06:16 UTC Started script as controls on zotws3.

The VCO Sweep ended on 03/28/2019 at 10:22:05.349809 (UTC), 03:22:05.349809 (PT).

Here's a quick analysis, we'll do a more thorough one later. The attached plot shows glitches in DARM as a function of VCO frequency. There are many different lines that give beatnotes. The widest clean area is around 78.8 Mhz, with 79.75 MHz second best. It would be best to spend another hour doing more restricted sweeps over any proposed areas we plan to use.

If the calibration from the VCO tuning offset to the frequency is needed, Detchar can fit this data and provide that.

(Edit: New plot uploaded without the axis labels cut of.)

Exact sweep times for the ramp from +5V to -5V:

• Start: 3/28/19 07:32:45 UTC / GPS 1237793583
• Duration: 10107 sec
• Stop: 3/28/19 10:21:12 UTC / GPS 1237803690
• VCO tune offset channel: H1:IMC-VCO_TUNEOFS
• VCO frequency channel: H1:IMC-VCO_FREQUENCY

Plot 1 shows Frequency against tune voltage. The VCO frequency is offset by 79.2 MHz.

Plot 2 shows a histogram of the VCO frequencies during ramping in 5 kHz bins. Looks like we covered the full range.

The VCO frequency can be estimated by VCO frequency = 79.2 MHz – 44 kHz + v 140.3 kHz – v³ 0.4447 kHz, with v the tune voltage.

For 78.8 MHz we get –2.59 V, and for 79.75 MHz we get +4.53 V. The full range of this VCO is from 77.9 MHz to 80.4 MHz. With the full tune voltage range of ±10 V, we can access frequencies from ~78.0 MHz to ~80.3 MHz.

I've exported the above inverse sensing function (the CAL-CS_DARM_ERR filter bank, module FM9 "O3_D2N") for the GDS pipeline to correct for foton's IIR warping of the 7000 Hz, high-frequency roll-off pole.

It now lives in 
    /ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/O3/H1/Measurements/Foton/2019-03-27_H1CALCS_InverseSensingFunction_Foton_SRCD-2N_Gain_tf.txt

I've updated and commited the 20190327 model file to read in this new export.
    /ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/O3/H1/params/modelparams_H1_20190327.py

Because we always ask ourselves how these are exported:
    (1) Open the H1CALCS.txt file in foton
    (2) On the "design" tab, select the appropriate D2N filter (in this case "O3_D2N")
    (3) Under "plotting" make fStart = 0.01, and fStop = 10000 and Number of Points = 1001. Leave all other parameters as their default.
    (4) Hit Bode Plot
    (5) Either "File > Export" or hit the "Export" button.
    (6) Select data type Transfer Function, select Column 0, and export A: current_in and B: current_out, with Conv. "as is."

Then save the file to the above mentioned Measurements/Foton/ directory with an appropriate date.

This should produce a 1000 line text file, with three columns: frequency, real part, imaginary part.

On the to-do list: automate this.

Remember -- we actually have to have *foton* process the transfer function coefficients because (of course) it has a different pre-warping algorithm than matlab or python. See G1801581.

Data templates taken to support this entry
Sensing Function:
/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/O3/H1/Measurements/FullIFOSensingTFs
    2019-03-27_H1DARM_OLGTF_5to1100Hz_30min.xml
    2019-03-27_H1_PCAL2DARM_TF_5t1100Hz_15min.xml

    2019-03-27_H1_OMCDCPDSUM_to_DARMIN1.xml

    2019-03-27_H1DARM_OLGTF_BB.xml
    2019-03-27_H1_PCAL2DARMTF_BB.xml

Actuation Function:
/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/O3/H1/Measurements/FullIFOActuationTFs/
    2019-03-27_H1SUSETMX_L1_iEXC2DARM_25min.xml
    2019-03-27_H1SUSETMX_L1_PCAL2DARM_8min.xml
    2019-03-27_H1SUSETMX_L2_iEXC2DARM_17min.xml
    2019-03-27_H1SUSETMX_L2_PCAL2DARM_8min.xml
    2019-03-27_H1SUSETMX_L3_iEXC2DARM_8min.xml
    2019-03-27_H1SUSETMX_L3_PCAL2DARM_8min.xml

Scripts that process the data
Sensing Function:
/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/O3/H1/Scripts/FullIFOSensingTFs
    process_sensingmeas_20190327.py

Actuation Function:
/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/O3/H1/Scripts/FullIFOActuationTFs
    process_actuationmeas_20190327.py

These parameters are now in the filter modules that Jeff quotes, and I have accepted in both the safe.snap and OBSERVE.snap to be using the filter modules with "O3" in the name.  For the last while (days?weeks?) we've been putting things in the filter modules with "ER14" in the name.  Those still exist, but aren't in use anymore. 

Also, the gain of H1:CAL-CS_DARM_ERR_GAIN had been reverted (likely on Tuesday?) to the value of 1.03.  I have again put it to the nominal value of 1.00 and accepted this in SDF.

I've processed the DARM open loop gain transfer function, and computed the relative delay between the Actuation and Sensing Paths. 

