40m QIL Cryo_Lab CTN SUS_Lab TCS_Lab OMC_Lab CRIME_Lab FEA ENG_Labs OptContFac Mariner WBEEShop
  Cryo Lab eLog, Page 21 of 59  Not logged in ELOG logo
ID Date Author Type Category Subjectup
  742   Fri Apr 26 19:38:39 2013 DmassLaserLaserLaser Tuning Coefficients

Got gamma~0.15 back after playing with the sideband setup for a bit (one bad cable, 1-4 cracked capacitors). Was better / more solid after.

 Rebuilt beat readout, realigned onto beat PD, found beat frequency:

f_beat = 117.8 MHz @ -14dBm

Rth_102068 = 9.237k

Rth_102085 = 8.018k


The viewer + card worked OK to see the dimmer spots. It was easy to see the ~2mW beam right after the window (~9 inches from cavity waist). It was very hard to see a single beam after the beamsplitter (18 inches after the cavity waist, and 1/4 the power because of beamsplitters) sans lenses. With an IR card and the IR viewer from the ATF I was able to see the dim spots without issue

  1343   Fri Dec 4 16:51:49 2015 BrianLaserLaserLaser actuation range

In our design of a new cavity, the size of the laser actuation range places a limit on the cavity FSR and thus the cavity length. I scanned through to see how far we can actuate with the temperature control.

Each point on the below graphs is a resonance of the TEM00 mode. Each point is seperated from its neighbors by one FSR which is 1.5 GHz. The red vertical line is where the laser mode hops. The power was read out from the RAM/RIN monitor and normalized to the maximum value.

If we need at least 10% of the max power, both lasers have an actuation range of about 15 GHz. If we need at least 50% of the max power, we only have an actuation range of about 9 GHz.

It is possible that the noise performance of the laser changes across this actuation range, which we should be able to measure from the beat frequency.

Attachment 1: westPower.png
Attachment 2: eastPower.png
  1138   Thu Aug 7 14:01:38 2014 DmassDailyProgressLab WorkLaser current sidebands

In order to free up the EOMs for a phase corrector path (and not have to figure out how to make a DC path on the resonant step up circuit without ruining the Q of said circuit) I decided to put sidebands on with laser current (again) and actually measure the RAM this time.

The input to the laser is as described in elog:676

There were three reasons we previously abandoned this scheme:

  1. The AM/PM ratio (at DC) was very high and made us nervous (AM/PM ~ 0.5: see elog:827)
    • Thinking slightly harder, it is clear that what we really care about is the changing RAM. Since the unknown source of RAM in the laser is unrelated to the known source of RAM in the EOM (drifting temperature -> drifting axis -> polarization misalignment).
    • It is completely possible to have a large DC RAM level and low RAM noise using the laser current modulation
    • Thus: we should measure RAM
  2. When we touched/stressed the cable connected to the current modulation input, we saw huge (~10MHz) low-frequency frequency shifts (seen while sweeping the cavity)
    • Some strain relief as done for the current modulation cabling (elog:1111)  can mitigate this problem to some degree
  3. In order to use PMCs in our scheme, we would have to put the PDH sidebands on after the PMC
    • This will still be a problem if/when PMCs are put into play. For the time being this can be ignored/tabled

After making a strain relief / mechanical isolation feedthrough for the current modulation path and putting the SMA to butterfly PCB adapter board back on the laser, I put sidebands back on the laser using current modulation

I measured modulation depth by sweeping the cavity and looking at transmission sidebands and was able to achieve gamma = 0.68 @ 33.6 MHz relatively easily (sending +7 dBm at the laser current modulation port)

I demodulated and used a lock in amplifier to monitor the I and Q f the RAM, and recorded it in the cymac - see attached drawing of the setup

I recorded the signals (as well as the measurement noise floor levels) in the Cymac with the following channel names:

  • AUX1: DC Power Level Monitor
  • AUX2: Theta (I/Q angle drift)
  • AUX3: R (total RAM amplitude)

Notable from coherence plot:

AUX3/AUX1 is not appreciably coherent anywhere, so we are not seeing a total power level based fluctuation here

AUX3/AUX2 IS coherent at low frequency (up to ~2 Hz), so whatever is driving the RAM also drives the phase angle of the signal. This is not shocking, just noteworthy.

Calibration needed:

  • AUX1: Counts/(Vdc at output of PD in 50 ohms)
    • [4182 cts] / [272 mV dc into 50 ohms]
  • AUX2: Counts/Deg
    • [32080 cts] / [360 deg]
  • AUX3: Counts/(RAM level measured at output of RAM mon PD into 50 Ohms)
    • [4980 cts] / [3 mVpp at output of PD into 50 ohms]

Applying this calibration we get:

  • Gamma_AM = 1.5/272 = 0.0055 (this is smaller than the old numbers I quoted, but also should be a more accurate representation of what is actually going into the cavity than previous measurements based on where we are measuring).
  • To double check / confirm the calibration: make sure driving 0 dB at the current mod input produces gamma_AM ~ 0.005 (or 3 mVpp RAM on a 272 mVdc signal)

Non DC Gamma:

[ASD in RAM] = [ASD in counts] * [1.5mVpk / 4980cts] * [1Vrms/1.41Vpk] * [1/272mVdc] 

[RAM/Count] = 7.9e-7 RAM/Count

[Freq Noise ASD] = [RAM ASD] * [P_incident] * [Hz/Watt]

P_incident = 1 mW

Plant = 2e-8 W/Hz

Freq noise = RAM * 1e-3 / (2e-8)

At 1 Hz we can see we have: 10 cts/rtHz * 8e-7 RAM/cts * 1e-3W/ (2e-8 W/Hz) = 0.4 Hz/rtHz

At 10 Hz we have 4e-3 Hz/rtHz

Compare to our noisebudget in elog:1099, we can see the RAM noise is above the coating thermal noise but well below the other experimental noises at 1Hz, and well below the coating thermal noise by 10 Hz.

The DC offset is 3.3mV at the error point from the RAM

The next  question: is the phase of the RAM changing such that, once demodulated, we are limited by the RAM (since what we care about is I, not sqrt(I^2 + Q^2))

The maximum amount of RAM noise from the RAM phase noise is about the same (within a factor of two) as the pure RAM noise (it seems not so surprising that a transformation of variables from I and Q to R and theta and back to I and Q would yield noise which is equal in amplitude for both variables) 


Attachment 1: 0807141425.jpg
Attachment 2: fmodramplot.png
  1140   Tue Aug 19 11:10:48 2014 DmassDailyProgressLab WorkLaser current sidebands

Can't use current modulation based sidebands in "as-is" setup - there were HUGE (5%) line harmonic (60, 120, 180) dips in transmission which got worse as we started gain peaking.

When I switched to using the Pockels cell for PM sidebands, here were no visible dips in transmission visible on the scope when triggered on the AC line.

This means:

If we want an actuator for the phase corrector path, we either need to hunt for and find/mitigate the source of this ground loop, buy another Pockels cell, or add a DC summing path into the resonant sideband circuit


  2457   Thu Jul 11 19:25:41 2019 ShubaElectronics Laser diode working

After the cryo weekly meeting, I proceeded to the Cryo lab. Aaron showed some Q measurements in cold condition. After that, I took the current driver circuit to the EE shop to test out the discussed procedure in the exact sequence. I was using a Thor Labs Laser L405P20 - 405 nm, 20 mW, datasheet attached below. The final outcome is that the violet Laser is working fine.

Steps I followed :

  1. Connected a 110 ohm resistance in series with the laser diode, powered by a DC voltage to find a estimate of internal resistance of the diode(attachment 1). Probbed the voltages across the resistors to find current through the circuit. When the supply voltage was 2.7V, the diode just started shining. I increased the supply voltage very slowly till 5V, because the laser can handle that much of voltage and I kept a constanf check on the current (always less than 40mA). When the supply DC voltage was 4.47V, the voltage across 110 ohm resistance was 0.87V, which gives an estimate of around 500 ohm internal resistance of the diode.
  2. In the next step, I connected +_ DC supply to the current driver circuit and provided an input signal of 0V from another voltage source. In place of the diode, I connected a 510 ohm resistance, and probed the voltage across it. The switch on the current driver circuit was in 'off' condition. At this instance, all the voltage supplies read 0V. I switched on the driver and slowly ramped up the biasing voltage to 14V keeping a constant check on the current drawn. Then I increased the voltage slowly on the input side, the rise was as expected. i went to a maximum of 3V input. I then reduced the input voltage to 0V. Ramped down the biasing voltage to 0V and also, switched the circuit off.
  3. I followed similar procedure as 2nd one, but with the laser diode connected and constantly probed the voltage across it, shown by the multimeter reading in the attached video (also the laser can be seen shining). I observe that, even for a 0.1V of input, the output is around 3.1V. I slowly increased the voltage to 1.1V input, which gave an ouput voltage around 4.4V to the diode. The intensity of the laser also increased. I decreased the input voltage to 0V and turned the driver switch to off position. Slowly ramped down the biasing voltage to 0V. Then I kept the things back to Cryo Lab.


