40m QIL Cryo_Lab CTN SUS_Lab TCS_Lab OMC_Lab CRIME_Lab FEA ENG_Labs OptContFac Mariner WBEEShop
  Cryo Lab eLog, Page 7 of 60  Not logged in ELOG logo
ID Date Author Type Category Subject
  2699   Wed Apr 14 17:15:23 2021 shrutiDailyProgressPSOMARoC measurement

Attachment 1:

Result. Measured radius of curvature seems to be closest to 0.4 m convex.


Attachment 2:

Beam profile bitmap images. Widths taken read from the image.


Attachment 3:

Code used to plot Attachment 1.

Attachment 1: RoCis0_4mConvex.pdf
Attachment 2: 20210412.zip
Attachment 3: test_roc.m
%% beam fit

close all
clear classes
clear all

zScan = [3 6 9] * 0.0254 - 0.023 + 3.7e-3; % inches to m
xradiusN = [874 1021 1186] / 2 * 1e-6; % um
yradiusN = [838 1008 1178] / 2 * 1e-6; % um

... 197 more lines ...
  2696   Tue Apr 13 11:31:37 2021 aaronDailyProgressGeneralsome parts to W bridge

enter: Tue Apr 13 11:31:35 2021

Picked up some boxes from Downs:

  • pin connectors for our EOM
  • 0.5" optic mount from newport
  • some batteries and battery holders for EE

Mostly just math.


  2695   Thu Apr 8 22:47:03 2021 shrutiDailyProgressPSOMACavity locking and temperature scan

[Rana, Shruti]

Today Rana pointed out several improvements to the setup including fixing some finicky knob situations. The cavity and setup are now in a good state with monitor and PD connected, although we still have not seen resonance (Will be Attachment 2 soon).

According to the calculations in Attachment 1 a 1V change in the 'TEMP TUNE' input of the ITC-502 temperature (and laser current) controller would correspond to a change of 1 GHz in laser frequency at the current operating point (8 kOhms 'T_ACT' as displayed).

We send in a 1 Hz triangle wave with V_pp=1V into the TEMP TUNE input and make tiny changes to the alignment while monitoring the monitor for flashes and the DC reflected power for a steep drop.

RXA edit:

  • We measured the input coupler transmission to be 2.5% for S-pol, so the cavity Finesse = 2*pi/(1 - T) ~ 250
  • From the overhead picture, you can see that the cavity length is ~15-16 screw holes, so L_{RT} = 16" = 40 cm. So the FSR = c/0.4 ~ 1 GHz.
  • So, assuming the cavity linewidth is then FSR/Finesse = 3 MHz, we want to make sure that our scan velocity is no faster than ~ 1 GHz/s, so that we are able to see the resonances easily on a 1 s oscope trace.
  • The diamter of the beam transmitted through the curved mirror of the cavity is ~3 cm on the monitor.
  • The mode-matching is still far off, since the beam after one round trip is much bigger than the input, but I estimate the mode matching is ~30-50%. Good enough for getting some flashes.
  • It would be good to find a mount for the 1" optics that allows us to cleanly get the 4-ports without clipping.

Other To Do

  • Get a new power adapter for the camera (weird flashes occur when shaking it, confuses the alignment/locking process)
  • Find/purchase a thread adapter to attach longpass filter to Watec camera
  • Eventually organize and safely store all ND filters and other optics in the cabinet
Attachment 1: TemptuneCalc.pdf
Attachment 2: Untitled.png
  2694   Tue Mar 30 15:00:56 2021 aaronHowToGeneralWhere is three corner hat data? Cryo repos?

Code and data for the cryo lab are in git.ligo.org within the Cryo Lab group

  • The data repo holds version controlled data, which should all be added using git lfs. File names should contain a date or other identifier that maps onto an associated elog entry.
  • The scripts repo contains scripts and notebooks for analyzing data
    • The notebook ThreeCornerHat.ipynb is what I used to generate three corner hat plots above
    • The notebook currentNoise.ipynb is what I used to look at the current noise.
    • Once I'm satisfied with these, I'll make a .py file that performs the measurement and generates plots in one go and add the useful sub-routines to labutils
  • Moku scripts
    • To take these measurements, I've been using the scripts in labutils/moku
    • The moku repo also includes liquid instruments' lireader repo for converting their proprietary .li binary format to a numpy object.
  2693   Tue Mar 23 11:07:56 2021 shrutiDailyProgressLab Worklab temperature logging

Lab temperature is now stabilized. The particle counter is now logging the temperature, but not the AD590 which records X1:AUX-LAB_TEMP_F.

The AD590 history also seems different from the previous record.

Attachment 1: Screenshot_from_2021-03-23_11-06-46.png
Attachment 2: Screenshot_from_2021-03-23_11-04-15.png
  2692   Mon Mar 22 21:41:09 2021 ChrisDailyProgressLab Worklab temperature logging

standalone_edc now works, and data from the temperature sensors (and other aux epics channels) is again being recorded.

The version of standalone_edc we had from RCG 4.0 silently fails to acquire data, unless symmetricom timing hardware is present. We don’t have such hardware in the cymac. I upgraded cymac1 to RCG 4.1 in order to get a newer version of standalone_edc. It can now use the IOP model as a timing source (--sync-to=x1iop_daq was added to the argument list). All models were rebuilt and reinstalled after the upgrade.

To minimize the chance for confusion, I removed the old nds/daqd service files, and archived the /opt/rtcds/tst/x1/target/fb directory. None of that is used anymore.

  2691   Mon Mar 22 13:28:32 2021 aaronDailyProgressLab Worklab temperature logging

Attachment 1 shows the most recent temperature data available on nds.

Separate from the dust monitor, we have an AD590 set up as a temperature sensor read by the Acromag, which interfaces with cominaux. This is the channel that cymac is supposed to write to frames as X1:AUX-LAB_TEMP_[K, F, C], but the value read by cominaux is currently (and has been) 0. I've confirmed the signal is reaching the acromax ADC on the correct channel.

The dust monitor is a particle counter (GT-526) that also shows temperature and relative humidity, and is read by cominaux over ethernet. Despite an "error reading and converting data from the counter" shown in cominaux:~/services/particlecounter.log between Feb 3 and today, caget'ing the channel C5:PEM-COUNT_TEMP (or the particle counts) returns the expected value (89-90 F). However, cymac has not been logging these frames and I can't access live or lookback data through nds.

I see that daqd.service and nds.service on cymac1 are both dead linked to /etc/systemd/system/[nds, daqd].service. Neither service script still exists in cymac1:/etc/systemd/system, only the 'old' version of these scripts. I think binaries for nds and daqd now live in /opt/rtcds/tst/x1/target/fb, and the standalone_edc service is meant to pull data from epics.

If I run standalone_edc, I get an error 'edc.ini' failed to open, and there are no channels to record; this service is not running on cymac1. /opt/rtcds/rtscore/edc_status.json is empty.


What happened to the temperature logging? Has the CDS system stopped recording data due to model changes, or is it just the python for the dust monitor?


Attachment 1: Screenshot_from_2021-03-22_14-08-10.png
  2690   Mon Mar 22 12:05:41 2021 ranaDailyProgressLab Work 

What happened to the temperature logging? Has the CDS system stopped recording data due to model changes, or is it just the python for the dust monitor?

