was only held for 2 hours or so, because after some unlock/drift event the power was below the servo trigger threshold. However, after gently nudging ETMX to get the transmission above the threshold, the servo kicked in, and brought it right back to TRX=1.0

Unfortunately, daqd was dead for much of the day, so I don't have much data to show; the trend was inferred from the wall striptool. 

It is not proven that there aren't further issues that prevent this from working with higher / more dynamic arm powers, but this is at least a point in favor of it working. 

EDIT: Here's a screenshot of the wall StripTool. Brown is TRY, blue is TRX. The downturn at the very end is me deactivating the servos. 

There is no scientific justifcation for the 0.9 threshold. Really, I should look at the noise/SNR again, now that there is some ND filtering on the QPDs. 

Attachment 1: trend.png
trend.png
  1432   Thu Mar 26 04:09:38 2009 YoichiUpdateIOOSingle X arm lock spectra with different MC lock schemes
The attached plots show MC_F, FSS_FAST_F and XARM IN/OUT spectra with different MC locking modes.
The conventional locking means the FSS is used. The direct frequency lock is the new way.
You can see that at low frequencies, the frequency actuator is working hard to suppress the MC pendulum motions.
The X-arm also sees a lot of frequency noise at low frequencies because of this.
The transmitted power of the X-arm fluctuates a lot making it difficult to align the mirrors.

The zoomed plots show that the structures in the kHz band are also present in the case of the direct frequency lock, although the frequencies are somewhat different.
Attachment 1: XarmSpectra.pdf
XarmSpectra.pdf
Attachment 2: XarmSpectraZoom.pdf
XarmSpectraZoom.pdf
  1757   Thu Jul 16 10:52:58 2009 ClaraUpdatePEMSingle Channel TRS-RNC Cable

I made and tested a female-to-female TRS(audio)-RNC cable. It only has a single channel, so it won't work for stereo speakers or anything, but I should only need one speaker for testing the microphones. The tip of the plug is the signal, the sleeve is ground, and the ring is null.

  16242   Fri Jul 9 15:39:08 2021 AnchalSummaryALSSingle Arm Actuation Calibration with IR ALS Beat [Correction]

I did this analysis again by just doing demodulation go 5s time segments of the 60s excitation signal. The major difference is that I was not summing up the sine-cosine multiplied signals, so the error associated was a lot more. If I simply multpy the whole beatnote signal with digital LO created at excitation frequency, divide it up in 12 segments of 5 s each, sum them up individually, then take the mean and standard deviation, I get the answer as:
\frac{6.88 \pm 0.05}{f^2} nm/ctsas opposed to \frac{7.32 \pm 0.03}{f^2} nm/ctsthat was calculated using MICH signal earlier by gautum in 13984.

Attachment 1 shows the scatter plot for the complex calibration factors found for the 12 segments.

My aim in the previous post was however to get a time series of the complex calibration factor from which I can take a noise spectral density measurement of the calibration. I'll still look into how I can do that. I'll have to add a low pass filter to integrate the signal. Then the noise spectrum up to the low pass pole frequency would be available. But what would this noise spectrum really mean? I still have to think a bit about it. I'll put another post soon.

Quote:

We attempted to simulate "oscillator based realtime calibration noise monitoring" in offline analysis with python. This helped us in finding about a factor of sqrt(2) that we were missing earlier in 16171. we measured C1:ALS-BEATX_FINE_PHASE_OUT_HZ_DQ when X-ARM was locked to main laser and Xend green laser was locked to XARM. An excitation signal of amplitude 600 was setn at 619 hz at C1:ITMX_LSC_EXC.

Signal analysis flow:

  • The C1:ALS-BEATX_FINE_PHASE_OUT_HZ_DQ is calibrated to give value of beatntoe frequency in Hz. But we are interested in the fluctuations of this value at the excitation frequency. So the beatnote signal is first high passed with 50 hz cut-off. This value can be reduced a lot more in realtime system. We only took 60s of data and had to remove first 2 seconds for removing transients so we didn't reduce this cut-off further.
  • The I and Q demodulated beatntoe signal is combined to get a complex beatnote signal amplitude at excitation frequency.
  • This signal is divided by cts amplitude of excitation and multiplied by square of excitation frequency to get calibration factor for ITMX in units of nm/cts/Hz^2.
  • The noise spectrum of absolute value of  the calibration factor is plotted in attachment 1, along with its RMS. The calibration factor was detrended linearly so the the DC value was removed before taking the spectrum.
  • So Attachment 1 is the spectrum of noise in calibration factor when measured with this method. The shaded region is 15.865% - 84.135% percentile region around the solid median curves.