Attachment 1: IMG_20190711_172056.jpg
Attachment 2: VID_20190711_180553.mp4
Attachment 3: L405P20-SpecSheet.pdf
L405P20-SpecSheet.pdf L405P20-SpecSheet.pdf
  2448   Tue Jul 9 17:40:02 2019 aaronDailyProgressLab WorkLaser driver circuit, cymac

[aaron, shubha]

I pulled the second laser driver circuit box from the cryo cavs table, and we took it to the EE shop. We tested it in stages:

  • Power supply is giving the voltage we want (+-18V for the L7812 in the circuit)
  • this voltage makes it to the power switch on the box
  • The polarity on the connector on the board is correct (78** gets positive, 79** gets negative)
  • When we power on the board, the op amp power legs get the correct voltages
  • We attached an LED where the laser diode would go, and when we give it the nominal voltage at the control point (~3V) it illuminates.

Shubha will post about what's still undiagnosed (including the real circuit diagram) and what is to come.

[aaron, anna]

Anna wasn't able to get modes exciting on the disk, so I'm taking a look. Shubha and I had disconnected the cable carrying the ESD excitation signal to the HV driver, so we could temporarily use that channel for the laser driver circuit. When we went to replace the cable, I noticed that the janky DB9-to-BNC I'd made had finally broken, so we replaced it with a PCB DB-BNC breakout.

The ESD excitation still wasn't reaching the HV drive, and though the DAC TP channels were reading the correct values, we weren't seeing any voltages out of the DAC chassis. I ran killAll/startAll, but am getting the characteristic increasing checksum errors (code 0x1000 on the DC status part of the GDS_TP screens). There were no bad frames, and in the past we've narrowed this down to a timing issue. At some point while digging through old elogs to find what we did last time, cymac1 crashed (all white boxes, no ping).

We pressed (not held down) the power button once, waited 5 minutes, then did it once more. With no response, we held down the power button for a hard reboot.

When the system came back online, I ran startAll and burtwb the 'measureReady' snapshot. All (usual) indicators green (attachment 1), but we still aren't getting any signal out of the DAC.


Attachment 1: 04_PM.png
  2447   Tue Jul 9 14:43:43 2019 ShubhaDailyProgressLab WorkLaser driver malfunctioning

There are two current drivers down in the lab. In the drive which uses a P channel MOSFET , It seems the voltage regulator on the current drive is blown up. Aaron took the second current driver out from the Johannes table. I provided similar input as yesterday, but still the laser isn't working. I opened the box and checked the circuit with multimeter. At the input of the laser, I see a voltage of 8.8V which is very high. The operating voltage of the laser diode is 2.3V with a maximum capacity of 2.5V.

  1597   Thu Jun 15 13:58:17 2017 johannesNotesLaserLaser frequency response

I gave measuring the laser frequency response another shot, this time in more open loop fashion and using the delay line to obtain the frequency tuning. What I did:

I used a Marconi FG to find a suitable zero-crossing of the delay line. I picked 433.03 MHz, tuned the laser beat note to that frequency, and determined that it had an amplitude of 0dbm at the input to the delay line (coming for the ET-3000A). Setting the Marconi output to that same amplitude, I determined a response coefficient of 4.4 mV/MHz for the delay line output.

I then set the Marconi to a frequency of 428.03 MHz and used the ~5 MHz output of a mixer that multiplies the two signals as the input to the Zurich HF2LI, whose internal PLL spits out the frequency difference between its internal oscillator and the input signal. I used a SR560 with .3Hz lowpass on the difference frequency signal and connected that to the frequency modulation input of the east laser. The idea was that instead of a direct phase lock I would have a not so aggressive frequency lock between the lasers that would keep the beat in the linear regime of the delay line.

I then measured the transfer function from west frequency modulation input on the current driver to delay line output and used the earlier obtained conversion factor to give it meaningful units. The result is the following (delay line response still included):

The magnitude response at low frequencies is a little less that what I would have expected based on cavity sweep findings from before. There seems to be an f^-1/2 slope and quite a bit of phase even at 10 kHz. I'm still working on disentangling the amplitude response and the delay line response from this, and also have to actually characterize the loop shape of the frequency lock.

Attachment 1: west_response_20170614.pdf
  885   Tue Oct 1 17:51:26 2013 nicolasLaserControl SystemLaser response according to RIO paper

Fig. 3

I don't understand how we can even get a 100kHz loop with this transfer function of the laser. The amplitude starts rolling off already at 1kHz and there is 90deg of phase as 10kHz.

In any case, it seems this laser is slow. So we should try to compensate it in the PDH control box, or perhaps with a phase correcting EOM.

  886   Tue Oct 1 17:54:01 2013 DmassLaserControl SystemLaser response according to RIO paper


Fig. 3

I don't understand how we can even get a 100kHz loop with this transfer function of the laser. The amplitude starts rolling off already at 1kHz and there is 90deg of phase as 10kHz.

In any case, it seems this laser is slow. So we should try to compensate it in the PDH control box, or perhaps with a phase correcting EOM.

This might be the transfer function through the RIO supplied ORION driver (which is a big part of what motivated us to make our own) - Disclaimer: I haven't checked the paper yet, so this is just a suspicion

  1631   Fri Jul 21 16:17:14 2017 VineethNotesDigital PLLLaser transfer function measurement

The updated diagram for the transfer function measurement is attached below.

Attachment 1: drawing.svg
  1633   Sun Jul 23 16:10:48 2017 ranaNotesDigital PLLLaser transfer function measurement

It seems strange to have a 50 Ohm in series with the laser drive. What is the impedance of the laser diode?

Also, for more production quality use, we would want to know about the rate of glitches. Is it possible to set up the device and record the phase signal in the DAQ for 24 hours to see how much it is bursting?

  1636   Mon Jul 24 11:08:44 2017 johannesNotesDigital PLLLaser transfer function measurement

The 50 Ohms are to ground, implemented in an off-the-shelf inline BNC package. It's just a mistake in the drawing. That input also doesn't go directly to the laser diode, but to the current driver's fast modulation port (differential amplifier)

  1623   Thu Jul 13 13:36:21 2017 VineethNotesDigital PLLLaser transfer function measurement

The phasemeter developed in the Red-Pitaya was used to measure the transfer function of the diode lasers.

The diode current of one of the free running lasers (not locked to the cavities) was modulated with a source of -13dBm. The beatnote obtained was around 20 MHz with a Vpp = 1.5V. The modulation changes the frequency of the beatnote, which in turn translates into the voltage change in the output of the DAC. The transfer function measured would then be the product of Laser TF, ADC->DAC TF, and the closed loop TF of the phasemeter. So, to obtain the Laser TF we divide the other factors out (which were measured with the same parameters).

The gains used for the measurement are Kp = 0.03 and Ki = 0.01


Attachment 1: PTF.pdf
Attachment 2: LTF.pdf
Attachment 3: CTF.pdf
Attachment 4: drawing.svg
  1624   Thu Jul 13 21:52:58 2017 ranaNotesDigital PLLLaser transfer function measurement

what are the input/output impedances of the pitaya?

  1625   Fri Jul 14 13:16:35 2017 vineethNotesDigital PLLLaser transfer function measurement

The input impedance of the ADC channels is 1MOhm and the output impedance of the DAC channels is 50 Ohm.


what are the input/output impedances of the pitaya?


  1627   Tue Jul 18 20:04:27 2017 ranaNotesDigital PLLLaser transfer function measurement

probably you ought to update the diagram to show how you implement the 50 Ohm termination for RF transfer functions

  1182   Wed Dec 17 13:54:19 2014 ZachLaserSiFiLasers mounted, energized, beat set up

On Monday, after I did some inventory of all the parts we have received from various companies, Dmass helped me mount the RIO lasers into their mounts so that I could get started with the optical setup. We cleaned the surfaces with methanol, applied a small layer of silver thermal compound, and then screwed them in.

I then borrowed the following to run the lasers:

  • The (separate) ThorLabs diode driver and temperature controller from Haixing's maglev setup
  • An integrated ThorLabs diode driver / temperature controller from the TCS lab

After finding the right cables, I powered up the lasers and verified the P-I curve for each as listed on the spec sheets.