  2689   Sun Mar 21 12:52:43 2021 ranaHowToNoise Budgetphase noise

Youtube on phase noise in osc (https://youtu.be/wByzymJ0Ppc)

  2688   Fri Mar 19 11:08:15 2021 aaronDailyProgressLab Work 

Lab temperature is high again (85 at the particle counter), but apparently hasn't been logging. Attachment 1 is the set of status screens -- still no GPS, and yellow CFC.

incoherent three corner hat

Measured beat note with 3 lasers incident on 1611, RF out sent through SLP-21.4 to moku phasemeter

  • python [path_to_scripts]mokuPhaseMeterTimeSeries.py -c ch1 -d 450 -i -s veryfast --useInternal --fileType bin --altFileName "L1L2_note"
    • where L1L2 are the lasers being measured (SX, NX, NS)

  • DC level on 1611 0.5 mW

  • In all three measurements, the phaselock was lost when the beat note drifted too close to DC. I'm trying again with the SLP-100 to give myself some more range.

  • really need to put these fibers in a box

While waiting for the measurements,


Attachment 1: Screenshot_from_2021-03-19_11-09-04.png
Attachment 2: lasers_ASD.pdf
  2687   Thu Mar 18 18:21:36 2021 aaronDailyProgressLaser3CH setup

Set up the three corner hat in the same optical configuration as previously.

All three beat notes are on the same 1611-FC, then sent to a 4-way RF splitter. RF components we could use more of:

  • SMA 50 Ohm terminators
  • low pass filters above 50 MHz
  • BP filters wider than 6 MHz (or some HP to match the LP)
  • N to SMA adapter
  • 1->2 SMA terminated RF splitters


Splitter output sent to (fig pending, I have it hand-scrawled):

  1. 50 Ohms
  2. DC block -> SHP-100+ -> pomona box 135 MHz lowpass (47 pF cap to ground) -> mixer R (for marconi)
  3. SHP-150 -> Moku 2
  4. SLP-100 -> Moku 1

Struggled to get the Marconi locked, though I did manage a couple times... tomorrow.


  2686   Thu Mar 18 16:44:33 2021 shrutiDailyProgressPSOMAcavity alignment, RF oscillator power and frequency

Proceeded to align the cavity as outlined in elog 2685.


I also measured the output of OCXO using the Moku; all channels have 10dB attenuation and 50 Ohm impedance.

Attachment 1: 32.7 MHz; Channel 1 (red) is 'TO EOM', Channel 2 (blue) is 'TO LO'.

Attachment 2: 33.59 MHz; Channel 1 (red) is 'TO EOM', Channel 2 (blue) is 'TO LO'.

Attachment 3: Channel 1 (red) is 'TO LO' at 32.57 MHz, Channel 2 (blue) is 'TO LO' at 33.59 MHz.

The data files for all the spectra can be found here.



Attachment 1: MokuSpectrumAnalyzerData_20210318_162955_Screenshot.png
Attachment 2: MokuSpectrumAnalyzerData_20210318_162552_Screenshot.png
Attachment 3: MokuSpectrumAnalyzerData_20210318_162358_Screenshot.png
  2685   Thu Mar 18 10:08:40 2021 shrutiUpdatePSOMACavity design

Using the latest beam profiles in Attachments 3,4 and the design in Attachment 1, I calculated the mode-matching solutions found here. The one I decided to go with is in Attachment 2.

Attachment 1: CavityDesign_L40cm_R2m.pdf
Attachment 2: MM_L40cm_R2m_RA.pdf
Attachment 3: North_corrected.pdf
Attachment 4: Sourth_corrected.pdf
  2684   Wed Mar 17 12:55:42 2021 aaronDailyProgressstuff happensnot much

enter Wed Mar 17 12:55:44 2021

Grabbed the Marconi from cryo lab.

I left some of the current noise data on the moku ipad, so I transfer that over to spirou to plot the current noise.

Thought about filters for a while. Stuffed an RC low pass in a pomona box, but realized an LC would be better for the job.

exit Wed Mar 17 17:13:55 2021

  2683   Mon Mar 15 12:11:51 2021 shrutiNotesLaserPower vs. Temperature plots for Rio PLANEX

South Laser [Attachment 1]

The power dependence on temperature at different diode currents:

  • The plots are linear in sensor resistance on the x-axis
  • (Top figure) At 130 mA, while the sensor resistance decreased, an abrupt change was seen at 6.73 kOhms, and while increasing it was at 8.27 kOhms. At other regions it is locally reversible, i.e., going back and forth causes the same curve to be plotted.
  • (Top figure) This trend is repeatable, as the black lines show the results of the same change performed a second time.
  • (Middle figure) At around 120 mA, three such abrupt changes was seen in the range the data was taken.
  • (Bottom figure) Similar plot at 100 mA.

Data and notebook for plotting in Attachment 3

Conversion to Temperature

This notebook uses Rtyp values mentioned on the Rio Planex Datasheet found on the wiki documents page to obtain the plot and fit in Attachment 2.

That was used to make the temperature in deg C axes in Attachment 1.

Attachment 1: South.pdf
Attachment 2: RtoT.pdf
Attachment 3: SouthData.zip
  2682   Thu Mar 11 13:56:47 2021 aaronDailyProgressElectronicscurrent noise

enter Thu Mar 11 13:56:46 2021

  • On Busby low noise box AC coupled input (logged by to 1 MOhm moku channel 2)
    • measure voltage noise of a 20.5 Ohm resistor for 28 minutes
    • measure voltage noise across the same resistor in driven by ITC502 at 119.38 mA, same duration
    • python [path/to/moku]mokuDataLogger.py -c ch2 -i -d 1680 -s 64e3 --useInternal --fileType bin --MOhmr2
    • started to make a box for sensing current noise... but perhaps this is not necessary. To measure 10 kHz/rtHz frequency noise on our 147 MHz/mA diode (-> uV/rtHz), even the metal film resistors (1% tol) with the steepest temperature coefficients on digikey (~100 ppm/K) could allow 10 mK/rtHz temperature fluctuations of the resistor. Still, the room is drafty, so I also measured the noise across the resistor driven by two AAA batteries (2.4 V total, for about the same current as before).
    • What is the voltage across the diode during operation, and can I check this at some of the butterfly pins? Does the current supply remain consistent operating at the same current but a different voltage? And, shouldn't a 'low noise current supply' be quieter than a battery (at least at the frequencies where it matters) anyway?
  • While waiting for the data, assembled the wire shelf on the West wall of PSOMA, between the vacuum cabinet and He cryostat.
    • Electronics on the 2nd shelf have rails for earthquake protection. We have sufficient rails for all 4 shelves, but I left them for another day.
    • Zip-tied the upper shelves to the piping on the wall. Seems sturdy.
    • Feet have plastic sliders to dampen the load on the floor. We also have label holders in need of labeling.
    • Photos are in the ligo.wbridge google drive and some attached.

exit Thu Mar 11 23:11:33 2021


Attachment 1: wire_shelf_5.jpg
  2681   Wed Mar 10 16:03:19 2021 aaronDailyProgressElectronicscurrent noise

enter Wed Mar 10 16:02:33 2021

I have "The Busby Low Noise Box" from the 40m and will be measuring current noise of our laser drivers... but today I mostly played around with liso. If anything useful, getting more accustomed to vectorizing python functions.

