We've finally arrived at the point where we have a beat note: East laser, in transmission of one of the SiCavs, beating with West laser, in transmission of the RefCav and going through a fiber to the main table. For the time being the beat note coming out of the NewFocus 1611 has an amplitude of about -7 dBm, but can likely be improved with some more careful modematching and balancing of light powers.

We evacuated both tanks (RefCav and cryostat) and found a beat near 200 MHz, which still seemed to have a drift of about 20 MHz/hour. The fact that the East laser temperature had to be human-servo-ed more frequently points at the SiCav as the culprit, which could explain this trend with a creep of about 40 mK/hour.

The last couple weeks were a race to get the beat note so we can see the temperature modulation from the visible laser pointer in Nizar's project, so the setups haven't been properly characterized, yet. We did get the Marconi locked to the beat, and saw the laser pointer modulation show up in the power spectrum of the control signal to the Marconi. We are thus only the data acquisition (and some calibration) away from an estimate of the beat noise, and the characterization of the setups will follow in the next couple of days.

These arrived today, at long last. There was some kind of fax number error in TechMart and apparently they hadn't gotten the P.O.

Anyway, they look good, and I can proceed with installing them into the system and continuing with the in-vacuum testing. These should make the opening and closing of the tank way less precarious, since the vertical modes can be kept to a much lower amplitude when lifting/lowering the upper assembly.

For a reminder of how these fit into the suspension assembly, see CRYO:1417.

that last attempt didn't work since nodus had been configured so as to have multiple copies of the Wiki directories and I upgraded the wrong one. EricQ has now solved that issue by replacing the directories in the /export/home (which is the nodus local disk) with symbolic links to the /users/public_html/ dir where the Wiki directories actually are. This was all due to the strange coincidence of me restarting nodus (rarely ever done) just a couple days after the Apache config files were accidentally deleted (even more rare) leading us to find that these conf files weren't being backed up (bad, but not so rare).

Looks like the new wiki version is now in the right place and actually running for CryoWiki, but not the AIC or ATF wikis.

I've made a backup of the current version of CryoWiki while updating its version of DokuWiki from 2010 to 2016.

The back up exists on nodus in public_html/ as well as in our shared CryoLIGO folder on caltech.box.com.

Unimportant note on compression (go back to work, DO NOT READ):

The standard gzip compression gave a file size of 19834173 bytes,

                               'gzip --best' gave 19774283, 

                           and 'bzip2 --best' gave 19242273

Mike Zucker has a nice review of 1064 cards. He found Lumitek was best, but other phosphor cards (which require charging) were in the same league. All were far more sensitive than upconversion cards.

Looks like ThorLabs VRC2 is phosphor based, while VRC4 may be an upconversion card.

IR Cards for 1550 nm:

• Thorlabs has several different types of Laser Viewing Cards. Of these, the NIR detecting VRC4 (see attachment #2) seems most aimed at 1064 and 1550 nm. But the VRC2 might be just as good. Need a lights off comparison with a dim 1550 beam to break the tie. add photo of empirical result to this elog
• Lumitek is listed as the place to get cards in the CryoWiki: (http://www.lumitek.com/Sensorcatalog.PDF). The Q-32 and Q-42 both have response at 1550 nm. Need to get one of each and compare.
• Newport has several NIR cards which they claim go up to 1600 to 1700 nm. The F-IRC2 and F-IRC2-S seem like they might be good. Also maybe the F-IRC4. ~ $80-160
• Edmund Optics has one for $75

The MZ doc that Chris mentions above has a minimum detectable power of 10 uW for a 1 mm dia beam. That's the equivalent of 1200 uW/cm². The Newport cards report numbers of more like 10 uW/cm², so there's a factor of 1000 difference in the detectability , according to these numbers.

IR Viewers for sale which cover 1550 nm:

• Newport sells a wide variety and also compares their spectral responsivity in some arbitrary units. Several do well below 1300, but not past there (see attachment #1). There are also a few which work up to 2000 nm. Notice that the sensitvity at 1550 nm is 100x less than at 1064 nm. It also has an optional face mask attachment so that you can operate it hands free like a Navy SEAL.
• The usual BLUE body, short handle viewer that we use in most places ($1800 on Edmunds), claims to have a sensitive band up to 1300 nm, but not sure if this really stops before 1550 nm.
• The older, grey body one with the long handle (also $1800 on Edmunds), claims to go to 1350 nm and uses a P-20 Phosphor screen.
• There is a version of this grey FIND-R-SCOPE which goes up to 1550 using a ??. Its available for $2155 from FJW Optical Systems. The response is 1000-10000x less at 1550 relative to 1064 so YMMV. They try to hand select the ones with high responsivity at 1550 if you ask them and they sell a visible light filter to help in spotting faint things.