We got a value of \frac{7.3 \pm 3.9}{f^2}\, \frac{nm}{cts}.  The calibration factor in use is from \frac{7.32}{f^2} nm/cts from 13984.

Next steps could be to budget this noise while we setup some way of having this calibration factor generated in realitime using oscillators on a FE model. Calibrating actuation of a single optic in a single arm is easy, so this is a good test setup for getting a noise budget of this calibration method.

 

Attachment 1: ITMX_calibration_With_ALS_Beat.pdf
ITMX_calibration_With_ALS_Beat.pdf
  16168   Fri May 28 17:32:48 2021 AnchalSummaryALSSingle Arm Actuation Calibration with IR ALS Beat

I attempted a single arm actuation calibration using IR beatnote (in the directions of soCal idea for DARM calibration)


Measurement and Inferences:

  • I sent 4 excitation signals at C1:SUS-ITM_LSC_EXC wit 30cts at 31Hz, 200cts at 197Hz, 600cts at 619Hz and 1000cts at 1069 Hz.
  • These were sent simultaneously using compose function in python awg.
  • The XARM was locked to mai laser and alignment was optimized with ASS.
  • The Xend Green laser was locked to XARM and alignment was optimized.
    • Sidenote: GTRX is now normalized to give 1 at near maximum power.
    • Green lasers can be locked with script instead of toggling.
    • Script can be called from sitemap->ALS->! Toggle Shutters->Lock X Green
    • Script is present at scripts/ALS/lockGreen.py.
  • C1:ALS-BEATX_FINE_PHASE_OUT_HZ_DQ was measured for 60s.
  • Also, measured C1:LSC-XARM_OUT_DQ and C1:SUS-ITMX_LSC_OUT_DQ.
  • Attachment 1 shows the measured beatnote spectrum with excitations on in units of m/rtHz.
  • It also shows resdiual displacement contribution PSD of (output referred) XARM_OUT and ITMX_LSC_OUT to the same point in the state space model.
    • Note: that XARM_OUT and ITMX_LSC_OUT (excitation signal) get coherently added in reality and hence the beatnote spectrum at each excitation frequency is lower than both of them.
    • The remaining task is to figure out how to calculate the calibration constant for ITMX actuation from this information.
    • I need more time to understand the mixture of XARM_OUT and ITMX_LSC_OUT in the XARM length node in control loop.
    • Beatnote signal tells us the actual motion of the arm length, not how much ITMX would have actuated if the arm was not locked.
  • Attachment 2 has the A,B,C,D matrices for the full state space model used. These were fed to python controls package to get transfer functions from one point to another in this MIMO.
    • Note, that here I used the calibration of XARM_OUT we measured earlies in 16127.
    • On second thought, maybe I should first send excitation in ETMX_LSC_EXC. Then, I can just measure ETMX_LSC_OUT which includes XARM_OUT due to the lock and use that to get calibration of ETMX actuation directly.

Attachment 1: SingleArmActCalwithIRALSBeat.pdf
SingleArmActCalwithIRALSBeat.pdf
Attachment 2: stateSpaceModel.zip
  16171   Tue Jun 1 16:55:32 2021 Anchal, PacoSummaryALSSingle Arm Actuation Calibration with IR ALS Beat

Rana suggested in today's meeting to put in a notch filter in the XARM IR PDH loop to avoid suppressing the excitation line. We tried this today first with just one notch at 1069 Hz and then with an additional notch at 619 Hz and sent two simultaneous excitations.


Measurement and Analysis:

  • We added notch filters with Q=10, depth=50dB, freq=619 Hz and 1069 Hz using foton in SUS-ETMX_LSC filter bank at FM10.
  • We sent excitation signals with amplitudes 600cts and 1000 cts for 619 Hz and 1069 Hz signals respectively.
  • We measured time series data of C1:SUS-ITMX_LSC_OUT_DQ and C1:ALS-BEATX_FINE_PHASE_OUT_HZ_DQ for 60s.
  • Then, spectrum of both signals is measured with Hanning window using scipy.welch function with scaling set to  'spectrum', binwidth=1Hz.
  • The beatnote signal was converted into length units by multiplying it by 1064nm * 37.79m / c.
  • The ratio of the two spectrums at teh excitation frequency multiplies by excitation frequency squared gives us teh calibration constant in units of nm Hz^2/cts.
  • At 619 Hz, we got \frac{5.01}{f^2}nm/cts
  • At 1069 Hz, we got \frac{5.64}{f^2}nm/cts.
  • The calibration factor in use is from \frac{7.32}{f^2} nm/cts from 13984.
  • So, the calibration factor from this methos is about 23% smaller than measured using freeswinging MICH in 13984.
  • One possiblity is that our notch filter is not as effective in avoiding suppresion of excitation.
    • We tried increasing the notch filter depths to 100 dB but got the same result within 2%.
    • We tried changing the position of notch filters. We put them in POX filter banks. Again the result did not change more than 2%.
  • The open loop gain of green PDH at 619 Hz and 1069 Hz must be large enough for our assumption of green laser perfectly following length motion to be true. The UGF of green laser is near 11 kHz.
  • The discrepancy could be due to outdated freeswinging MICH measurement that was done 3 years ago. Maybe we should learn how to do the ITMX calibration using this method and compare our own two measurements.
Attachment 1: SingleArmActCalwithIRALSBeat-1306624785.pdf
SingleArmActCalwithIRALSBeat-1306624785.pdf
  16192   Tue Jun 8 11:40:53 2021 Anchal, PacoSummaryALSSingle Arm Actuation Calibration with IR ALS Beat