exit Wed Mar 10 21:07:52 2021

  2679   Tue Mar 9 14:03:12 2021 aaronComputingDAQoma model
  • installed cds-workstation on spirou (for some reason, was being held back by an incorrect version of python3-gpstime. Manually installing the correct version of that package let me apt install cds-workstation).
  • fixing bugs in x1oma model, following instructions on dcc.
    • Need at least 2 filter modules to build the model (it was just a dummy model with ADC and DAC).
    • 'site' designator is deprecated, I've replaced it with 'ifo=X1'
    • Need to use a different testpoint node (dcuid) from the existing models. 
    • from cymac1
      • rtcds build x1oma
        rtcds install x1oma
      • restart rtcds. The x1oma model wasn't added to the startup sequence or status screens.
        • Manually running rtcds start x1oma completes successfully, and /opt/rtcds/tst/x1/medm/x1oma contains the usual .adl files.
        • Now, after building and installing x1oma is in the rtcds startup sequence, but doesn't load successfully
        • Possibly just built the model with an incompatible version of matlab. Gaston is running 2015b, which works historically, and is also the latest Matlab I see mentioned in RCG release notes.
          • When building the x1oma model (with either Matlab 2020 or 2015 versions), I get a warning before the model eventually compiles successfuly
            • cat: '/opt/DIS/lib/modules/4.19.0-6-rtcds-amd64/*.symvers': No such file or directory
            • This warning is also there when I compile other models (like x1siq), even though they were built with Matlab 2015b.
          • Saving the model as a 2015b compatible model does not lead to an installed x1oma
        • I'm using some 'already in use' DAC channels in x1oma. In fact, the only 2 free DAC channels (of our 16 channel fast DAC) are 1 and 2. 0 is occupied by SIQ ESD; 3-7 occupied by Si cantilever Qs; and 8-16 by cryo cavs. I swapped the OMA to use DAC channels 1 and 2, but this x1oma still fails to load (and doesn't auto-populate status medm screens).
      • This has to be lower than the x1oma model. rtcds status shows several kernel modules are not loaded (dis_*).
        • On cymac1 I apt install advligorts-cymac, I'm told the package was not fully installed and it updates to 4.1.0-1 (was 4.0.1).
        • I confirmed that these modules are running on one of the 40m frontend machines.
        • I reinstalled cds-workstation and advligorts-cymac (no change)
        • rebuilt and reinstalled all models. No change, in particular status screens not updated with x1oma model.
          • Oddly, x1oma is no longer in the startup sequence or list of models...
Cymac isn't crashing, so I'll vacate the lab and continue from home
  2678   Tue Mar 9 13:57:38 2021 aaronDailyProgress update

enter Tue Mar 9 13:57:31 2021

  • picked up a fiber microscope and cleaner from Aidan's office
  • I'd like to record data with cymac1 to assist in the 'coherent 3 corner hat' measurement. To that end, I'm setting up the x1oma model (again).
    • Not much success

exit Tue Mar 9 16:42:41 2021

  2677   Mon Mar 8 12:11:52 2021 aaronDailyProgressLab Workallan covariance

I've attached a new diagram for this measurement. Instead of using two RFPD, I'll mix all three beams on one RFPD and separate the notes with passive filters. I'll use the N and S PSOMA lasers as reference clocks to measure the frequency noise of the Teraxion laser.

  • Rearranged the fiber components as shown in the diagram, trying to minimally disrupt the PSOMA path
  • Connected an SMA power splitter to some HP and LP BP minicircuits filters
  • Moved the ITC510 combination controller that was controlling cryo cavs E laser TEC (not in use) over to PSOMA rack. Now, both PSOMA N and S lasers are being controlled by separate ITC502 controllers; PSOMA N TEC is still on the TED 200 C.
    • Labelled disconnected cables (cryo cav E TEC and E temperature tuning) so we can find them again
    • Now, PSOMA S laser LD and TEC are being controlled by our ITC502, N laser TEC is controlled by TED 200 C, and N LD controlled by ITC510.
    • ITC510 has a larger range but worse quoted current noise properties than ITC502. In principle this shouldn't matter, as the measurement will be insensitive to frequency noise on the N and S lasers.
    • I set the LD current limit on ITC510 to 144.9 mA. It was at 125 mA.
  • DC power levels in measurement configuration
    • 550 uW total
    • 150 uW due to PSOMA S laser
    • 52 uW due to PSOMA N laser
      • I realized I sent 0.6 mW in the wrong direction through the PSOMA N EOM. After swapping the connectors (so 0.663 mW is entering the 'in' port of the EOM), I get 304 uW out of the EOM.
        • This is worse than the specified 2.7 dB insertion loss, which I measured in January.
        • I think Rana's measurement last week was transmitting Teraxion through this EOM.
        • I've taken the N EOM out of the path for now. Hoping I haven't damaged this.
    • 348 uW due to TeraXion laser
  • RF beat notes
    • I tuned the N and S LD currents to achieve the desired beat note spacing
      • S-X beat note (-12 dBm) at 32.5 MHz. I have a 27-33 MHz bandpass filter. I was using a 22 MHz lowpass filter, but realized I could separate the other two peaks a bit more by increasing this beat frequency and using a BPF.
      • N-X beat note (-27 dBm) at 98 MHz. I'm using a 90 MHz highpass filter, and want to be somewhat away from the corner where the phase changes rapidly.
      • This puts the NS beat note (-59 dBm) at about 65.5 MHz, outside the bands of both passive filters.
    • After fixing the beat frequencies, I send 1611's RF out to a splitter, followed by passive filter, and onto the two moku input channels in phasemeter mode.
      • The filters are
  • Measurement
    • Cleared moku ram with 
      • python [path_to_scripts]mokuCleanRAM.py -i
    • Left the room, and started a measurement from spirou with
      • python [path_to_scripts]mokuPhaseMeterTimeSeries.py -c both -d 450 -i -s veryfast --useInternal --fileType bin
      • Use DC coupling to avoid the 100 Hz corner for Moku's AC coupling.

      • I observed substantial frequency drift (>30 MHz) during the measurement time (possibly from cycling the lights earlier). I took a second measurement with lights on after keeping the lights on for ~10 min, haven't yet seen if there was further drift. 

    • not sure why the moku has a new IP? Also sometime during my measurements the IP address was reassigned back to (from earlier today), though I didn't change its ethernet port. 

    • To measure the high frequency noise (> 7 kHz), I ran

      • python [path_to_scripts]mokuPhaseMeterTimeSeries.py -c both -d 90 -i -s ultrafast --ac --ac2 --useInternal --fileType bin
      • This time I did ac couple the inputs

Update: Meh, the frequency drifted too much to be useful (20 MHz over 7.5 minutes measurement time).

Attachment 1: 9EEB7450-A584-4085-8F4F-0F70F164AF13.jpeg
  2676   Mon Mar 8 12:11:15 2021 aaronDailyProgressLaserbusby, sr554 to cryo lab

enter Mon Mar 8 12:11:07 2021

Making the Allan covariance measurement today.

Picked up the busby low noise box and SR554 from gautam at 40m, which I'll use to measure the current noise of our laser drivers.