I then built a quick (temporary) optical beat setup, combining the two beams on an 1811. I had the temperatures (actually, thermistor resistances) set to what was listed as the testing set point on the datasheet, and as soon as I overlapped the beams and focused them onto the PD, there was already a strong ~50 MHz optical beat.

diagram.jpg setup_with_beat.jpg

I have spent some time since then trying to lock various kinds of PLLs, both to interrogate the free-running frequency noise and to get used to controlling the lasers. Some things I've tried:

  • Locking a Marconi to the free-running beat, which I think might be an exercise in futility due to the relatively small range of the Marconi FM
  • Locking one laser to the other directly using a PLL, which I think might be an exercise in futility due to the bandwidth of the current actuation from the ThorLabs driver
  • With Dmass's help, locking a Zurich PLL to the free-running beat. This appeared to work, and we saw a preliminary frequency noise spectrum that looked about right, but I'm skeptical because the control signal doesn't seem to respond to my slewing one laser's frequency.
  • Briefly, locking one laser to the other at low frequencies using the Zurich PLL control signal as a frequency discriminator. This didn't work, adding to my suspicion.

The first two were not helped by the fairly basic loop shaping afforded by attenuators and an SR560.

I think my next step will be to simply use the I-Q demodulation method (like I did to measure the no-FM Marconi noise in ATF:1877) to measure the frequency noise. I'll compare that to what I get with the Zurich PLL.

  1183   Wed Dec 17 14:40:15 2014 DmassLaserSiFiLasers mounted, energized, beat set up

  • With Dmass's help, locking a Zurich PLL to the free-running beat. This appeared to work, and we saw a preliminary frequency noise spectrum that looked about right, but I'm skeptical because the control signal doesn't seem to respond to my slewing one laser's frequency
  • Briefly, locking one laser to the other at low frequencies using the Zurich PLL control signal as a frequency discriminator. This didn't work, adding to my suspicion.

If the "locked indicator" light is not green on the Zurich (first tab, under "Reference", then what you get out is junk (e.g. you have unlocked the lock in, and i hasn't re-acquired yet) - you can do this by kicking it too hard with a frequency shift, which would be easy to do if you were slewing laser frequency, as the coefficients of the laser [Hz/mA] is so big. When the lock in loses the signal, you have to manually re-lock it (toggle off and on the button which has the mouseover text: "enable the fixed center frequency mode of the PLL"). You can get  something which sort of looks like a PLL signal which has terrible noise and weird glitchy response when the lock in isn't locked in.

Your instinct to look for slewing at the PLL control point is correct, and a sign that the state of the PLL is healthy/unhealthy

  1184   Wed Dec 17 18:11:38 2014 ZachLaserSiFiLasers mounted, energized, beat set up



If the "locked indicator" light is not green on the Zurich (first tab, under "Reference", then what you get out is junk (e.g. you have unlocked the lock in, and i hasn't re-acquired yet) - you can do this by kicking it too hard with a frequency shift, which would be easy to do if you were slewing laser frequency, as the coefficients of the laser [Hz/mA] is so big. When the lock in loses the signal, you have to manually re-lock it (toggle off and on the button which has the mouseover text: "enable the fixed center frequency mode of the PLL"). You can get  something which sort of looks like a PLL signal which has terrible noise and weird glitchy response when the lock in isn't locked in.

Your instinct to look for slewing at the PLL control point is correct, and a sign that the state of the PLL is healthy/unhealthy


 Yes, I noticed this effect. I'm talking about immediately after acquiring---or re-aquiring---PLL lock. I did this several times at different beat frequencies to see what effect it had on the noise (the spectrum changed considerably, which is another bad sign).

  904   Mon Oct 14 04:24:25 2013 DmassDailyProgressLab WorkLast week's progress
  • Lock West cavity with LB1005 (again)
  • Switch Gold PDs
  • Put sidebands on E laser with EOM
    • Measure gamma ~0.09 w/ +21.5 dBm in
  • (Attempt to) minimize RAM via EOM alignment
    • Was able to make RAM go down and convinced myself that I wasn't clipping beam
    • Level:
    • Measure gamma; make sure it's still big - yes
  • Swapped splitters in PDH path (resistive splitters -> reactive)
  • Recover beat signal (+13.5 dBm @ 240 MHz, so no significant loss of beat amplitude)
  • Recover beat readout
  • Measure beat spectra at different DC levels for the beat readout (elog:903) <-- CONFUSING, AND POSSIBLY INTERESTING
  760   Tue May 14 00:36:27 2013 DmassDailyProgressLab WorkLeak testing update

 I borrowed a He leak detector from a benevolent K Schawb group.

The whole unit is on a single rolling cart, and it has a nice UI / control setup that keeps you from doing anything (too) dumb to the turbo / leak checker.

Initial leak testing:

  1. I left the 300K shield clamped down to the table so that I could retain alignment of the windows w.r.t. the table
  2. I undid the outer vacuum seal and lifted the cryostat up (minus the 300K shield)
  3. Put some bases on the giant (~2" x 10") posts to use these as support legs
  4. Put legs underneath the inner vacuum shell
  5. Rested cryostat on legs
  6. Got a hypodermic needle, filed the tip dull, and attached this to a flow-rate-limited 4He gas tank for leak checking purposes via plastic tubing
  7. Skyped Warren to have him help walk me through the proper procedure for leak checking a vacuum chamber:
    1. Seal leak checker off and test it's calibrated leak source (supposed to read 10^-6 mbar x liter / sec, read this to the two digits reported by the instrument)
    2. Valve off the inner ("experimental") vacuum from the leak checker, and leak test the hosing / KF flanges. No leaks detectable at the 1.5e-8 mbar x liter / sec level
    3. Test whole thing using the needle to shoot the He at all the seals (windows and large indium seal: 5 total)
  8. Found giant leak in the large indium seal at the 10^-2 mbar x liter / sec level (this is probably bigger than it was during the run as evidenced by the inner and outer chambers venting to each other during the warmup when they didn't do this during the cold part of the run).
  9. Found large leak in one of the windows at the 10^-4 mbar x liter / sec level
  10. Marked where the leaks were with sharpie (to be cleaned off later) and opened up inner seal

Inspecting and remaking the inner seals:

  1. There was an obvious indent in the indium where the huge leak normal to the seal. (The indium is easy to stretch a little bit when making this seal, so the thickness is not always entirely uniform when laying the indium in the groove before making the seal) - I will be even more careful not to let the indium stretch when making the seal in the future
  2. A number of the screws in the window seals were not tight - I tightened these and replaced the window I had marked as leaky (there was no obvious dirt or failure in that particular seal)
  3. Cleaned the main indium groove and its mating parts, redid the indium seal, though I changed the way I made the indium seal a little bit:
    1. I used to put some of the UHV vacuum grease on the indium wire on Warren's advice before putting it in the seal
    2. Last time when making the seal, the weaker points (partial kinks, thinner bits, etc) in the indium wire seemed to get a little more stretched by greasing it up, since my method for greasing involved having the wire be in the air and me running the fingertips of my greased up gloves along the length of the wire.
    3. This time I pushed the indium in dry, then put a tiny bit of grease on the upward facing surface of the wire once it was already in its groove
  4. I sealed the inner vacuum chamber back up, again resting it on post so that I could have access to the seals with the 4He gas needle
    1. I tightened the indium seal bolts repeatedly over ~2 hours to give the indium time to flow / get squished down
    2. I did not follow a "star pattern" for the indium, since I believe it does not get bunched in the same way that a viton o-ring does; I went in a circular pattern

Leak testing #2

  1.  I measured no leaks in any of the seals at the 1.5e-8 mbar x liter / sec level!~
  2.  I didn't tighten the bolts of the large indium seal while the inner vacuum was under pressure (this may have been ridiculous - the pressure is ~1 ton when under vacuum)
  3.  I did tighten the window seals while it was under vacuum

Inspecting and remaking the outer seals:

  1. Cleaned and regreased the viton o-ring in the usual way:
    1.  Alcohol + lint free wipes to clean and remove old grease
    2. Regrease with vacuum grease
    3. Remove the bulk of the grease with clean gloves
  2. Remade outer seal:
    1. Lowered upper part of cryostat most of the way into outer can (still clamped to the table)
    2. Threaded screws into holes to guide cryostat onto the outer can with proper alignment
    3. Lowered cryostat rest of the way onto the outer can, taking no weight with the crane
    4. Tightened bolts in ~star pattern*** to prevent the viton o-ring from bunching
      1. There were 16 holes, I tightened (in order): 1, 9, 5, 13, 2, 10, 6, 14, 3, 11, 7, 15, 4, 12, 6, 16, repeat, doing no more than 1/4 turn each time
    5. Checked alignment for east cavity by scanning laser and seeing 00 flashes on the transmission camera - saw flashes without having to adjust input mirror position at all

Leak testing #3:

  1. Turned on leak checker, started pumping outer vacuum while inner vacuum was at ~1atm
  2. While leak checking, I got a big signal on the leak checker which I was not able to reproduce => not leaking from the room into the outer vacuum
  3. Noticed the inner pressure had gone down to ~470 torr (the inner vac chamber was still sealed off, even though it was at ~1 atm)
  4. Theory: I had a one time leak from the inner chamber into the outer chamber (this is maybe not so crazy: when I was leak checking the inner chamber, I had +1 atm across it, which translates to ~2200 lbs of force pushing it closed. When I was leak checking the outer chamber, I had -1 atm across it, which means I had ~2200 lbs of force pushing out against it.)