We attempted to simulate "oscillator based realtime calibration noise monitoring" in offline analysis with python. This helped us in finding about a factor of sqrt(2) that we were missing earlier in 16171. we measured C1:ALS-BEATX_FINE_PHASE_OUT_HZ_DQ when X-ARM was locked to main laser and Xend green laser was locked to XARM. An excitation signal of amplitude 600 was setn at 619 hz at C1:ITMX_LSC_EXC.

Signal analysis flow:

  • The C1:ALS-BEATX_FINE_PHASE_OUT_HZ_DQ is calibrated to give value of beatntoe frequency in Hz. But we are interested in the fluctuations of this value at the excitation frequency. So the beatnote signal is first high passed with 50 hz cut-off. This value can be reduced a lot more in realtime system. We only took 60s of data and had to remove first 2 seconds for removing transients so we didn't reduce this cut-off further.
  • The I and Q demodulated beatntoe signal is combined to get a complex beatnote signal amplitude at excitation frequency.
  • This signal is divided by cts amplitude of excitation and multiplied by square of excitation frequency to get calibration factor for ITMX in units of nm/cts/Hz^2.
  • The noise spectrum of absolute value of  the calibration factor is plotted in attachment 1, along with its RMS. The calibration factor was detrended linearly so the the DC value was removed before taking the spectrum.
  • So Attachment 1 is the spectrum of noise in calibration factor when measured with this method. The shaded region is 15.865% - 84.135% percentile region around the solid median curves.

We got a value of \frac{7.3 \pm 3.9}{f^2}\, \frac{nm}{cts}.  The calibration factor in use is from \frac{7.32}{f^2} nm/cts from 13984.

Next steps could be to budget this noise while we setup some way of having this calibration factor generated in realitime using oscillators on a FE model. Calibrating actuation of a single optic in a single arm is easy, so this is a good test setup for getting a noise budget of this calibration method.

Attachment 1: ITMX_Cal_Noise_Spectrum_1307143423.pdf
ITMX_Cal_Noise_Spectrum_1307143423.pdf
  8299   Fri Mar 15 02:14:27 2013 JenneUpdateCDSSimulink linking to wrong library part

Jamie and I discovered a problem with Matlab/Simulink earlier today. 

In the end suspension models, there is a subblock (with top_names) for ALS stuff.  Inside there, we use a library part called "ALS_END".  When the model was created, it included the part ...../userapps/release/isc/c1/models/ALS_END.mdl .  However, if you open up the c1scy diagram and look in the ALS block for this part, you see the part that is in ..../userapps/release/isc/common/models/ALS_END.mdl .  Note the difference - the one we want is in the c1 directory, while the one that was created (by Jamie) for the LHO One Arm Test is in the common directory. 

If you compile the c1scy model, the RCG is using the correct library part, so the information regarding which part we want is still in there. 

However, if you delete the ALS_END part from the model, put the correct one in, save, close, then reopen the model, it once again displays the wrong model.  The right click "go to library part" option brings you to the library part that is displayed, which is currently the wrong one.  THIS IS BAD, since we could start modifying the wrong things.  You do get a warning by Matlab about the file being "shadowed", so we should take heed when we see that warning, and make sure we are getting the file we want.

We are currently running Matlab version 7.11.0.584, which is r2010b.  Step 1 will be to update Matlab to the latest version, in hopes that this fixes things.  We also should change the name of our c1 part, so that it does not have the same name as the one for the sites.  This is not a great solution since we can't guarantee that we will never choose the same names as other sites, but it will at least fix this one case.  Again, if you see the warning about "shadowed" filenames, pay attention.