I returned the fiber components to the previous state after measurement (PSOMA S beam incident on cavity, N beam to the other fiber launch, both beams pass through their respective EOM).

exit Mon Mar 8 19:22:50 2021

  2675   Sun Mar 7 19:32:32 2021 ranaDailyProgressElectronicsPSOMA S mA->MHz, ITC 502 current noise

something weird in the current noise plot - left axis log, right axis linear

also, something is up with the overall scale. With 5 nV of noise on a 20 Ohm resistor, the noise floor above 10 Hz should be ~250 pA/rHz. I recammend plotting the noise floor of the measurement apparatus with a 20 or 50 Ohm input term on the 560 (and grab the Busby Low Noise box from the 40)

  2674   Fri Mar 5 14:36:00 2021 aaronDailyProgressLab Worklearned something about measuring clocks

enter about 2pm


  • chatted with Radhika, who is going to get our particle counter logging again.
    • We found the services script that Duo and others had worked on in controls@cominaux:~/services, but many of the corresponding files in /etc/systemd/system were null linked.
    • In the process, I remembered that I'd tried to initialize the x1oma model again, but was unsuccessful. Consequently, cymac crashes during starting 'all' frontends.
    • Regardless, one could always caget the slow channels to see what's up (eg caget C5:PEM-COUNT_05UM returns the 5 um particle count). The data may not be saved to frames by cymac1, but at least the particle counter seems to be talking with cominaux.
  • untangling cymac1
    • Undid the steps I had taken to try to install x1oma:
      • Removed x1oma entry from /etc/advligorts/systemd_env
      • removed x1oma entry from /etc/advligorts/master
      • Manually restart rts-local_dc and rts-daqd services using systemctl
      • Rebooting cymac after these changes seems to have resolved the issues with frontends, but I still have white boxes for the GPS signal on X0DAQ (attachment 1).
    • meanwhile spirou's cds-workstation package is broken (no medm, for example).

Three Corner Hat to Allan Covariance [aka Cross-variance, aka Groslambert Variance]

A conversation with Rana had me thinking again about the 2-channel, 'coherent hat' measurement, where one compares the beat note between a laser of interest (X) and two other lasers (N, S). It felt like there should be enough information there. If I know $(\phi_{X}-\phi_{S})^2\equiv \Delta_{SX}$ and likewise for $\Delta_{NX}$, I ought to know the difference between I and Q.

Indeed I found this paper comparing the two methods... looks promising! The authors have further work quantifying the confidence intervals for the so-called Allan covariance.

The method is a natural extension of the Allan variance, where the phase noise on an oscillator as a function of frequency (1/t) is estimated by the autocorrelation of the oscillator's frequency at varying time separations (t). "Allan covariance" then is the same method applied to a set of different oscillators, where the oscillator frequencies are cross-correlated. I'll adapt the equations below from the paper to use more familiar notation.

Let $\phi_i$ denote the phase error (additonal phase relative to the carrier frequency) on the i'th laser (N, S, and X, for our cryo lab measurement). We want to estimate the phase noise of the TeraXion laser (\phi_X) by measuring the phase errors of the beat note between X and the N and S lasers (\phi_{NX}, \phi_{SX}). To see why the covariance of $\phi_{NX}$ and $\phi_{SX}$ gives a good estimate of the variance of $\phi_X$, consider

\phi_{NX} = \phi_N -\phi_X+\theta_{NX}

\phi_{SX} = \phi_S -\phi_X+\theta_{SX}

\implies \langle \phi_{NX}\phi_{SX}\rangle = 2\langle\phi_N\phi_S\rangle + 2\langle\phi_X^2\rangle -\langle\phi_N\phi_X\rangle - \langle\phi_S\phi_X\rangle + \langle\phi_S\theta_{NX}\rangle + \langle\phi_N\theta_{NX}\rangle - \langle\phi_X\theta_{SX}\rangle - \langle\phi_X\theta_{NX}\rangle

Since the expectations of all the uncorrelated terms are 0, we can conclude that as long as the measurements were taken simultaneously,

\langle\phi_X^2\rangle_\tau\to \langle \phi_{NX}\phi_{SX}\rangle

We should (and the paper does) use the phase errors $\phi$ to define a corresponding frequency error at each timestep $k$ set by our sampling period $\tau$, $f_i(k) \equiv \frac{\phi_i(k)-\phi_i(k-1)}{\tau}$. The frequency error is then $\Delta_i(k) = f_i(k)-f_i(k-1)$. The Allan variance is the variance of that frequency error, while the Allan covariance is the covariance of $\Delta_i$ with $\Delta_k$. Averaging the residuals makes explicit the dependence on $\tau$ and according to David Allan "effectively modulates the bandwidth in the software allowing one to distinguish between white-noise phase modulations (PM) and flicker-noise PM." I'm a bit unclear why it is necessary to take the second difference (that is, correlating frequencies separated by \tau rather than directly phases separated by \tau), but I think this paper from Allan goes into a bit more depth. My best guess is, it's because flicker phase noise (which asymptotes with infinite slope at low frequencies) will show up as a DC offset that is consequently not averaged away in terms like $\langle\phi_N\phi_S\rangle$. Taking a second difference means rather than negating this DC offset once (as I did above by defining $\phi_i$ as a phase error), phase noise below the sampling rate is discarded with each measurement time step.

With the frequency noise estimate from Allan (co)variance and some knowledge of the power law dependence of the noise, one can derive a corresponding spectral density. The Allan (co)variance with a particular sampling rate \tau corresponds to sampling the frequency-domain noise distribution with a function related to the Fourier transform of the chosen time-domain estimator. See the paper on converting between time and frequency domain noise estimates from Allan (linked above and here).


Very exciting! Sketch of the measurement with equipment on our tables in attachment 2. Whoops, we only have a free space 1811, not a fiber coupled one. Between that and the floppy path to PSOMA table, I'm going to think further on this measurement and try it next week. Maybe I can pick up some spare fiber parts from the 40m when I stop by to search for a transformer. In the meantime, yesterday's log has been updated with the current noise spectra.

exit Fri Mar 5 20:47:39 2021

Attachment 1: Screenshot_from_2021-03-05_15-14-24.png
Attachment 2: 8B3029C0-A1D3-4044-B238-68D545187BE5.jpeg
  2673   Thu Mar 4 13:37:14 2021 aaronDailyProgressElectronicsPSOMA S mA->MHz, ITC 502 current noise

enter Thu Mar 4 13:37:09 2021

Picked up from Downs (see PSOMA hardware inventory for specifications):

  • label maker and tape
  • 4x longpass filters

current to frequency

  1. Plug in windows laptop to Teraxion driver box
  2. Turn on driver box, then 'connect' to driver from the GUI. The interlock automatically switches to 'on' and the laser powers up at 1550.12 nm. I didn't change any settings on the Teraxion laser.
  3. Power from the blue fiber carrying Teraxion beam to PSOMA table puts out 0.6 mW. I replace the PSOMA N beam with the Teraxion beam just before the Faraday isolator.
  4. With the PSOMA South laser current drive at 149.36 mA and TEC at 9.305 kOHm, the beat is at 2.078 GHz (HP 8560E). I tune the S laser to 120.74 mA to get the beat to around 50 MHz.
    • There seems to be some thermalization happening with either or both lasers, because the beat note is moving monotonically at about 200 kHz/s. I'm searching for an SR554 in the meantime.
    • Didn't locate a transformer amplifier. I took the following data by hand, but as you can see by the apparent hysterisis, the laser frequency is still monotonically changing.
    • Second set of data after the beat note settled down
current (mA) 118.30 118.11 117.68 117.56 117.35 117.50 117.59 117.65 117.74 117.83 118.04 118.11 118.29 118.47 118.55 118.62 118.76 118.89
beat frequency (MHz) 148 120 58 42 9 35 47 55 69 83 114 123 149 175 186 197 2162 235
current (mA) 117.89 117.73 117.56 117.41 117.22 117.13 117.07 117.01 117.13 117.23 117.35 117.63 117.76 117.95 118.06 118.11 118.20 118.26
frequency (MHz) 137 112 88 66 38 26 17 7 26 40 58 101 120 148 164 170 183 192