Takeaway messages:

  1. I need to tighten the screws on the inner can when the inner chamber is under vacuum
  2. I need the inner vacuum to hold w.r.t. the outer when it is overpressured by ~1atm (the exchange gas pressure we want to use),
  3. This is ~700 times less than the amount I overpressured the inner chamber w.r.t. the outer while leak testing so:
  4. I should pump the inner chamber down to ~1 atm while leak testing the outer chamber
  5. I should tighten screws while I have a vacuum sucking them in


  761   Tue May 14 17:58:29 2013 DmassDailyProgressLab WorkLeak testing update - attempt #2

Operation: Retest the inners can's seal and tighten the bolts under a vacuum

I repeated the calibration procedures outline in elog:760, found the calibrated leak source to agree with the displayed leak rate value (10^-6 mbar x liter / sec), and the KF hosing to not show any leaks at the 8e-10 level.

I measured no leaks in the inner vacuum at the 1e-8 level.

I tightened all bolts while under vacuum - some where noticably looser in this state.

I redid the viton o-ring seal (just used gloves to touch up the grease along the exposed top surface and take away any visible detritus.

I used a similar tightening pattern: 1, 9, 13, 5, 3, 11, 15, 7, 2, 10, 14, 6, 4, 12, 16, 8, repeat .

This time I pumped down the inner vacuum first to 800 mtorr and sealed it off (this is ~ the max pressure we will want to have exchange gas with)

I proceeded to pump down the outer vacuum to leak check it, and noticed (again) and spike in leak rate while pumping down, this time while P_outer ~ 5 x P_inner.

This makes me suspect that it was a virtual leak inside the outer vacuum (maybe somewhere in the superinsulation? who knows. I am, of course, uncertain that it was not a leak from the inner volume to the outer, but it seems less likely now.

I leak tested the outer vacuum (room to outer vac) - and found no leaks above the 1.2e-6 mbar x liter / sec level, where P_inner = 1.5 torr > P_outer =  45 mtorr. I was testing from the room to the outer vacuum. 

I have hooked our little turbo back up to both vacuums in parallel and am pumping them down in prep for going cold again. How quickly the pressure goes down will determine when I start cryopumping.

I do not yet have a number for what pressure it is reasonable to start going cold at. I will talk to Warren about this

  762   Wed May 15 21:14:27 2013 DmassDailyProgressLab WorkLeak testing update - attempt #2

Going to cooldown soon. Currently pumping both vacuum chambers in parallel using out little turbo unit. Current pressures:

P_inner = 6.4e-4 torr 

P_outer = 2.1e-3 torr

At Warren's suggestion I will use the leak checker to get a background rate of He leaking for the outer chamber. He said if it was < 10^-5 then I was good to go

I will need to switch some vacuum stuff around, plan:

backfill outer vacuum with N2 gas (so that I can switch the to the leak checker without letting in a huge He background)

once outer vac background < 1e-5 mbarr x liter / sec, cool down.

  2490   Mon Aug 19 17:13:17 2019 aaronUpdatestuff happensLeak valve not operable

I disconnected the SS-8BS leak valve from the shutoff valve so I could bake it, but found an unpleasant surprise. The photos probably tell it best. The labelled parts in attachment 1 are:

There was a glass sleeve of some kind inside the valve, shattered; I can't find mention of it on the elog, but maybe I don't know what it was used for in the past?

  403   Fri Jan 6 18:04:10 2012 DmassCryostatLab WorkLeaking Dewar

There appears to be what Warren described as a "honking leak" between the insulating vacuum and the experimental chamber after all (my bad). I will either reseal the dewar with the nice vacuum grease, or use the leak detector, depending on whether or not I get it on Monday.

  1458   Fri Oct 14 17:35:42 2016 ChrisThings to BuyGeneralLens and cryostat window specifications

Specs for the substrates and coatings of our new super-polished lenses and windows are posted for review/comment here:

Windows - E1600310

Lenses - E1600311

We need to finish up the optical design in order to insert lens RoCs into the lenses spec.

John Tardif stated that Coastline can accommodate about 40 1" optics in a run (i.e. we can have about 20 lenses in addition to the 20 windows).

  1655   Thu Aug 3 09:26:25 2017 brittanyUpdateelectrostatic suspensionLevitation in the OMC lab
Earlier this week, Aaron and I worked with Mariia to identify a place to setup her levitation experiment. The central question that we hope to achieve before the end of the summer is can we lift the disk given the electrode patterns Mariia has designed.
In the OMC lab, there was a big unused vacuum tank under an optics table. Koji agreed to allow us to work in there over the next few weeks and sent detailed instructions on keeping the space pristine clean. 
Mariia is working on finding all the parts for the experimental set-up.


  49   Tue Jan 4 17:12:47 2011 DmassLaserLaserLinewidth

Here is a picture of the beat measurement between the Emcore and Thorlabs laser diodes.

PD: Newport 818-BB-30 (2.5 GHz)

Analyzer: HP 8560E (~3 GHz?)


I used a couple points from the picture (via the analyzer cursor) to estimate the linewidth to be ~800 kHz (+/- 200 kHz). This is consistent with what Emcore told us the linewidth is.

  • I used 3 points and CRUDELY fit a lorentzian...this is why my error bars are huge.
  • I see what appears to be sidebands 2.3 MHz away from the center peak, 5.2 dB down (in amplitude)
  • Both lasers are (supposedly) single mode in their respective operating regions, which encompass the current and temperatures we were running them at according to their data sheets and an email to one of the Emcore engineers.
  • The sideband suppression ratio (SBSR) is quoted as > 30 dB for both of these lasers

At longer time scales, the laser frequency fluctuated a lot. Here are some videos:

Beat: 50 MHz/div and 10 dB/div (Youtube)

Beat: 10 MHz/div and 10 dB/div (Youtube)

The second video shows an uglier duckling.




Attachment 1: IMG_0376.JPG
  2089   Thu Jun 28 14:31:41 2018 Mandy Misc Lock-In Amplifier Notes

Note: Koji showed me how to use the lock-in amplifier about two weeks ago. I did some more studying by reading articles online, this is what I feel like I've learned about it. Please don't hesitate to tell me I got something wrong, thank you!