  11564   Thu Sep 3 02:12:08 2015 ranaUpdateCDSSimulink Webview updated

Back in 2011, JoeB wrote some entries on how to automatically update the Simulink webview stuff.

Somehow, the cron broke down over the years. I reran the matlab file by hand today and it worked fine, so now you can see the up to date models using the internet.

https://nodus.ligo.caltech.edu:30889/FE/

  11567   Thu Sep 3 13:25:40 2015 ranaUpdateCDSSimulink Webview updated

added the cron script for this to megatron to run at 8:44 AM each morning. Here's the new MegaCron attached :-()-

** it takes ~13 minutes to complete on megatron

Attachment 1: crontab_150903.rtf
MAILTO=ericq@caltech.edu

# m h  dom mon dow   command
#0 */1 * * * bash /home/controls/public_html/summary/bin/c1_summary_page.sh > /dev/null 2>&1
#15 5 * * * /ligo/apps/nds2/nds2-megatron/test-restart

# MEDM Screen caps for the webpage
2,13,25,37,49 * * * * /cvs/cds/project/statScreen/bin/cronjob.sh

# op340m transplants -ericq
... 18 more lines ...
  12727   Tue Jan 17 20:47:23 2017 ranaUpdateCDSSimulink Webview updated

Seems like this stops working every ~2 years. Its been busted since early 2016 according to cron, so I fixed up the paths and restored some missing files and committed things to the SVN (with comments!) and now its working and grabbing the Web viewable versions of the front end models. Just need to restore its viewability and then the world can watch our models any time.

Quote:

Back in 2011, JoeB wrote some entries on how to automatically update the Simulink webview stuff.

Somehow, the cron broke down over the years. I reran the matlab file by hand today and it worked fine, so now you can see the up to date models using the internet.

https://nodus.ligo.caltech.edu:30889/FE/

 

  16930   Mon Jun 20 19:46:04 2022 TomislavUpdateASCSimulation plots

In the attachment please find IMC ASC simulation plots. Let me know what you think, if you want some other plots, and if you need any clarification.

Attachment 1: pit_mot_cl_MCs.png
pit_mot_cl_MCs.png
Attachment 2: loc_damp_cl_MCs.png
loc_damp_cl_MCs.png
Attachment 3: contr_output_cl_MCs.png
contr_output_cl_MCs.png
Attachment 4: sens_output_cl_MCs.png
sens_output_cl_MCs.png
Attachment 5: BS_motion_cl_MCs.png
BS_motion_cl_MCs.png
  16934   Tue Jun 21 18:41:46 2022 TomislavUpdateASCSimulation plot

In the attachment please find a comparison of error signals of simulation and reality. For C1:IOO-WFS1/2_PIT_IN1 excess signal ('belly') between a few Hz up to 70-80 Hz might be caused by air turbulence (which is not included in the simulation).

Attachment 1: sens_output_comparison.png
sens_output_comparison.png
  9326   Fri Nov 1 17:01:46 2013 GabrieleSummaryLSCSimulation of REFL_3f signal when the arms come in

 I simulated how the 3f signal is affected by the resonance condition of the arms.

To keep it simple, I only simulated a double cavity. The attached plot shows the result. In x there is the arm cavity detuning from resonance (in log scale to show what happens close to the 0 value). In the y axis there is the PRC detuning. So every vertical slice of the upper plot gives a PDH signal for a given arm detuning. The bottom plot shows the power build up inside the arm, which is dominated by the carrier.

refl_3f_I_vs_cavity_tuning_noresf2.png

The 3f signal is not perturbed in any significant way by the arm resonance condition. This is good and what we expected.

However, in this simulation I had to ensure that the 1f sidebands are not perfectly anti-resonant inside the arms. They are indeed quite far away from resonance. If the modulation frequency is chosen in order to make the 1f sidebands exactly ant-resonant, the 2f will be resonant. This screws up the signal: REFL_3f is made of two contributions of equal amplitude, one on the PRC sidebands resonance and the other on the PRC carrier resonance. When the arm tuning goes to zero, these two cancels out and there is no more PDH...

refl_3f_I_vs_cavity_tuning.png

However, this is a limit case, since the frequency show match perfectly. If the modulation frequency is few arm line widths away from perfect anti-resonance, we have no problem.

  9327   Fri Nov 1 17:44:06 2013 KojiSummaryLSCSimulation of REFL_3f signal when the arms come in

Yes, the resonance of the 2nd-order sidebands to the IFO screws up the 3f scheme.

2f (~22MHz) and 10f (~110MHz) are at x 5.6 and x 27.9 FSR from the carrier, so that's not the case.