ITC 502 current noise

  1.  Though I don't have a transformer, I expect about 1 nA/rtHz from the ITC502 I'm measuring. Driving a 20 Ohm dummy load with a , this should be 20 nV/rtHz at the input of the SR560 -- well above its 4 nV/rtHz noise floor. Seems fine for now.
    • Repeating the measurement here, with the following pin mapping
      • 1, 5 have a dummy load to lift the interlock
      • 3, 7 are sending current across a 20 Ohm dummy load
      • SR560 is on battery power measuring in AC-coupled, low noise mode with G=100 and a 100 kHz low pass filter (in addition to the implicit 0.03 Hz high pass due to AC coupling).
      • Moku is set to 'data acquisition' mode, and the signal is sent to DC-coupled 50 Ohms with 1 Vpp range. Acquiring 7 min data at 256 kS/s.
    • After setting up the moku for remote access (below), I killed the lights, left the lab, and ran the following from the directory where I wanted the data file stored:
      • python [path_to_script]mokuDataLogger.py -c ch2 -i -d 420 -s 256e3 --useInternal --fileType bin

Update: Plotted the current noise ASD, using the 'modified welch' method found In labutils/moku/modifiedPSD. I used 'median' averaging for the welch spectrograms, and the script from Anchal and Scott improves averaging at high frequencies by building the spectrum frequency decade by deacde, and calling welch with as much averaging as possible for the chosen decade. Shaded region is the 1 sigma confidence around the median. The right axis shows the current noise referred to the S laser's frequency noise, assuming 147.86 MHz/mA.

Hm, not sure if it makes sense yet.

Update: No, it did not make sense. I converted A^2/Hz to A/rtHz as if my data were already in dB, but they were in magnitude units; and the twinned y axis for frequency noise was not log scaled. I've updated the plot attached and the script in the git repo, and added a line where the noise floor from voltage noise of SR560 across 20 Ohms should be (not measured). I'm still surprised to see current noise so far above the nominal noise floor of the SR560.


  • Opened one of the SR560 lsthat was constantly hitting overload in either DC or GND mode.
    • front panel offset was turned too far. Adjusted offset as suggested in calibration guide until pin 14 on U407 was at the same voltage as unit's ground (was ~6V).
    • Unit still overloading, identified that one leg of U106 (front end op amp) was not at 0V. Manual suggests this constant overloading is a sign U106 is busted.
  • uninstalled pip and conda on spirou, which were improperly installed by sudo user. All python on spirou now through controls user installed anaconda. Also initialized moku and psoma environments with some of the usual packages. Pip is installed in the virtual envs (not in base), and all pip installations should occur after conda package installations.
    • sudo python -m pip uninstall pip
      conda activate moku
      conda install pip
      pip install pymoku
      moku update fetch
    • sudo apt-get install libavahi-compat-libdnssd1
  • connected moku to ethernet on router port
    • first need to configure moku to accept ethernet connection
      • when connecting via ipad, click gear icon before 'go'ing to splash page listing available instruments.
      • Turn 'on' connection via ethernet. I used DHCP, and was assigned IP

Returned ITC502 to driving S laser, still below 120 mA.

exit Thu Mar 4 23:04:28 2021

Attachment 1: mA_to_MHz.pdf
Attachment 2: 002F0263-0615-4566-9192-F57A14C8B543.jpeg
Attachment 3: ITC502_ASD.pdf
  2672   Thu Mar 4 09:44:24 2021 aaronMiscNoise Budgetdata files for laser noise

no data for current driver, three corner hat is on the cryo lab data repo with git lfs.

  2671   Wed Mar 3 14:52:43 2021 ranaMiscNoise Budgetdata files for laser noise

How about posting the data files for the driver current noise and for the 3 corner hat?

for data files, you can just upload here. If you want to put it in gitlab, you should use the GIT LFS paradigm

  2670   Tue Mar 2 13:50:35 2021 aaronLab InfrastructureDrawingsPSOMA table clean enclosure

enter Tue Mar 2 13:50:31 2021

Just finalizing the layout and dimensions for a table enclosure. 

Please take a look at the attached drawings. The drawings from F&L, measurements and photos of the room, and notes are linked on the wiki. We need to finalize

  • Height
    • The drawing assumes we open the cantilevers cryostat at the midsection (which lets us open with only 11.5" height increase, rather than doubling the height to 38" by opening from the top). I left 12" for rigging a pulley, which seemed reasonable based on my trial attaching the crane to the cryostat -- but I didn't actually open up, since I don't want to disturb the suspended optics.  
    • The overall height would put any overhead electronics just below 7'. This is just close enough for me to reach knobs without a stool, and the space for electronics over the enclosure is constrained by the HEPA FFU. Should we plan for a shelf scopes and other electronics inside the enclosure, or is this too dusty a solution?
    •  It will be difficult to change the over-table clearance later, because the panel doors will only extend to the surface of the table. If we do allow a shelf inside the enclosure and can keep the HEPA FFU away from the ceiling beam, it may behoove us to increase the height a bit more in case someday we want a taller cryostat or elaborate piping a la green monster.
    • QIL uses a MAC 10 HEPA filter. I think we want the room-side replaceable version, and I assume those dimensions in the drawing.  
    • They also have a 'low flow, low vibration' version of this already "low vibration" FFU. They specify that the original version has 51 dB acoustic noise during normal operating conditions, but not how much reduction to expect with the low vibration version. The original version provides up to 610 CFM, while the low flow model provides up to 490 CFM. Since we would almost certainly turn off the FFU entirely during quiet measurement, perhaps this doesn't matter. 
  • Outlets 
    • Drawing puts 10 outlets on a power strip, one strip on each long side of the table on the top rail. This was based on a 'rule of thumb' I found online to give 20% overhead on the wall outlet amperage (we have 120V/20A outlets), and an assumption that what we plug in could draw up to 1.5A in steady operation. Reasonable, more fewer outlets?
  • Cable passthrough
    • Do we need to request openings for later installation of the the "40m EX/EY end table style" cable passthrough? Or could we cut these openings ourselves later?
  • Rail around enclosure
    • I've set the height of this rail to protrude 1" above the tabletop to help avoid leaning on the table. Since we are using panels, not curtains, that means the enclosure will extend 2" below the surface of the table.

Exit Tue Mar 2 17:55:46 2021

Attachment 1: table_enclosure_side.pdf
Attachment 2: table_enclosure_end.pdf
  2669   Fri Feb 26 13:24:31 2021 ranaNoise HuntingLasersetup beatnote to Teraxion laser

Aaron and I set up the beatnote measurement between the Teraxion and (either south or north) RIO:

  • mounted all the Teraxion stuff on the NW corner of the Green Monster table.
  • Used a high power (300 mW!!) fiber coupler to take ~1% of the beam and send that to the PSOMA table on the overhead PM fiber.
  • Aaron measured ~0.2 mW at the PD ?
  • The beat note amplitude is ~-15 dBm out of the New Focus 1611 (1000 V/A AC transimpedance)
  • Split the output with a RF splitter (NOT a BnC tee) to the 3 GHz spetrum analyzer and rack mounted Moku.
  • Recorded some phase meter data with the beat at ~148 MHz.