What is a Lock-In Amplifier? 
Simply defined, the lock-in amplifier is an instrument capable of detecting and measuring extremely small AC signals. It is often used to recover signals buried in a significant amount of noise. Extremely sensitive, the lock-in amplifier can provide filter-sensitivity values "Q" upwards of 100 000. For context,"a normal bandpass filter becomes difficult to use with Q’s greater than 50" [2]. 
How Does the Lock-In Amplifier Work? 
The lock-in amplifier utilizes phase-sensitive detection. Essentially, phase-sensitive detection is a method of highlighting the signal at a specific reference frequency and phase. The measurement signal is multiplied with a reference signal through the phase-sensitive detector, or multiplier. The phase-sensitive detector outputs the product of two sine waves [1].
The output is interpreted as two AC signals, one at the difference frequency, (\omega_{sig}-\omega_{ref}), and the other at the sum frequency, (\omega_{sig}+\omega_{ref})
Using a low-pass filter and given that the measurement and reference frequencies are equal, the resulting output is the difference frequency component where the measured phase is \theta=\theta_{sig}-\theta_{ref} : 
V_{psd} \approx V_{sig}cos\theta
This output is interpreted as a DC signal proportional to the amplitude of the measurement signal. 
In dual-phase lock-in amplifiers, this process is utilized twice. The second time, the reference signal is passed through a 90 degree phase-shifter before being multiplied with the measurement signal:
V_{psd2} =1/2 V_{sig}V_{LO}sin\theta(\theta_{sig}-\theta_{ref})
V_{psd2} \approx V_{sig}sin\theta
As a result, the lock-in amplifier outputs two values, which are labeled X and Y
The X output is known as the in-phase component and the Y output is known as the quadrature component. These two components collectively represent the measurement signal as a vector, relative to the reference signal [1]. 
The magnitude R and the phase shift \theta can be calculated by converting from Cartesian to polar coordinates:
\theta =tan^{-1}(Y/X)
Practice Exercise 1: Homodyne Detection Set-Up
Homodyne detection refers to the method in which  \omega_{sig}=\omega_{ref}
7 BNC cables
4 Power cords
2 BNC T-connectors
2 Synthesized function generators (SRS D345)
Oscilloscope (TDS 3014B)
Lock-in Amplifier (SRS830) 
  1. Attach T-Connector to FUNCTION of function generator 1
  2. BNC cable 1, connect one end of T-Connector to CH 1 of oscilloscope
  3. BNC cable 2, connect other end of T-Connector to INPUT A of lock-in amplifier
  4. Attach T-Connector to FUNCTION of function generator 2
  5. BNC cable 3, connect one end of T-Connector to CH 2 of oscilloscope
  6. BNC cable 4, connect other end of T-Connector to REF IN of Lock-in amplifier
  7. BNC cable 5, connect OUTPUT X of lock-in amplifier to CH 3 of oscilloscope
  8. BNC cable 6, connect OUTPUT Y of lock-in amplifier to CH 4 of oscilloscope
  9. BNC cable 7, connect TIMEBASE of function generator 1 to TIMEBASE of function generator 2 (back panel)
  1. FREQ button, set frequency of function generator 1 to 100Hz
  2. AMPL button, set amplitude of function generator 1 to 1V
  3. PHASE button, set phase of function generator 1 to 0 degrees
    1. Consider function generator 1 as the measurement signal 
  4. Set frequency of function generator 2 to 100Hz
  5. Set amplitude of function generator 2 to 1V 
  6. Set phase of function generator 2 to 0 degrees
    1. Consider function generator 2 as the local oscillator
  7. Using oscilloscope MEASURE button, display the the current X and Y outputs of the lock-in amplifier and record
  8. Increase the phase of the measurement signal from 0 to 360 degrees with a step of 20 degreesUsing python notebook, Record the X and Y outputs with each increase
  9. Create plots of Phase vs. X, Phase vs. Y, and X vs. Y 
    We can see that the Phase vs. X plot resembles a rectified cosine wave and the Phase vs. Y plot resembles a rectified sine wave.
    It also makes since that the X vs. Y plot would form an arc in the first quadrant. Notice that calculating the maximum voltage for our measurement signal is only possible when the signals are in-phase. If we continue to increase the phase-shift of one of the signals, the mean-level would move from positive to zero once reaching a 90 degrees phase difference. At 180 degrees, our output values would be opposite. Then, at 360 degrees, the output values would be similar to the initial values at the 0 degrees phase-shift. 
Practice Exercise 2: Heterodyne Detection Set-Up 
Heterodyne detection refers to the method in which  \omega_{sig}\neq\omega_{ref}
5 BNC cables
4 Power cables
2 Synthesized function generators (SRS D345)
Oscilloscope (TDS 3014B)
Lock-in Amplifier (SRS830)
  1. BNC cable 1, connect FUNCTION of function generator 1 INPUT A of lock-in amplifier
  2. BNC cable 2, connect FUNCTION of function generator 2 to REF IN of Lock-in amplifier
  3. Set frequency of function generator 2 to 100 Hz
  4. Set amplitude of function generator 2 to 1 V 
  5. Set phase of function generator 1 to 0 degrees
  6. BNC cable 3, connect OUTPUT X of lock-in amplifier to CH 1 of oscilloscope
  7. BNC cable 4, connect OUTPUT Y of lock-in amplifier to CH 2 of oscilloscope
  8. BNC cable 5, connect TIMEBASE of function generator 1 to TIMEBASE of function generator 2 (back panel)
  1. Using oscilloscope DISPLAY button, change to X-Y mode 
  2. Check that X signal is measured by X-axis and Y signal is measured by Y-axis 
  3. Set the frequency of the measurement signal to 110Hz, 101Hz, 100.5Hz, 100.1Hz, 99.9Hz, 99.5Hz, 99Hz …Observe the changes in the trajectory (direction and period) and the amplitudes of X and Y, displayed on the oscilloscope
[1] "SRS830 Lock-In Amplifier Manual" http://www.thinksrs.com/downloads/pdfs/manuals/SR830m.pdf
[3] “Principles & Operation of the Lock-In Amplifier” https://www.youtube.com/watch?v=8qJHahU3rtg
[4] “Tutorial, Teardown, & Experiments with SRS530 Lock-in Amplifier" https://www.youtube.com/watch?v=rzzliN_vTKs
[5] “Improving Student’s Understanding of Lock-In Amplifiers” https://arxiv.org/ftp/arxiv/papers/1601/1601.01285.pdf
[6] “Basics of IQ Signals and IQ modulation & demodulation” https://www.youtube.com/watch?v=h_7d-m1ehoY
  630   Mon Dec 17 18:31:56 2012 DmassDailyProgressLab WorkLocking "Progress"

Nic, Rich and myself spent some Q.T. in lab trying to get the first cavity locked. What was done today:

  1. Used an iris right before cryostat window to record the cavity axis (going through center of iris + retro beam overlapping = full constrained to cavity axis)
  2. Added EOM to the table (Thorlabs broadband EOM - 4.5e-3 rad/Volt @ 1550)
  3. Realigned beam to cavity (because EOM wedging displaced it)
  4. EOM has little homemade resonant circuit attached (elog:438) - gain ~3 @ 31 MHz
  5. Got the PDHv2 box out (schematic (DCC)) and printed this out
  6. Went through chain with Rich to see what was currently in it and refresh our memories
  7. Changed a couple resistors / caps so that it has the gain we want for ~10-50k UGF, with a pole @ ~200Hz, and a zero at the estimated cavity pole of 150kHz (there is also a switchable integrator from 200Hz and below)
  8. Checked the xsfer function of the PDHv2 box with the SR785 to make sure it was as expected (yup)
  639   Fri Dec 21 15:19:24 2012 nicolasElectronicsControl SystemLocking "Progress"

After talking to Rich and Rana I think the extra lowpass is just the cavity pole where it should be (~45kHz). We need to undo the modification for the pole compensation, but Rich is not here and this elog does not tell me what modifications were made...


Nic, Rich and myself spent some Q.T. in lab trying to get the first cavity locked. What was done today:

  1. Used an iris right before cryostat window to record the cavity axis (going through center of iris + retro beam overlapping = full constrained to cavity axis)
  2. Added EOM to the table (Thorlabs broadband EOM - 4.5e-3 rad/Volt @ 1550)
  3. Realigned beam to cavity (because EOM wedging displaced it)
  4. EOM has little homemade resonant circuit attached (elog:438) - gain ~3 @ 31 MHz
  5. Got the PDHv2 box out (schematic (DCC)) and printed this out
  6. Went through chain with Rich to see what was currently in it and refresh our memories
  7. Changed a couple resistors / caps so that it has the gain we want for ~10-50k UGF, with a pole @ ~200Hz, and a zero at the estimated cavity pole of 150kHz (there is also a switchable integrator from 200Hz and below)
  8. Checked the xsfer function of the PDHv2 box with the SR785 to make sure it was as expected (yup)

 [dym: you want to look @ U5, that is where all the shaping is happening for now. R22 and C19 are responsible for the zero (which should be around 40kHz). R22 is what I was changing to move the cavity pole compensating zero around. It should currently be 10 Ohms. U4 should be bypassed like the other stages for now: e.g. there should be a jumper in C18. The marked up schematics are @ home in LA, so I can't scan and post them, but will do so when I am back down there]

  94   Tue Mar 1 14:34:52 2011 Frank the Funkasaurus RexDailyProgressCavityLocking the Diode Lasers to the Cavity

I tried locking the diode lazers to the ULE-mirror cavity but the noise was too big and I didn't get no error signulz.

  1302   Mon Aug 17 11:49:29 2015 MattDailyProgressSiFiLong Glasgow Q measurement

After talking to Zach on Friday I learned that I need to do much longer averaging on phi/Q measurements. Assuming that the Q of the cantilever is around 1e6, this gives a timescale of around 8 hours. With this in mind I turned on the continuous measurement system for the first mode of the Glasgow cantilever on Friday afternoon around 5 and planned to collect data over the weekend. The mode stayed in lock for around 20 hours. I'm not surprised that I lost the lock as the clamp began to warm up, I usually need to realign optics when the cyrostat makes large changes in temperature. When I came in Monday morning I had completely lost the amplitude signal.

The start time of the plots is at 5pm on Friday. The amplitude set point is 1500.

The average phi was -2.5e-7, and this was after a lot of slow low pass filtering so clearly something is wrong. I also don't know what's going on with the jumps in temperature.