The systematic error in the DARM open loop gain transfer function below 30 Hz is understood based on the results discussed above, and for now, we must tolerate them.

However, for the first time, we appear to have an appreciable, non-integer 16 kHz clock-cycle delay between the Actuation and Sensing paths: 7.653 clock cycles.
Sounds like it's finally time for me to import the thiran filter delay technology from LLO.

Scripts to produce these plots:
/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/O3/H1/Scripts/DARMOLGTFs/
    process_darmolg_20190327.py

/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/O3/H1/Scripts/CALCS_FE/
    compute_relativedelay_AvsC_20190327.py

I've used the above mentioned model to write new reference model parameters for the computation of time dependent correction factors, a.k.a. "the" EPICs records.


Script to create the records:
/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/O3/H1/Scripts/CALCS_FE
     createEPICS_for_20190327.py

Location of generated EPICs records:
/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/O3/H1/Results/CALCS_FE/
     epicsrecords_model-H1_20190327_created-20190327.txt

Model file used:
/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/O3/H1/params
     modelparams_H1_20190327.py

Attached are screenshots of the corresponding MEDM screen and SDF acceptance *after* the EPICs records were pushed.

H1 control room matlab (matlab2015b) was too old to be able to use autoquack, but joeb was able to do that for us using matlab2018.

cd /ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/PostO2/L1/Scripts/CAL_FE_DELAY

fs=2^14

ts=1/fs

td=7.653*ts

populate_cal_fe_delay('H1', td)

Took the diff with the original filter in the control room and it only has one filter "delay" defined in CS_DARM_CTRL_DELAY, everything else is the same.

diff /opt/rtcds/lho/h1/chans/H1CALCS.txt ~/H1CALCS.txt

780a781,787
> # DESIGN   CS_DARM_CTRL_DELAY 0 zpk([-3495.762131159507+i*542.3606112515328;-3495.762131159507-i*542.3606112515328;
> #                               -3679.520087768798+i*1130.184746679708;-3679.520087768798-i*1130.184746679708;
> #                               -4186.607279464874-i*1784.0262722568;-4186.607279464874+i*1784.0262722568;-3445.944679249666;
> #                               -6825.617444279973],[3495.762131159505+i*542.3606112515323;3495.762131159505-i*542.3606112515323;
> #                               3679.520087768799+i*1130.184746679708;3679.520087768799-i*1130.184746679708;
> #                               4186.607279464874+i*1784.0262722568;4186.607279464874-i*1784.026272256801;3445.944679249668;
> #                               6825.617444279969],-1.000000000000004,"n")
781a789,792
> CS_DARM_CTRL_DELAY 0 21 4      0      0 delay      -1.546135588596514307051190e-06  -0.2162350800222442   0.0174983720374032   3.6925487491949833  20.2756341393376971
>                                                                  -0.0968548102303960   0.0247770268878642  -3.9090569933487744  40.3599675023881517
>                                                                   0.1821175467962750   0.0493202823215208 -12.3574398555497709  57.1481734336473011
>                                                                   0.4517859341185633  -0.0723061411729160  -6.2482373805309441 -13.8300839151207331

The delay filter has about the right delay (attached, -168.729deg @ 1kHz ~7.68 clock cycles for 2^14Hz sampling).

The new file was installed, the new filter loaded.

The schedule file and waveforms have been uploaded to the h1hwinj1 machine and I asked Gary (at LLO) to ask Patrick to reload the INJ TRANS guardian.

Reload completed.

   At 11:55 (04:55) IFO dropped out of Observing due to SDF Difss with CALINJ SDF Diffs. At 12:01 (05:01) Accepted SDF Diffs and put IFO back into Observing Mode. 

   It may be unrelated but lostlock at 12:04 (05:04). 

It would be better to un-monitor these channels in the SDF as they change when we carry out hardware injections. The injections should have finished at 5:01:30 PDT, which doesn't quite match up with the lockloss at 5:04.

These injections do not appear to have been successful in as much as H1:DMT-INJECTION_DETCHAR:1 was not active during any of the injection times. We're re-scheduling these for the morning of Fri March 29, 2019.

To un-monitor the channels open the SDF screen for CAL-INJ (while in Observe), set the "TABLE SELECTION" to "FULL TABLE", set "MONITOR SELECT" to "ALL" (this part may not me necessary). Channels with the final green tab lit up are currently being monitored. To un-monitor a channel click on the "MON" tab and hit the "CONFIRM" button. A list of the channels that we want the monitoring removed are shown in this screenshot (the ones with the yellow box in front of MON).

I'm sorry, I got distracted and didn't get around to uploading the schedule file and waveforms in time for these first set of injections. Hopefully, the next set will go through.

These injections do not appear to have been successful in as much as H1:DMT-INJECTION_DETCHAR:1 was not active during any of the injection times. We're re-scheduling these for the morning of Fri March 29, 2019.

Forgot to post some SDF diffs from last night (2019-03-26) that I accepted to get to observing. These are about the Cal line magnitude for EX (first screenshot), and some ALS_Y differences (second screenshot).