Cameras:

The main issue with cameras is that they either use silicon or InGaAs for the detection array. Silicon has a negligible absorption at 1550 nm, so only InGaAs is useable. Presumably, we should also be able to use InGaAs cameras for 2000 nm. The CryoWiki has a section on IR Cameras. Can someone please reply to this entry with the model and make of the cameras we have today and your feelings about them?

Another variety is based on how IR viewers work: they have an IR absorbing phosphor which is deposited on the silicon CCD. This anti-Stokes phosphor then emits light in the visible (typically greenish) so that the CCD can pick it up. As a comparison to the IR cards above, the phosphor coated cameras take ~10 mW/cm2 to use up the full range of the signal. Assuming that we're not integrating and that the camera has a dynamic range of ~50 dB (10 bit), the camera's minimum detectable power would be ~30 uW/cm². Bleh. See attachment #3.

• Why we should use InGaAs cameras instead of phosphor coated CCDs, from Ophir. Phosphor is nonlinear. Difficult to measure the wings of the beams. OK for finding spots though.
•  

When I went to re-measure the noise after having cleaned the optical surfaces, I noticed a strong hop-like behavior in the laser (judging from flip-flopping in the TRANS between two DC values). I have seen this before, and it usually goes away if I scan the laser frequency to a different cavity axial mode, but this time it persisted. It was bad enough that there were clear non-stationary jumps in the control noise spectrum, so I figured something had to be done.

I remembered that I had bought fiber isolators for eventual use in the setup, which I had not yet permanently installed due to their extra loss, which makes alignment all the more difficult. This seemed like a good time to try them out. I couldn't locate them in my stash, but I eventually found them on the central optical table. One is currently being used in the new refcav setup---I took the second one back and installed it in mine. (NOTE: If the new setup requires an isolator, we need to buy more as I bought these two on our budget for myself.)

With that, the hopping went away. A new spectrum is plotted below, alongside a recent in-air measurement and the laser frequency noise.

Conclusion: I wish I could have verified an improvement in the scattering noise without having to install the isolator, but in any case, it's gone now.

Time to proceed with the West setup alignment, beat measurement setup and more vacuum testing!

Tonight, I completely disassembled the cryostat and cleaned every surface. I also removed, cleaned and reinstalled the viewports, and I drag-wiped all cavity mirrors.

I then put it all back together, replaced it in the optical chain, realigned the laser and did some preliminary locking. Despite using the "L-bracket trick", the f*#@ing o-ring slipped out as I was closing the chamber, so I was unable to begin pumpdown at this moment. 

After the suspension rings down tomorrow, I will make a new measurement to see if the scattering shelf has improved.

When I came in this evening, the vacuum was holding at around 2 x 10^-4 Torr, which of course is not that great, but I figured it was fine at least for some initial in-vacuum measurements.

I'm not sure if this is cause for suspicion or not, but it seemed like the cavity needed realignment. I may have tricked myself into thinking so, but I could not find the TEM₀₀ resonance and had to proceed with fine alignment using the camera and then the photodetector again. Typically, once the system is aligned and the cryostat is not moved, the alignment maintains itself quite well. Weird.

Anyway, here is a noise measurement from tonight, compared with the measurement from last night (which was shown in the replied-to post to be nominal) and the laser frequency noise:

Notes:

• The first obvious thing is that the scattered light shelf is way worse. Resolving individual averages, I could see that the cutoff frequency varied from as low as ~300 Hz up to the longer-term average seen here of around 600 Hz. Some more on this later.
• The other obvious (and expected) thing is that the fundamental suspension modes are much larger now in vacuum. I probably rang them up considerably while realigning the system, but they did not attenuate strongly over the time I was in the lab. This of course goes hand-in-hand with the larger scattering shelf.