  1799   Sun Oct 29 21:48:41 2017 johannesNoise HuntingTransfer FunctionsLoop Modelling

I repeated the measurement of the Planex Frequency response I had described in https://nodus.ligo.caltech.edu:8081/Cryo_Lab/1597. I had only measured the West laser back then, now I recorded the transfer functions for both.

Brief Recap:

I'm using the delay line to look at the laser frequency modulation transfer function at frequencies above 100kHz, so I can resolve what happens around the UGF better. For this, I did the following:

  • For each measurement there is one 'Master' laser (ML) and one 'Slave' laser (SL) - the ML is the one whose response function is recorded and is free-running except for the applied modulation. The SL's thermistor setpoint is tuned to give a 433.4 MHz frequency difference, at which the delay line has a zero crossing
  • I'm using the beat note out of the ET3000A, which has an amplitude of 11dBm going into a power splitter. I then use a Marconi to generate a signal with a 20MHz offset to the beat at 453.4MHz. This serves as the LO for a mixer that demodulates the beat to 20MHz, which is then fed into the Red Pitaya.
  • The RP's digital PLL locks to the downconverted beat with ~500kHz bandwidth generates a voltage proportional to its control signal on its DAC. I'm using this as the error signal into an SR560 that connects to the SL frequency modulation port to drive this control signal to zero with <10Hz bandwidth. This keeps the delay line on it's zero crossing for an extended amount of time but doesn't interfere with the transfer function measurement.
  • The other splitter output goes into the delay line with 8dBm signal power, where it is split again, slightly underdriving a level 7 mixer. At the end of the long cable inside the delay line I measured an RF power of -3dBm.
  • The AG4395 source then modulates the current of the ML. For the record: since our current drivers do not have 50Ohm input impedance I externally terminated the drive signal.
  • I calibrated the delay line output with a high-Z scope channel while it was connected to the AG4395 with a Marconi dialed to 433.4MHz.
  • The last time I did this I used the HF2LI, which has trouble keeping up with the frequency fluctuations of the free-running beat. Now it worked a lot better
  • I repeated this for both lasers


A .tar.gz file with the jupyter notebook I used to produce all these plots and all data files is attached to the post.

I also recorded the amplitude response of both lasers since it can affect the scale factor of the delay line at higher frequencies. I did this with two different detectors to ensure I'm not just interpreting the PD response as amplitude modulation. I used a PDA10CF ('DC') and the ET3000A ('AC') and got very good agreement between the two - note that the ET3000A had less incident power, so I boosted the 'AC' responses by an artificial numerical factor of 1000. I regrettably did not record the power on the PDs so I can't scale it to laser output power as of now.



I divided the frequency modulation transfer functions I recorded by the square root of this amplitude response, I didn't do anything to the phase because I'm unsure how that folds in. In any case, the phase response is much flatter for amplitude modulation. The low frequency end is in rough agreement with the DC tuning coefficients I had measured (6 and 8 MHz/V)


The following can be observed: There is non-negligible phase already at 10kHz, or even 1kHz, and at 100kHz it has grown to ~45 degrees. With the cavity pole we are pretty much at 135 degrees at 100kHz, which is why we're struggling to achieve much higher locking bandwidth. It is interesting to note that it's accompanied by some amount of loss in magnitude response though.

I dug out some RFPD TF measurements from June to find out where the loop can be optimized. I found that modelling them with two poles gives a good fit to the peak out to several MHz. I made this fit in the complex plane with the python leastsq function, using the absolute value between points as the residual. I also found that the peak locations are off by about 100kHz in both cases, maybe I'll need to try again to adjust them.


Folded into the recent cavity ringdowns we now have a complete model for the plant: Current Driver, Laser Diode, Cavity, RFPD all combined give


I added a model for the new servo boxes I have been using to complete the loop and plotted both the servo alone and the resulting open loop gain (red and blue).

servo_comparison.pdf    open_loop_comparison.pdf

I also added a second, proposed modification to the servo. Because the combined downward slope of the OLG exceeds 1/f it would be possible to add a small differential contribution around 100kHz to recover some phase and cross unity gain at 300 to 400 kHz instead. Note I was using only a model for the servo for this, but I know from open loop measurements that the phase loss out to 500 kHz is marginal and dominated by the lowpass filter after the mixer.

I still want to confirm this with closed loop measurements, which have given me trouble in the past because I always find a dependency of the phase on the gain, which shouldn't be in a strictly linear system.


Attachment 1: planex_freq_response_amp_corrected.pdf
Attachment 2: rfpd_res_response_fit.pdf
Attachment 3: servo_comparison.pdf
Attachment 4: open_loop_comparison.pdf
Attachment 5: planex_amp_response.pdf
Attachment 6: plant_response.pdf
Attachment 7: servo_optimization.tar.gz
  1110   Mon Jun 30 16:35:58 2014 EvanDailyProgressPDHLoops lock with on-board rf electronics

New rf powers

I have transitioned the west PDH loop (resp. east PDH loop) so that it now locks with the CRYO–001 (resp. CRYO–002) on-board rf electronics.

This necessitated some changes to the rf power distribution. The new situation is as follows:

  • West Marconi carrier at 32.70 MHz, and east carrier is at 33.59 MHz. For each, the carrier power is 6 dBm. Each Marconi drives a hybrid splitter.
  • For both west and east, power from one port of the splitter is sent directly to the amplifier for the resonant EOM (ZX60-100VH+). This means I have removed the 16 dB of attenuation that used to precede each amplifier.
  • Power from the other port of the splitter is sent through a 3 dB attenuator and then into "LO in" on CRYO–001 for west (resp. CRYO–002 for east). This means there is 12 dBm into each ERA–5+ onboard amp, and therefore +8 dBm into each mixer (not accounting for the insertion loss of the LO phase shifter).

For some reason, the east LO phase shifter doesn't seem to be shifting. It seems like the AD587 and 10kΩ trimpot are working fine. This will require some more investigation.

West loop vitals

  • Power on cavity, off resonance: 409(1) uW 11(1) uW = 398(2) uW
  • Power on west refl PD, off resonance: 350(1) uW 20(1) uW = 330(2) uW
  • Voltage on west refl PD, off resonance: 370(2) mV (dark voltage has been subtracted)
  • On-resonance voltage dip for west refl PD: 313(8) mV
  • Error signal peak-to-peak: 335(10) mV
  • Carrier power on west trans PD: 1.96(2) V (dark voltage has been subtracted)
  • Single sideband power on west trans PD: 9.1(5) mV (dark voltage has been subtracted)

From this we infer:

  • Modulation depth is 0.136(7) rad
  • Visibility is 85(2) %

East loop vitals

Data not yet taken

  328   Mon Oct 31 23:49:31 2011 DmassElectronicsLaserLow Noise Diode Driver

I picked up the low noise diode driver from Sam Abbot. Some info:

Some measurements to be done:

  • DC Transfer function (this tells us how low noise our voltage ref needs to be)
  • RF Transfer functions
    • Modulation -> output
    • Modulation-> monitor
  •  Maximum current under load from diode
  335   Wed Nov 2 21:35:12 2011 DmassElectronicsLaserLow Noise Diode Driver Modifications

Went to downs to modify the low noise driver (DCC). The story:

The maximum current of the diode driver, loaded with the laser diodes we have, was below their nominal operating voltages. Sam and I fixed this.

WANT: Maximum current to be above the nominal operating voltage, but below the damage threshold listed by RIO


  • The maximum current is set by the voltage labeled "Vreg" on the schematic
  • The maximum current happens at an input voltage of 0V
    • Actually, it happens from 0V up to a voltage of order 1V (discussed in elog:333 under "THINKING")
  • We modify R10 to lower Vreg, which determines the maximum voltage


  1. Lifted one of the legs of R24, disconnecting Vreg
  2. Used an external power supply (with a high current limit) to change Vreg
  3. Put some 0.7V diodes in series with a fluke in current mode at the output (across P3)
  4. Adjusted Vreg (with the power supply) until we had close to the maximum desired current (within 5mA of)
  5. Measured the Vreg with a fluke
  6. Measured voltage drop across our fake diodes at this current with a fluke
  7. Wrote down difference in voltage drops across PLANEX diode and "fake" diode we used
  8. Added (or subtracted) this number to (or from) desired Vreg
  9. Calculated value of R10 needed to generate this Vreg...(Vreg = 1.25 x (R10 / R11 +1))
  10. Found resistors close to this value, chose the one on the "safe side" (smaller R, smaller Vreg)
  11. Swapped out the existing R10's for the new desired ones
  12. Confirmed Vreg with a fluke after replacing R10

The envelope math:

Channel 1 (102068)