Could we also see how much gain fluctuation of the 3f signals we would experience when the arm comes into the resonance?

  9337   Mon Nov 4 14:11:23 2013 GabrieleSummaryLSCSimulation of REFL_3f signal when the arms come in

Quote:

Yes, the resonance of the 2nd-order sidebands to the IFO screws up the 3f scheme.

2f (~22MHz) and 10f (~110MHz) are at x 5.6 and x 27.9 FSR from the carrier, so that's not the case.

Could we also see how much gain fluctuation of the 3f signals we would experience when the arm comes into the resonance?

 From the simulation there is no visible change in the gain.

  14635   Thu May 23 15:37:30 2019 MilindUpdateCamerasSimulation enhancements and performance of contour detection
  1. Implemented image level noise for simulation. Added only uniform random noise.
  2. Implemented addition of uniform random noise to any sinusoidal motion of beam spot.
  3. Implemented motion along y axis according to data in "power_spectrum" file.
  4. Impelemented simulation of random motion of beam spot in both x and y directions (done previously by Pooja, but a cleaner version).
  5. Created a video file for 10s with motion of beam spot along the y direction as given by Attachment #1. This was created by mixing four sinusoids at different amplitudes (frequencies (0.1, 0.2, 0.4, 0.8) Hz Amplitudes as fractions of N = 64 (0.1 0.09 0.08 0.09). FPS = 10. Total number of frames = 100 for the sake of convenience.  See Attachment #5.
  6. Following this, I used the thresholding (threshold = 127, chosen arbitrarily), contour detection and centroid computation sequence (see Attachment #6 for results) to obtain the plot in Attachment 2 for the predicted motion of the y coordinate. As is evident, the centering and scale of values obtained are off and I still haven't figured out how to precisely convert from one to another.
  7. Consequently, as a workaround, I simply normalised the values corresponding to each plot by subtracting the mean in each case and dividing the resulting series of values by their maximum. This resulted in the plots in Attachments 3 and 4 which show the normalised values of y coordinate variation and the error between the actual and predicted values between 0 and 1 respectively.

Things yet to be done:

Simulation:

  1. I will implement the mean square error function to compute the relativer performance as conditions change.
  2. I will add noise both to the image and to the motion (meaning introduce some randomness in the motion) to see how the performance, determined by both the curves such as the ones below and the mean square error, changes.
  3. Following this, I will vary the standard deviation of the beam spot along X and Y directions and try to obtain beam spot motion similar to the video in Attachment #2 of elog post 14632.
  4. Currently, I have made no effort to carefully tune the parameters associated with contour detection and threshold and have simply used the popular defaults. While this has worked admirably in the case of the simple simulated videos, I suspect much more tweaking will be needed before I can use this on real data.
  5. It is an easy step to determine the performance of the algorithm for random, circular and other motions of the beam spot. However, I will defer this till later as I do not see any immediate value in this.
  6. Determine noise threshold. In simulation or with real data: obtain a video where the beam spot is ideally motionless (easy to do with simulated data) and then apply the above approach to the video and study the resulting predicted motion. In simulation, I expect the predictions for a motionless beam spot video (without noise) to be constant. Therefore, I shall add some noise to the video and study the prediction of the algorithm.
  7. NOTE: the above approach relies on some previous knowledge of what the video data will look like. This is useful in determining which contours to ignore, if any like the four bright regions at the corners in this video.

Real data:

  1. Obtaining real data and evaluate if the algorithm is succesful in determining contours which can be used to track the beam spot.
  2. Once the kind of video feed this will be used on is decided, use the data generated from such a feed to determine what the best settings of hyperparameters are and detect the beam spot motion.
  3. Synchronization of data stream regarding beam spot motion and video.
  4. Determine the calibration: anglular motion of the optic to beam spot motion on the camera sensor to video to pixel mapping in the frames being processed.

Other approaches:

  1. Review work done by Gabriele with CNNs, implement it and then compare performance with the above method.
Attachment 1: actual_motion.pdf
actual_motion.pdf
Attachment 2: predicted_motion.pdf
predicted_motion.pdf
Attachment 3: normalised_comparison.pdf
normalised_comparison.pdf
Attachment 4: residue_normalised.pdf
residue_normalised.pdf
Attachment 5: simulated_motion1.mp4
Attachment 6: elog_22may_contours.mp4
  14638   Sat May 25 20:29:08 2019 MilindUpdateCamerasSimulation enhancements and performance of contour detection
  1. I used the same motion as defined in the previous elog. I gradually added noise to the images. Noise added was uniform random noise - a 2 dimensinoal array of random numbers between 0 and a predetermined maximum (noise_amp). The previous elog provides the variation of the y coordinate. In this, I am also uploading the effect of noise on the error in the prediction of the x coordinate. As a reminder, the motion of the beam spot center was purely vertical. Attachement #1  is the error for noise_amp = 0, #2 for noise_amp = 20 and #3  for noise_amp = 40. While Attachment #3 does provide the impression of there being a large error, this is not really the case as without normalization, each peak corresponds to a deviation of one pixel about the central value, see Attachement #4 for reference.
  2. While the error does increase marginally, adding noise has no significant effect on the prediction of the y coordinate of the centroid as Attachment #5 shows at noise_amp = 40.
  3. I am currently running an experiment to obtain the variation of mean square error with different noise amplitudes and will put up the plots soon. Further, I shall vary the resolution of the image frames and the the standard deviation of the Gaussain beam with time and try to obtain simulations very close to the real data available and then determine the performance of the algorithm.
  4. The following videos will serve as a quick reference for what the videos and detection look like at
    1. noise_amp = 20
    2. noise_amp = 40
  5. I also performed a quick experiment to see how low the amplitude of motion could be before the algorithm falied to detect the motion and found it to occur at 2 orders of magnitude below the values used in the previous post. This is a line of thought I intend to pursue more carefully and I am looking into how opencv and python handle images with floats as coordinates and will provide more details about the previous trial soon. This should give us an idea of what the smallest motion of the beam spot that can be resolved is.
Quote:
  1. Implemented image level noise for simulation. Added only uniform random noise.
  2. Implemented addition of uniform random noise to any sinusoidal motion of beam spot.
  3. Implemented motion along y axis according to data in "power_spectrum" file.
  4. Impelemented simulation of random motion of beam spot in both x and y directions (done previously by Pooja, but a cleaner version).
  5. Created a video file for 10s with motion of beam spot along the y direction as given by Attachment #1. This was created by mixing four sinusoids at different amplitudes (frequencies (0.1, 0.2, 0.4, 0.8) Hz Amplitudes as fractions of N = 64 (0.1 0.09 0.08 0.09). FPS = 10. Total number of frames = 100 for the sake of convenience.  See Attachment #5.
  6. Following this, I used the thresholding (threshold = 127, chosen arbitrarily), contour detection and centroid computation sequence (see Attachment #6 for results) to obtain the plot in Attachment 2 for the predicted motion of the y coordinate. As is evident, the centering and scale of values obtained are off and I still haven't figured out how to precisely convert from one to another.
  7. Consequently, as a workaround, I simply normalised the values corresponding to each plot by subtracting the mean in each case and dividing the resulting series of values by their maximum. This resulted in the plots in Attachments 3 and 4 which show the normalised values of y coordinate variation and the error between the actual and predicted values between 0 and 1 respectively.

Things yet to be done:

Simulation:

  1. I will implement the mean square error function to compute the relativer performance as conditions change.
  2. I will add noise both to the image and to the motion (meaning introduce some randomness in the motion) to see how the performance, determined by both the curves such as the ones below and the mean square error, changes.
  3. Following this, I will vary the standard deviation of the beam spot along X and Y directions and try to obtain beam spot motion similar to the video in Attachment #2 of elog post 14632.
  4. Currently, I have made no effort to carefully tune the parameters associated with contour detection and threshold and have simply used the popular defaults. While this has worked admirably in the case of the simple simulated videos, I suspect much more tweaking will be needed before I can use this on real data.
  5. It is an easy step to determine the performance of the algorithm for random, circular and other motions of the beam spot. However, I will defer this till later as I do not see any immediate value in this.
  6. Determine noise threshold. In simulation or with real data: obtain a video where the beam spot is ideally motionless (easy to do with simulated data) and then apply the above approach to the video and study the resulting predicted motion. In simulation, I expect the predictions for a motionless beam spot video (without noise) to be constant. Therefore, I shall add some noise to the video and study the prediction of the algorithm.
  7. NOTE: the above approach relies on some previous knowledge of what the video data will look like. This is useful in determining which contours to ignore, if any like the four bright regions at the corners in this video.

Real data:

  1. Obtaining real data and evaluate if the algorithm is succesful in determining contours which can be used to track the beam spot.
  2. Once the kind of video feed this will be used on is decided, use the data generated from such a feed to determine what the best settings of hyperparameters are and detect the beam spot motion.
  3. Synchronization of data stream regarding beam spot motion and video.
  4. Determine the calibration: anglular motion of the optic to beam spot motion on the camera sensor to video to pixel mapping in the frames being processed.