Turned Teraxion laser OFF around 9 PM. Analysis in progress if I can figure out how to get the data.

Next we'll have to do some careful temperature tuning to get the RIO laser into a low noise state and simultaneously have a beatnote < 200 MHz.

  2668   Fri Feb 26 11:52:31 2021 ranaComputingDAQCymac & Matlab

I would advise against using the latest Matlab. The RT CDS stuff uses Simulink for editing the .mdl model files, but they switched to .slx a few years ago and I,m not sure if the modern one is supported by CDS yet. Best to ask Chris/Jon what the officially supported one is and make sure thot the cryo workstations use that one by default.

  • downloaded Matlab 2020b (!!) from Caltech imss onto Spirou
  • When I open it, am alerted that the following directories are missing:
    • /opt/rtcds/userapps/local


  2667   Fri Feb 26 10:26:05 2021 ChrisDailyProgressDAQ 

To help with the missing files and permissions issues, I did the following:

  • Reset ownership of all files to the controls user, and changed the UID/GID of the advligorts user on cymac1 (it now functions as an alias of controls with equivalent permissions)
  • Shared via NFS the /usr/share/advligorts directory from cymac1. It's now available on the workstations under /opt/advligorts.
  2666   Thu Feb 25 22:34:48 2021 shrutiComputingPSOMACavity design

Two possible good configurations.

Code for creating these are found on our GitHub repo

Explaining the plot:

  • xaxes: Frequency detuning of laser wrt cavity
  • yaxis: Power transmitted through curved optic/ Incident power on cavity
  • Non-black lines labeled in the legend refer to m+n and the same line contains contribution of both the carrier and the first order phase modulation side-bands.
  • Dashed lines correspond to different (m,n) modes coloured so that all modes with same m+n have the same colour
  • Black line is the total power in all modes for the carrier and sidebands
  • Cavity finesse ~312 with input coupler transmission = 0.01
  • The frequencies written near the central peak are the separation from TEM00 of the closest two peaks


Attachment 1: CavityDesignPossibility1.pdf
Attachment 2: CavityDesignPossibility2.pdf
  2665   Thu Feb 25 14:21:35 2021 aaronDailyProgress  

enter Thu Feb 25 14:21:33 2021

picked up more packaged from Downs

  • heavy duty wire shelving for electronics
  • ipad case for imoku
  • ergonomic keyboard heart
  • 5 in-lb and 4-6 ft-lb torque wrenches

Installing OMA model

  • downloaded Matlab 2020b (!!) from Caltech imss onto Spirou
  • When I open it, am alerted that the following directories are missing:
    • /opt/rtcds/userapps/local
    • The *rtscore* directories are properly linked on cymac1, but the link is dead on spirou and gaston. There must be something amiss with the nfs service. Following the instructions on atf wiki that Jon used to set up LDAS backup for cryo frames...
      • The /opt/rtcds and /opt/rtapps directories are hosted on cominaux (they appear in /etc/exports), and cymac1 and the workstations are set up as clients (both directories are mounted in /etc/fstab).
      • On cominaux, running sudo systemctl restart nfs-common gives an error that the nfs-common.service is masked.
      • On the client side (cymac1), /opt/rtapps is modifiable by controls and owned by root as suggested on wiki, but /opt/rtcds is owned by advligorts (set by Chris, so probably this is correct).
      • I mounted all and set up an rsync server on cymac1 using the steps in the wiki, but no change to /opt/rtcds directory. Subsequently deleted the freshly created /etc/rsyncd.conf, and kill rsyncd.pid, to return cymac1 to its original state.


  • The AI chassis is working, there's no short. It draws 0.48 A on the positive supply, 0.46 A on negative supply in steady state. I gave the Sorensens about 0.5 A overhead by ramping down the current limit until the limit was reached, then increasing the limit by about 1A, then turning on the AI chassis.
  2664   Tue Feb 23 13:43:57 2021 aaronDailyProgressPSOMAcondensed fiber layout

enter Tue Feb 23 13:43:48 2021

  • photographed the AI chassis innards. Where is the mini hdmi cable for the camera?
    • When I turned on the AI chassis, the Sorensen hit a current limit on + supply. I didn't notice this at first, so haven't checked it out (could be the limit is set too low, as AI board was fine on the benchtop). For now, do not turn on the AI chassis.
    • Without the AI chassis, the current draw is 2.2 A on positive supply, 1.2 A on negative. I'll try to identify where the extra 1 A is going, and see if I have a photo from earlier this year. I did find this elog from Johannes that shows an extra 1A on positive supply. Some of the cryo cavs equipment is still on, so I am not overly concerned.
  • redid the fiber components (attachment). Didn't touch any free space mirrors, but rotated the launches slightly (axisymmetric about z, no pit/yaw/z changes) when unscrewing fiber.

Next planning to add some foam around the fibers and compare beat spectrum before/after.

exit Tue Feb 23 22:16:33 2021

Attachment 1: 2AAF19D0-3F40-4146-9975-8EF4545C91A7.jpeg
  2663   Fri Feb 19 22:01:44 2021 aaronNotesLaserPower vs. Temperature plots for Rio PLANEX

Not that it was well motivated, but we've been using the "West Bridge Labs" (fomerly ATF?) wiki for PSOMA. Planex datasheet on the documents page.


I added the RIO Planex datasheet to the Cryo Wiki.

  2662   Fri Feb 19 14:46:04 2021 ranaNotesLaserPower vs. Temperature plots for Rio PLANEX

I added the RIO Planex datasheet to the Cryo Wiki.

I believe the equation to relate resistance to temperature is this one:

and for us, B = -3950 K

This Wolfram Alpha link gives what we really want: T(R). To be safe, I recommend we never go below 6 kOhm, or above 12 kOhm.

  2661   Fri Feb 19 14:43:09 2021 aaronComputingDAQAI chassis diagnosis

Opened up the AI chassis

  • Interface card with pin mapping is D070101-v1. AI board is D070081-v2.
  • Rana and I identified that op amp U3A needed replacement on channel 1 only; all other channels were functioning normally
  • I removed the front panel and flipped over the offending circuit board ("A). I powered on the board to identify the correct op amp, but something was shorting the negative supply. I disconnected the flipped board from power, and the supply was still shorted -- this was happening somewhere on the board for channels 9-16 ("B"), which I hadn't touched. I disconnected power from board B, wiggled the connectors, and reconnected it -- power was restored. 
  • I again flipped over board A, and this time was able to identify the offending op amp by testing its input and output pins, and nearby resistors. I replaced the op amp by 
    1. apply tack flux SMD291 to all pins of IC
    2. melt chip quik alloy on pins, and with the alloy molten lift off the IC
    3. Clean off the brittle chip quik with a flux coated solder wick. Wipe away the remaining flux with a kim wipe and isopropyl
    4. Solder on the new surface mount op amp. It is an AD8672ARZ, which we had in the electronics shop.
  • Realized I had replaced the wrong op amp (U2, not U3). I hadn't realized the AD8672 is a dual op amp, so misunderstood the pin mapping from before. Fortunately, the swap was good and the changed op amp is still working.
  • After the swap, the negative supply short returned. Again, it was on board B, not the board I had been working on. After wiggling cables again, taking apart the connector and confirming no stray wires and good connections, the short went away and power was restored. Pretty confused by this. I might have just had the current limit too low? Could not reproduce the problem, and didn't do anything that should have fixed a short. 
  • After replacing the correct op amp, channel 1 no longer has an offset. yes

I meant to take a photo while still opened up, so I left the DAC to AI unplugged to remind myself to get to that next time; should also post some transfer functions to confirm everything is functioning. 