  • I_max =125 mA (from RIO)
  • Vdiode at Imax = 1.746 V
  1. Check
  2. Check
  3. Used two 0.7 V diodes in series
  4. Chose I_max = 122mA
  5. Vreg = 8.43
  6. Vdrop = 1.567 (across the fake diodes)
  7. PLANEX Vdrop is 0.174 V higher
  8. Vreg2 = 8.6
  9.  R10 = (8.6/1.25-1)*R11 = (8.6/1.25-1)*273 =1605 Ohms
  10. Used 1600 Ohms
  11. Check
  12. Check

Channel 1 (102085)

  • I_max =150 mA (from RIO)
  • Vdiode at Imax = 2.143 ** (extrapolated from data in elog:334 - should verify this is OK before I fire up the diodes)
  1. Check
  2. Check
  3. Used three 0.7 V diodes in series
  4. Chose I_max = 147mA
  5. Vreg = 10.65
  6. Vdrop = 2.36 (across the fake diodes)
  7. PLANEX Vdrop is 0.217 V lower
  8. Vreg2 = 10.43
  9. R10 = ( 10.43 / 1.25 - 1 ) * R11 = (10.43 / 1.25 - 1 ) * 273 = 2.012 kOhms
  10. Used 2kOhms
  11. Check
  12. Check (Vreg final = 10.46)
  1810   Sat Nov 4 16:06:08 2017 ZachNoise HuntingSiFiLow beat noise recovery

After a series of unfortunate events, I have recovered the lowest beat noise from 10/26/2017 (actually, slightly better---c.f. CRYO:1798):


Below is a quick summary of what went down...

Scope tumble

I started trying to investigate the spookily higher noise from earlier in the week:


Something frighening has happened on Halloween. For some reason, the noise appears to be ~3x higher almost everywhere in frequency. This seems like it can't be real (and hopefully it isn't).

Before making a beat noise measurement, I noticed that the beat signal was a little weaker than it should be, so I set about realigning the TRANS setup. Unfortunately, I placed the scope I was using to monitor the beat while tweaking too precariously, and it fell, tugging a bunch of cables with it. Luckily, everything was relatively well strain-protected, but the force of the fall was enough to rip the ends off 2 cables (1 BNC, 1 SMA) and break an isolated SMA bulkead feedthrough on the RF oscillator chassis:

I was unable to find a replacement isolated part either in Bridge or in Downs, so for the moment I've installed a non-isolated one---we should order some isolated SMA feedthroughs.

Solving the Halloween noise mystery

With that behind me, I got back to the TRANS tuning and was able to increase the beat amplitude by a factor of ~2. At this point, I made a beat noise measurement and found that it had returned to its pre-Halloween level everywhere except at higher frequencies. In playing around with the Red Pitaya PLL controls, I noticed that---depending on how the parameters were set---the loop seemed to lock on harmonics of the fundamental beat note. In particular, I could make it lock on the 3rd harmonic, and the resultant ASD looked exactly like the one I measured on Halloween (i.e., exactly 3x higher than the usual level, everywhere). So, it seems like this mystery is "solved": one should make sure that the loop is actually locked on the fundamental beat note at all times.

Of course, it begs the question of why the loop locks so easily on what should be negligibly small harmonics. I don't think the distortion is generated at the RP input itself, as it's always a struggle for me to make sure the beat signal fills the ADC sufficiently for the loop to run well (in addition to adjusting the digital input gain). I am using the standalone Agilent RF amplifier box, so I should probably look into nonlinearities there, as well.

Recovering the low-noise state at high frequencies

At this point, I had recovered a high-frequency noise state that was ~3x higher than the best previous measurement on 10/26/2017. The zoomed plot below shows a recap of how I eventually recovered the lowest noise state.

Starting from the initial state (BLUE), I simply pushed the servo gains as high as I could and recovered the previous low-noise level at the highest frequencies (RED). However, the level between 500 Hz and 10 kHz was still ~2x higher than it was before. I continued troubleshooting for a while with no luck (in particular, I played with the PLL settings at length, as it can often yield much higher apparent noise if the digital UGF is too low). Eventually, I started playing with the laser power. Interestingly, I found that the elevated mid-band noise could be suppressed if the E laser power was increased, though the W laser power didn't exhibit the same effect. With this increased E laser power and optimized loop gains, I actually obtained the best high-frequency noise level yet (ORANGE).

The E laser power sensitivity is a nice new clue, but it's confusing. The naive explanation is that it's simply an increase of optical gain, leading to a commensurate suppression of sensing noise. However, I tried replicating this effect by reducing the optical attenuation at the E REFL PD (in lieu of increasing the power at the laser head), and I could not achieve similarly low noise. A complicating factor is that the incident light power on the REFL PD is close to saturation when unlocked. Any non-idealities from this relatively high incident power should not care about why the power is greater (i.e., from increased laser output or from less attenuation locally), so I'm still scratching my head.

I'm not going to lose too much sleep on this until I install the new PDH boxes, hopefully this weekend. If the ultimate noise floor here is dominated by sensing noise (which I know to be in the ballpark), this should get much better with the much-improved electronics noise.


  434   Tue Mar 6 12:50:49 2012 DmassElectronicsLab WorkM-I-C! K-E-Y!

Put the reflected beam on the PD. Some problems went away when I did this.

Below is a picture of the setup with labels.

Photographs of the error signal sweeps (wrote down levels - will enter these later)


Did some sweeps with the power turned down to check my gammas!

Gamma = 0.05 - 0.01 (From noisy sweeps)

Gamma = 0.165 (Calculated from cavity transmission)

Attachment 1: InAirLayoutPics.png
  435   Tue Mar 6 23:42:10 2012 DmassElectronicsLab WorkM-I-C! K-E-Y!

Moving forward...here are the mods I am planning on making to the PDHv2 box. Changes are in pink, blue, and red.

I am working with a lower power level now on the 1811. I may be able to turn this up in the future.

Using a discriminator of 5e-6 V/Hz (about what I measured in the lab)...

  • Gain of 10 in first stage
  • Tunable gain 1-11 in second stage
  • Switchable boost in stages 1-3 (all with gain of 3, and crossover freq of 5kHz)
  • Switchable integrator in stage 4 with gain of 1, and pole ~1 Hz

The opening of the switches happens in order 4, 3, 2, 1, so in this configuration I can:

  • Lock with a single pole at 1 Hz and a UGF ~40 kHz
  • Tune the UGF via the trim pot in the second stage
  • Engage the integrator
  • Engage boosts 3 -> 1

 [EDIT: fixed the plot - I said "open" for a number of resistor values when I meant "0 ohms"]

Attachment 1: PDHv2mod.png
  436   Wed Mar 7 01:00:18 2012 FrankElectronicsLab WorkM-I-C! K-E-Y!

your blue cuts and red shorts don't make sense at all, same for the changes to the values. The whole thing won't work at all.

DYM: I was sleepy. All "OPEN" resistors should be instead 0 ohms. updating.

FRANK: still, cuts and shorts won't make sense (WHAT DOES THIS MEAN???) - i suggest: come up with TF for each stage, then make a simple drawing of the schematic (opamp circuitry)  you want to implement and then check which parts in the PDH2 schematic you have to modify to implement that function - there is absolutely no need to cut traces or make some other shorts

DYM: TOO VAGUE! I don't really know how to parse this into useful information, considering I stared at the board for several days, talked to Koji antrying to understand how it works, talked to Rana about what we want, and the board topology doesn't seem to support what we want. My modifications are the only thing I came up with which satisfy those requirements...which are:

  • Have a true switchable integrator
  • Have a tunable gain
  • Be unconditionally stable so locking isn't an excercise in uncertainty
  • Have switchable boost stages
  • Be (initially) a single simple pole with no crazy triple DC boost


Moving forward...here are the mods I am planning on making to the PDHv2 box. Changes are in pink, blue, and red.

I am working with a lower power level now on the 1811. I may be able to turn this up in the future.

Using a discriminator of 5e-6 V/Hz (about what I measured in the lab)...

  • Gain of 10 in first stage
  • Tunable gain 1-11 in second stage
  • Switchable boost in stages 1-3 (all with gain of 3, and crossover freq of 5kHz)
  • Switchable integrator in stage 4 with gain of 1, and pole ~1 Hz

The opening of the switches happens in order 4, 3, 2, 1, so in this configuration I can:

  • Lock with a single pole at 1 Hz and a UGF ~40 kHz
  • Tune the UGF via the trim pot in the second stage
  • Engage the integrator
  • Engage boosts 3 -> 1



  437   Thu Mar 8 00:50:52 2012 DmassElectronicsLab WorkM-I-C! K-E-Y!

I saw Frank in person and communication happened. I learned that something I was trying to do was naive:

  • Having a gain of 10^5 at while acquiring lock might be rather foolish
  • This is a consequence of trying to acquire with a UGF of 100 kHz, and a pole at 1 Hz

I will move my pole up to 100-1kHz, and proceed with caution.