Other approaches:

  1. Review work done by Gabriele with CNNs, implement it and then compare performance with the above method.

 

Attachment 1: residue_normalised_x.pdf
residue_normalised_x.pdf
Attachment 2: residue_normalised_x.pdf
residue_normalised_x.pdf
Attachment 3: residue_normalised_x.pdf
residue_normalised_x.pdf
Attachment 4: predicted_motion_x.pdf
predicted_motion_x.pdf
Attachment 5: normalised_comparison_y.pdf
normalised_comparison_y.pdf
  14667   Wed Jun 12 22:02:04 2019 MilindUpdateCamerasSimulation enhancements

Today, Rana asked me to work on improving simulations based on the ideas we discussed last week. As of the previous elog the simulation accomodated only

  1. Simulation of Gaussian beam spot.
  2. Arbitrary motion.

Today, I added the simulation of point scatterers.

What?

The image on the sensor (camera) is produced in roughly the following steps.

  1. Motion of the Gaussian beam on the optic (X,Y coordinates) which is what has been simulated so far.
  2. Reflection from the surface of the optic which can be modeled using knowledge of the BRDF has not been included as of this elog as I wish to do a little more reading before doing so.
  3. Reflection from point scatterers (dust particles burnt into the optic surface by the laser and so forth) which are characterised as peaks (impulses) in the TIS vs position plot. The laser beam is incident nearly normally on the optic and this behaviour is independent of the angle of observation. This is what has been added to the simulation.

How?

  1. Increased the frame resolution to 720 x 480.
  2. Defined an array of the same size and set values of at most "num_scatter" number of points at random positions to values determined randomly between 1 and "scatter_amp" + 1 where scatter_amp is non-negative.
  3. Multiplied the resulting array by the resulting Gaussian beam. The motivation was to imitate the bright specks obtained on various camera feeds in the lab. Physically, this also implies normal incidence and normal observation which is not the real case at all. I shall add these features in a day or two.

Herewith, in attachments #1, #2, #3 I am attaching videos obtained by varying scattering amplitude and number of scattering points in a vain attempt to reproduce this data. I shall work more on this simulation on Friday.

 


Scripting stuff:

  1. Previous elogs detail how to take gige images at various exposure times. I am still waiting on Kruthi to use the script.
  2. Tomorrow I shall work on the scripting software to interact with the GigE and take video for a fixed duration etc. I shall also begin working on a script to autolock the PMC based on what Rana showed me on Monday. I will also take a look at the the contents of this elog and try to pick up from there. I hope to make significant progress by the next lab meeting.

Neural network stuff:

GANs for simulation:

  1. Other than putting the physics into simulation i.e the first portion of this elog, GANs can be trained to generate images similar to the original data. I am unfamiliar with training GANs and the various tricks that are used specifically for them. I will do a bit of reading and make an update by Friday. As of now, the data I plan to use is this and I will train it using the GTX 1060 on my machine.

Networks for beam tracking:

  1. I will use the architectures suggested in this work with a few modifications. I will use MSE loss function, Adam optimizer and my local GPU for training.
Attachment 1: simulated_motion0.mp4
Attachment 2: simulated_motion0.mp4
Attachment 3: simulated_motion0.mp4
  14698   Tue Jun 25 23:52:37 2019 MilindUpdateCamerasSimulation enhancements

Yesterday, Rana asked me to look at Hiro Yamamoto's docs on the DCC to improve the simulation. I'm performing a first pass (=> Just skimming through to see if they're relevant, I will go through them more carefully soon!) and putting up stuff here for future reference. @Kruthi's help much appreciated!

  14714   Mon Jul 1 20:11:34 2019 MilindUpdateCamerasSimulation enhancements

Today, I read a lot more about BRDF and modelling but could not make much headway regarding the implementation in the simulation. I've stopped for now and I'll take a crack at it tomorrow again.

Quote:

Yesterday, Rana asked me to look at Hiro Yamamoto's docs on the DCC to improve the simulation. I'm performing a first pass (=> Just skimming through to see if they're relevant, I will go through them more carefully soon!) and putting up stuff here for future reference. @Kruthi's help much appreciated!

  16948   Sat Jun 25 22:18:41 2022 TomislavUpdateASCSimulation and reality comparisons

In the attachment please find plots comparing controller output, local damping output, and error signals.

Input noises of the simulation are seismic noise, osem noise, input power fluctuations, sensing noises of WFSs and QPD, and air turbulence noise for WFSs. There is also optical torque noise (radiation pressure effect). 