Update: photo is now on the west bridge wiki and ligo.wbridge google drive

  2660   Fri Feb 19 14:34:44 2021 aaronNoise HuntingPSOMAITC 502 current noise

Moku was set to 50 Ohm input impedance, SR560 set to 50 Ohm output. 

The resistor an diode are in series and I measured the voltage across the resistor only, as in Johannes' figure. I just wasn't sure what the current drive wanted to see, I also measured with just the resistor (no diode) and got something similar. 


did you have the Moku set to 50 Ohm impedance during the current noise measurement? Was the SR560 using its 50 Ohm output?

what's the idea with using a diode + resistor, rather than just a resistor? I'm not sure what the actual resistance - are they in series?


  2659   Fri Feb 19 14:26:08 2021 ranaNoise HuntingPSOMAITC 502 current noise

did you have the Moku set to 50 Ohm impedance during the current noise measurement? Was the SR560 using its 50 Ohm output?

what's the idea with using a diode + resistor, rather than just a resistor? I'm not sure what the actual resistance - are they in series?

  2658   Fri Feb 19 14:14:29 2021 shrutiNotesLaserPower vs. Temperature plots for Rio PLANEX
I turned on the temperature control and TEC for the North laser with the TED 200C laser controller to measure the power dependence on temperature.
The power was measured right after the fiber launch. Laser -> Faraday Isolator -> Intensity modulator -> 90% PO -> Fiber launch

The data and code for plotting is in Attachment 1.

The plot is in Attachment 2.
While the sensor resistance decreased (corresponding to an increase in the laser temperature) an abrupt decrease in power was seen at 8.72 \pm 0.01 kOhms [close to 28 C];
When decreasing the temperature (measured by an increasing sensor resistance) an abrupt increase in measured power was seen at 10.23 \pm 0.01 kOhms [close to 24 C].

The two sensor resistance values for abrupt power change was determined visually from the graph.


Attachment 1: PvsT_20210219.zip
Attachment 2: PvsT.pdf
  2657   Fri Feb 19 13:24:10 2021 aaronDailyProgressLab Work 

Enter Fri Feb 19 13:24:02 2021

  • The power strip for the lab workstations was unplugged. Was there a reason for this? I plugged it back in. Nothing wrong necessarily, but since it seems unrelated to anything on the elog and the hall lights were also off want to be sure there wasn't an unauthorized/undocumented lab entry.  
  • I found a slotted lid for rack electronics next to the cymac / cryo cavs rack, and used it to close the OCXO preamp box on PSOMA rack. I think it belonged to the ISS box on cryo cavs rack, which is currently not in use.
  • The Valon 3010 frequency divider had a loose screw rattling around inside. I opened the box (grounded, no power), secured the loose screws, and put the divider in the RF drawer.
  • Moved over the l-He dewar a few inches to confirm we have enough room for a 48" shelf (gap is now 49"). Also lowered its rubber feet, so it is resting on the fixed feet rather than the wheels. 
  • The label makers in EE and cryo are in the Casio family (for reference) 


A lot of the electronics around the lab use green for the negative supply line, and black for ground. I figured I just remembered the convention wrong, but from what I can find online the standard is green for ground and black/red for line in both US and EU... I don't want to change this local convention, but is there a reason we use it? Is this just cryo, or other WB labs / 40m? Since our colors are nonstandard, it's really necessary to label all power wires.

Mostly was fixing the AI chassis.

exit Fri Feb 19 22:01:12 2021

  2656   Fri Feb 19 09:39:35 2021 ranaNoise HuntingPSOMAITC 502 current noise

How about rack mounting the moku somehow? We can also then connect it to an ethernet cable so that it can have its freedom to talk to its Moku siblings all over the world.

Recomended iPad cases


In the theme for the week, I caught my leg on a BNC cable and dropped the Moku from benchtop height indecision. Tried to convince myself nothing broke by measuring white noise from an SRS function generator, and comparing the LD driver current noise spectrum to that measured by HP 8560E. There's a spurious peak (floating inputs) on ch 1 at 356 MHz (attachment 1), though this is not the first time I've noticed something similar. Still, it's troubling; when I send the current noise from SR560 to either channel of Moku, the peak remains in the signal-carrying channel, even though the signal is well above the noise floor.


  2655   Fri Feb 19 00:52:40 2021 ranaElectronicsElectronicsinspecting W diode driver

I have the old "W" driver which was labeled as noisy in the EE shop. This is the one that Aaron saw a ~10 MHz oscillation on. E-traveler # S1500270.

With the Moku, I didn't see any oscillation. There was some noise at a few of the monitor points, but not sure yet if that was the Moku noise or the circuit.

There has been some modification of this "V3" circuit. Although the knobs make the correct voltage change in the early stages of the circuit, the resulting voltage does not make it through the summing and filtering stages. Need to look up Johannes's elog and see what was done to this and why.

  • Zach elog where he changes the knob range and measures the resulting Amperes/Revolution


also, here's a link to Johannes's measurement of current drive noise with a transformer. Apparently he used a resistor (no diode), and our sub-nV transformer. SR560 by itself might be fine too.

On the Moku, and also on RF spectrum analyzers, "dBm" stands for "dB mW", not "dB mV". Its a measure of RF power into a 50 Ohm load. i.e. mW = 1000* (Vrms^2)/50. And 0 dBm  ~ 222 mV rms, -40 dBm ~ 2 mVrms.

  2654   Thu Feb 18 16:16:38 2021 aaronComputingDAQcymac to Debian buster

The offset could be a problem with the AI chassis. Attachment 1 is a ~3 Hz sine measured directly from the DAC, and attachment 2 is after the AI chassis. There might be a minor offset from the DAC (at most 6 mV), but there is a ~1 V offset after the anti-imaging.

I removed the chassis, haven't done anything inside.

Attachment 1: 15A01777-A1C5-4710-8180-28C8B68A07AE.jpeg
Attachment 2: F90CD5EF-1A26-4C4A-9C84-046E49642545.jpeg
  2653   Thu Feb 18 14:06:56 2021 aaronNoise HuntingPSOMAITC 502 current noise

enter Thu Feb 18 14:06:53 2021

  • finished routing AC power cables for PSOMA table


Commercial current driver noise

  • LD Datasheet 
  • ITC 502 manual
  • Measurement configuration
    • ITC 502 LD out connected to DB9 breakout board
    • Pins 1 and 5 connected with a 50 Ohm dummy load, to lift the driver interlock
    • Pins 3 and 7 connected across a 2V diode and 20 Ohm resistor. 
    • SR560 channel A measuring voltage across the resistor, AC coupled and high-passed (6dB rolloff at 1 Hz) with G=100. The preamp is running on battery power in 'low noise' mode. 

While searching for a functioning SR560, I found 2 with bad batteries (can run on AC but not on batt power), one that overloads with no gain or load attached, and one with a 32 MHz oscillation when running on battery (outside the 1MHz amplifier bandwidth, but still odd). All are now labelled as such. 