  • If you look at the schematic posted in elog:435, TF stages 1-3 have space for a series RC in the feedback, with a switchable R in parallel (e.g. R10, R11 and C16).
  •  I wanted to make a boost stage which goes from flat to a P-I shape (flat with a low frequency integrator)
  • The closest I could come to this in the current topology was to make R10 and R11 approximately equal
  • If I do this...the high frequency gain changes by a factor of 2 when I switch R11 in and out. Cascading a few of these would mean I was shifting my UGF by up to a factor of EIGHT by putting in boost
  • This seems bad.
  • If I make R11>>R10, I can make something which doesn't have change my UGF by much (~R10/R11), but it means I would have to lock with a finite DC boost on.
  • This seems bad (but it may not be as bad as I initially thought)
  • I can make something which is flat, and has a switchable integrator without too much trouble as described in elog:435

After talking to Frank / Rich, I decided to solve a simple problem before I solved the most complicated craziness I could think of, so I am making my first pass of the servo look like:

  • UGF tunable from ~20k -> 200k
  • Single pole ~200 Hz (40-60 dB DC gain on lock acquisition)
  • Pole switchable to integrator

A list of what part subs/additions I will make is here google docs.

For now I will do no trace cutting or jumping


  1023   Tue Feb 25 23:52:18 2014 ZachElectronicsGeneralMAX333 is quiet

Since we are thinking of using the MAX333 quad analog switch in a low-noise environment for the omniPD (see thread at CRYO:1016), we want to make sure it is not noisy.

I measured its noise tonight and found that the shorted-input noise of both NC and NO channels was limited by thermal noise from the ~130-ohm closed-circuit resistance, to the level measurable using a LT1128 preamp:


Here are a photo and sketch of the measurement setup:

20140225_214609.jpg 20140225_234647.jpg


  1557   Fri Apr 7 18:15:48 2017 brittanyDailyProgressCryo QMEDM screens

I updated the current sitemap.adl medm screen to have links to the SiQ experiment (which is the 3 letter name for the CryoQ exp't)


I designed an overview page of the experiment with links to pages for driving the ESD and monitoring the output of the QPD.


The next thing is that I will be testing the inputs into the anti-aliasing chassis to see if I understand how the whole thing is hanging together.

  1345   Mon Jan 4 20:05:01 2016 ZachElectronicsSiFiMade five 2-m LIGO power cables for rack electronics

With Ben Abbott's help, I finally got around to making up some LIGO power cables to go from the 18V power supply (see CRYO:1305) to the various rack-mount units. Rich and Gary had ordered enough supplies (cable, crip-on pins, connectors and backshells) to make on the order of ten 2-m cables of each the 18V and 24V varieties. I made five of the 18V only, since that's all I need.

I found the cables fairly easy to make using the tools at Downs. If anyone is interested in making some of their own, just ask me and I'll walk you through it.

  1123   Wed Jul 9 01:10:42 2014 DmassDailyProgressPDHMade new low pass filters for after mixer

I made some new filters for the IF port of the PDH mixer - pics below


I don't know how to correctly model what appears to be a self resonance due to the series capacitance of the inductors used (3.3uH)

I can raise the SRF by decreasing the inductors and increasing the capacitors, but this costs us phase at low frequency (b/c of the impedance of the inductors w.r.t the 50 ohm output resistor)


Attachment 1: MeasSummary.pdf
  213   Mon Jun 27 10:26:50 2011 JennyDailyProgressGeneralMaking COMSOL solution match analytical solution

I’ve made a basic COMSOL model.

I only modeled conduction through the silicon cylinder.


Here’s my cylinder: cylinder1.png


The length is 50 cm and the radius is 10 cm. The entire cylinder was initially at 120K. I set one end boundary to 130K at time t=0 and thermally insulated the other boundaries. I ran a time-dependent simulation and recorded the temperature at the other end of the cylinder (in the center of the circular boundary) as a function of time.


I generated this plot:cylinderprobe1.png


The blue trace is temperature at the center of the circular boundary at which I set the temperature. Obviously it stays at 301 K throughout the simulation. The green trace is temperature at the other end. This simulation produced a time-constant of 1500 seconds.



As a first approximation I treated the cylinder as a 1-D heat problem assuming the cylinder could be modeled as a resistor and a capacitor in series. The solution to this problem T(x, t) assuming the initial/boundary conditions T(0<x≤L, 0)=300K and T(0, t)=301K is (1-exp[t/(RthCth)])+300 with Rth=L/(A*k) and Cth=rho*V*Cp


The time constant for the system is tau=RC=L^2*rho*Cp/k. In my problem L=0.5 m, rho=2330 kg/m^3, Cp=703 J/(kg*K) and k=163 W/(m*K). This gives a time constant of 2512 s.


Looking at the simulation, the time constant is 1500 s. At first I wasn’t sure why the values didn’t match up. Doubling the length of the cylinder did quadruple the time constant in my simulation as expected. Changing mesh size, and radius of the cylinder did not change the time constant (as expected).


I realized that the problem was that since the silicon is the only material in the system, it produces the dominating thermal resistance and thermal capacitance, which cannot be modeled as a lumped resistance and lumped capacitance one after the other. The cylinder does not heat up uniformly, as shown in this plot of temperature gradients across the cylinder at time t=1500 s.


Instead the system is more like an infinite number of little resistors and capacitors R—C—R—C—R—C—…. This problem is harder, so instead I think it makes more sense to change my initial model so that the simple approximation works.


To ensure that my simulated solution and my analytical solution match up, I am choosing to change the COMSOL model so that the approximation in which I treat it as a single resistance and a single capacitance is valid.  I’ll add a piece of insulating material at one end of the cylinder. Since foam has a low thermal conductivity, and a low specific heat, I can treat it as a resistance with no capacitance and the silicon as a capacitance with no resistance. Adding the foam should make the time constant of the system much larger, and it should make the analytical approximation more closely match the COMSOL output.


Here’s the new model: cylinderfoam1.png


The red is foam and the blue is silicon. I may change the dimensions from what's shown but this is the general picture. I will update soon with my findings.


  214   Mon Jun 27 15:28:43 2011 JennyDailyProgressSimulationMaking COMSOL solution match analytical solution part II


I ran this new simulation of silicon and foam. Here's a picture of the model (again with the foam in red and the silicon in blue):


I changed the dimensions to two cylinders, each of length 1 mm and each of radius 10 cm for two main reasons:

1) A short, wide silicon cylinder has a lower thermal resistance than a long one. The approximation in which I neglect its resistance becomes more accurate.

2) I didn't want to make the foam cylinder so long that its thermal resistance was so large that the time constant of the system was on the order of days. These dimensions gave a time constant (in my pen-and-paper calculations) of 43.6 seconds.


My first run popped out a time constant of 50 seconds. That's a 13% difference from my calculations.

Then I noticed that the solver was only updating the solution every ~10 seconds. I told it to update every second instead and got a time constant of 46 seconds, a 5% difference.


In summary:

Percent difference between calculated time constant and COMSOL time constant

  • Si w/o foam, computer-generated timesteps: 40%
  • Si w/ foam, computer-generated timesteps: 13%
  • Si w/ foam, max timestep of 1 second: 5%


Here's a plot of temperature at each end of the silicon cylinder. disk_si_foam_probe.png

It just looks like one line because there is essentially no temperature gradient over the cylinder. Both ends are at the same temperature to the 4th decimal place.

  840   Wed Aug 28 18:37:09 2013 EvanElectronicsGeneralMaking the invert switch behave

The addition of the 2.5 MΩ bypass path (from the output of U1 to the inverting input of U6) breaks the symmetry of the optional-invert output buffer (U6) on the PDH2 board. Consequently, the bypass only works when the invert switch is engaged (i.e., when S1 on K6 is connected to NO, thereby making the output buffer act as an inverter).

To first order, the solution to this is to add a second, identical bypass path going from the output of U1 to the non-inverting input of U6. For CRYO-002, I've gone ahead and added 2.5 MΩ going from the output of U1 to pin 2 of relay K6. The resulting transfer functions are shown in the first attachment, for two gain settings. The second attachment gives the inverting TF divided by the noninverting TF. They seem to be pretty well matched up to a few hundred kilohertz (although I'm not sure what's going on around 100 Hz). Unfortunately the new noninverting TF seems to be missing 40° relative to the inverting TF at 1 MHz. I'm hoping that we can recover this by tweaking the resistance of the second bypass path.

Attachment 1: cryo-002_tf_inv.pdf
Attachment 2: cryo-002_tf_inv_quot.pdf
Attachment 3: cryo-002_tf_inv_code.zip
ELOG V3.1.3-