The procedure to get optical gains and sensing noises:
Having the actuator response A rad/cnts @ 3 Hz. I was shaking MC1/2/3 in pitch with B cnts @ 3 Hz and getting WFS1/2 QPD signals of C cnts @ 3 Hz, which means WFS1/2 QPD optical gain is D cnts/rad = C / (A * B) cnts/rad. So, if WFS1/2 QPD IN1 has a noise spectrum (at higher freqs) of E cnts/rtHz, that corresponds to E/D rad/rtHz of sensing noise for WFS1/2 QPD.

Actuator response [rad/cts] I was getting shaking mirrors at 3 Hz and measuring amplitudes of OSEM output (knowing the geometry of the mirror). I scaled it to DC. From here I was getting ct2tau_mc (knowing the mirror's moment of inertia, Q, and natural pitch frequencies). OSEM calibration factors [cts/rad] I was getting from the input matrix and geometry of the mirror.

The flat noise at higher frequencies from the local damping and controller output channels is presumably quantization/loss of digits/numerical precision noise which I don't include in simulations for now?!

Regarding air turbulence, in KAGRA it has been reported that air turbulence introduces phase fluctuations in laser fields that propagate in air. According to Kolmogorov’s theory, the PSD of phase fluctuations caused by air turbulence scales as ∝ L*V^(5/3)*f^(−8/3). Here, L is the optical path length and V is a constant wind speed. Since it is not obvious how can one estimate typical V in the beam paths I was taking this excess noise from the error signals data between 10 Hz and 50 Hz, extrapolated it taking into account ∝ f^(−8/3) (not for frequencies below 2 Hz, where I just put constant, since it would go too high). I expect that I won't be able to get a parameterized model that also predicts the absolute value. The slope is all I can hope to match, and this I already know. QPD chamber is much smaller (and better isolated?) and there is no this excess noise.

Regarding other things in simulations (very briefly): beam-spots are calculated from angular motions, length change is calculated from beam-spots and angular motion, cavity power depends on length change and input power, and torque on the mirrors depends on beam-spots and cavity power. From other things, local-sensor basis conversion (and vice versa) is worth noting.

Attachment 1: sens_output_comparison_23_6_new.png
sens_output_comparison_23_6_new.png
Attachment 2: contr_out_comparison_23_6_new.png
contr_out_comparison_23_6_new.png
Attachment 3: local_damp_out_comp_23_6_new.png
local_damp_out_comp_23_6_new.png
  4388   Tue Mar 8 16:59:47 2011 josephbUpdateCDSSimulated Plant Work

The screens for the simplified c1spx model have been updated.  I re-introduced the suspension point information into the sensor output matrix so we can take into account the fact that as the entire supporting structure moves, the osems moves relative to the optic.

Master screens for the noise filters (i.e. 60 Hz, suspension point motion, and optic noise) have been created.

I have currently set the matrix values of the c1spx model to handle just longitudinal motion.  I.e. Coils drive only in the POS degree of freedom and sensor read outs are also only in the POS degree of freedom.  I've turned off all the noise inputs.

I added a simple double pole at 1 Hz in the C1:SUP_ETMX_PL_F2P_0_0 filter bank.

  16751   Fri Apr 1 14:26:50 2022 TegaUpdateOptical LeversSimplified sketch on MC table

Here is an early sketch of the MC table. 

 

Quote:

I have made an editable draw.io diagram of the planned simplified BHD setup on the ITMY table (see attached).  10 pts = 1 inch.

This is very sketchy but easily adjustable since we are removing everything but the ITMY Oplev from that table.

 

Attachment 1: MC_Table.drawio.pdf
MC_Table.drawio.pdf
  16752   Fri Apr 1 17:02:02 2022 KojiUpdateOptical LeversSimplified sketch on MC table

We are supposed to have BS Oplev Beams. We don't like the shallow angle reflections (i.e. AOI>45deg).

The laser is too big but I suspect the other components are too small. So it'd be check the actual size of the components including the optical mounts that are missing on the figure so far.

  16753   Fri Apr 1 22:22:29 2022 KojiUpdateOptical LeversSimplified sketch on MC table

Possibility to swap BS and ITMX tables:
BS table, which Tega said MC table, is 2ft x 4ft. The ITMX table is 3ft x 5ft and only the central 2ft x 4ft area is used. The area around the BS table is the narrowest for the east arm. We need at least (2+delta) ft of the hallway width so that we can move the instrument. I'm not yet sure if the ITMX table can be placed there without precise investigation.
 

  16755   Mon Apr 4 15:49:06 2022 TegaUpdateOptical LeversSimplified sketch on BS table

I have updated the BS table using feedback from Koji and Paco and the attached pdf document is the latest iteration.

Attachment 1: BS_Table.drawio.pdf
BS_Table.drawio.pdf
ELOG V3.1.3-