In the theme for the week, I caught my leg on a BNC cable and dropped the Moku from benchtop height indecision. Tried to convince myself nothing broke by measuring white noise from an SRS function generator, and comparing the LD driver current noise spectrum to that measured by HP 8560E. There's a spurious peak (floating inputs) on ch 1 at 356 MHz (attachment 1), though this is not the first time I've noticed something similar. Still, it's troubling; when I send the current noise from SR560 to either channel of Moku, the peak remains in the signal-carrying channel, even though the signal is well above the noise floor.

Not to mention, the current noise spectrum looks wack. The signal was barely in range of the 8560, but I ended up with a qualitatively similar spectrum after boosting SR560 gain to 1000 (attachment 4). The spectrum was the same with both functioning SR560 I tried. 

Note y axis is in dBm, so -100 to -80 dBm/rtHz is 1-10 nA/rtHz out of the driver. The ITC 502 manual quotes < 5 uA rms from 10 Hz to 10 MHz, so 1nA/rtHz is in the right ballpark. 

P_{\mathrm{SA}}=\frac{V_\mathrm{out560}^2}{50\Omega}=\frac{(100*20\Omega)^2}{50\Omega}I_{\mathrm{noise}}^2 \implies \mathrm{dBA}\approx \mathrm{dBm}/2 - 40

\sqrt{\int_{10\mathrm{Hz}}^{10^7\mathrm{Hz}}(10^{-9}\mathrm{A}/\sqrt{\mathrm{Hz}})^2df}\approx 3\mu \mathrm{A}

exit Thu Feb 18 21:00:47 2021

Attachment 1: IMG_0017.PNG
Attachment 2: IMG_0018.PNG
Attachment 3: IMG_0019.PNG
Attachment 4: 2DBAD4BD-C058-42FB-903F-98128E99F789.jpeg
  2652   Thu Feb 18 13:05:50 2021 aaronUpdatePSOMA 

Here are the frequency noise ASD from the beat note and individual lasers from the three corner hat.

Update: the plotting notebook is now in the Cryo scripts repo.

Attachment 1: beat_ASD.pdf
Attachment 2: lasers_ASD.pdf
  2651   Thu Feb 18 12:30:09 2021 aaronNoise HuntingLaser"North" current driver having slow current fluctuations

I dug through the log to figure out what created the 8s noise in the current driver. I suspect it was my power cable connectors on Feb 3 , but it seems Rana measured this clean beatnote spectrum on Feb 4. Still, all other changes were downstream of the current drives -- was the broken current driver involved in that clean beat? 

I've added another datapoint of the laser frequency noise from my measurement Jan 25.

  2650   Wed Feb 17 13:25:52 2021 aaronDailyProgressPSOMAelectronics management

enter Wed Feb 17 13:25:50 2021

I'm clearing electronics from under the PSOMA table, and temporarily storing them on the lH2 dewar's base. It's quite dusty under there, and I wiped down what was below. I've added the wire shelf we want to order to the PSOMA hardware inventory ('other'), along with 80-20 components to make a tabletop "soundproof" enclosure.

While under the table, I noticed the N2 line (for floating the table) is touching a strut attached to the floated table, creating a mechanical short. I've added zipties to avoid this. Also mounted the power strips to the table legs' struts, and labelled power cables as I went.

The MA-40 HEPA filter unit indicates it's time for a filter replacement. I've contacted Calum, who might have a supply of spare filters for these units. Perhaps it's also time to check the FFU on cryo cavs table.

Began but did not finish rerouting cable on the tabletop.

exit Wed Feb 17 20:18:39 2021

  2649   Wed Feb 17 11:56:20 2021 shrutiComputingPSOMACavity design

The code for this can be found here. I have also updated the PSOMA Wiki with similar information.


Attachment 1

This is an updated diagram including the sidebands at around 33 MHz since that is what the available oscillator at the Cryo lab provides. The exact modulation frequency (within 1 MHz) does not seem to matter much since it only changes slightly the relative position, with the carrier, of the peaks created by the sidebands.

The numbers in the legend refer to m+n

L_{rt} is the round trip length, R_c =2m the radius of curvature of the only curved optic in the triangle cavity with the specifications of the one currently on hand, \theta is the vertex angle at the curved optic (twice the angle of incidence).

The angle of incidence at the input coupler is 45^\circ because of its coating specifications.


Without any optimization the cavity scan can look like that in Attachment 2 (the sidebands are shown separately for a little clarity).

If anyone is interested in running FINESSE and other packages on Google Colab: this may help. [Access with Caltech credentials]

Attachment 1: Optimized_incl_sidebands_old_2m_33_5MHz.pdf
Attachment 2: Unoptimized.pdf
  2648   Tue Feb 16 14:29:04 2021 aaronComputingDAQcymac to Debian buster
rtcds start --all

works to start the models, though FIFO status light is still red.  There is also still an offset on DAC outputs, but not the same as the previously noted offset (-740 mV, down from -2.2 V).

The RCG runtime diagnostics guide (p18) suggests this bit indicates either FIFO is empty (DAC module clocking more often than it should, or IOP code running too long), or FIFO contains too many samples (DC not clocking properly or at all).  It also suggests this error should manifest as an "errant, noisy, DAC output or... complete loss of DAC output signal," which doesn't sound like the fixed offset I'm seeing. Nonetheless, the IOP shouldn't be running long (we've run these models together before), so I check the timing cables to DAC and ADC.

The ADC (but not DAC) haphazardly connected (attachment 1 is the ADC cable). After reseating the DAC cable and rebooting cymac1, the FIFO error is resolved but I no longer get any signal when driving the DAC. I don't see any change in status if I unplug the DAC, so perhaps there just isn't an indicator for 'DAC is unplugged.'

Reseating the ADC cables caused the FE to crash (predictably), but restarting the FE with rtcds restart --all (or stop followed by start) left timing errors. I tried to rmmod the x1iop kernel after stopping the models, but rmmod was not a recognized command. I installed kmod (the package containing rmmod), but the command was still not recognized; when I uninstalled kmod, several rts packages were also uninstalled (odd, since kmod couldn't have been a dependency before it was installed). Fortunately, simply reinstalling advligorts-cymac recovered the state with FIFO status light red. Not sure why the FIFO error returned, but I re-reseated the cables through this so maybe it is just flaky connectors. 

In the end, I left the system as it was before, but mounted the ADC+timing breakout board inside cymac1 next to the DAC+timing board. (attachment 2).

I also checked that the timing signals looked OK (attachment 3 shows the DAC clock on trace 1, ADC clock on trace 2; attachment 4 is zoomed in near 0V, where you can see a slight difference in decay to 0 on the two clocks).

Attachment 1: 55B5D301-4243-45DA-9B03-45EF87B2A684.jpeg
Attachment 2: 4C7FE7DC-E8A0-4770-9A49-14EB491F76E2.jpeg
Attachment 3: 13D67AB6-5A77-4C94-A2A4-FE7FDF065B01.jpeg
Attachment 4: C84B3054-269B-4969-817F-1429BB671C7A.jpeg
  2647   Tue Feb 16 14:27:40 2021 aaronDailyProgressPurchasesupcoming wire shelf

enter Tue Feb 16 14:20:40 2021

We'd like to put a heavy duty wire rack on the W wall of cryo lab, to store the electronics currently under PSOMA table. The space available is 43'' x 32'' x 102 '' (width x depth x height, have ~9'' of spare width if we shift existing cabinets).

exit Tue Feb 16 17:12:17 2021

ELOG V3.1.3-