Interested to see if the vacuum pump was doing some of the ringing up, I shut the valve and disconnected the hose. After waiting about half an hour, there was no difference in the scattering shelf. I also noticed that the chamber pressure rose by a factor of 10 to about 2 x 10^-3 Torr within that time.

Conclusions:

1. The vacuum system is not sealed very well at the moment. I will have to re-check all the seals and viewports.
2. It's time to clean all the optical surfaces, including the cavity mirrors and the windows. This will be a good opportunity to kill 2 birds with one stone.
3. Depending somewhat on how the cleaning goes, I may have to work on a suspension damping solution sooner rather than later.

This weekend, I completed the recovery of the system from the recent suspension-wire-breaking episode.

As of last week, the wire was re-pulled, but the payload required a re-balancing. I did the balancing with the payload suspended from the cantilevered-breadboard test setup as seen in the replied-to post, as I've found this to give an acceptable leveling of the board once transferred to the chamber. I then reinstalled the suspension upper assembly onto the coldplate, suspended the payload from it and hoisted the experiment back down into the cryostat. Once it was in place, I checked that the apertures were clear.

As we've had issues with the o-ring slipping out of place (its groove faces downward---Steve at IRLabs informed me that it was made this way because he was under the impression that we were going to close the cryostat upside down and then flip it, which is apparently common), I used a technique recommended by the manufacturer whereby a number of L-brackets are taped in place to hold the o-ring in until the chamber is closed, then slid out (see below). This seems to have worked.

Since the payload has been newly suspended, the cavity axis had shifted in an unknown way, and so I had to redo the alignment. For this, I used the HeNe back-injection and co-alignment scheme that I did in CRYO:1370. After co-aligning the beams, I saw flashes on the CCD in transmission and then roughly tuned the input alignment further to maximize the TEM₀₀ transmission. Then, as usual, I replaced the CCD camera with a photodetector and fine-tuned the alignment by sweeping across the resonance. The cavity then locked with the usual test loop (see CRYO:1418).

After letting the suspension ring down somewhat, I re-measured the noise and found it at the usual level, albeit with slightly more action in the scattering shelf due to the high suspension velocity:

With the o-ring now in place, I was able to begin pumpdown of the chamber to do our first noise measurements in vacuum. I left it pumping overnight:

NOTE: I found two MKS gauges sitting on the workbench by the stereo. One was labeled NFG and so I used the other in my setup. If this is not the one I'm supposed to be using, please let me know and I'll replace it.

Next up:

• Check the vacuum performance (for leaks, etc.)
• Check the single-cavity noise in vacuum against the in-air level that we've been measuring
• Align and lock the West system
• Rebuild the transmission beat setup and take a differential measurement
  • As has been reported before, the lock appears to be marginal due to the high amplitude of body modes. I have tried to do some crude damping to no avail, and my other option is to improve the locking loops, which must be done eventually anyway. Until I do this, I'm not sure how successful the differential measurements will be, or for that matter if we're seeing nonlinear noise in the single-cavity measurements.

A while back, one of the suspension wires spontaneously snapped off (apparently due to fatigue from bending at the upper clamp launch point):

This afternoon, after much procrastination and despite numerous offers from Chris for help, I finally re-pulled the 3rd wire and crimped the copper ferrule to the end. A test hang convinced me that everything went well, though the payload will need a meticulous re-balancing (there is extra ad-hoc weight added to balance in these photos).:

(Quick summary of last two weeks' beam profile task)

For the east laser I obtained the following parameters using ten datapoints from the razor edge obstruction technique:

I received the long awaited quote from Laseroptik, and also from Coastline. I put them on the Wiki page https://nodus.ligo.caltech.edu:30889/CryoWiki/doku.php?id=documents:2nd_gen_cavities#drawings_and_pricing

I threw together a design for the structural framing for the clean hood for the cryostat table. We'll be using u-channel strut framing, all can be bought from McMaster, I opted for black powder-coated steel. The parts to hold all the joints together are available on the website.

The HEPA filter will sit on the two intermediate struts at the very top. There's a sprinkler above the cryostat table that restricts its placement (it's very close to the exact center of the table length-wise), and we need to get the building management involved to  fire-safety-proof assembly.

There will be a rail-guided pulley system on the singular cross-beam suspended from the top, and the cross-beam itself will be able to move along the side-struts, giving us full 2D freedom for the suspension point, but of course we'll be able to clamp it in position. Using the volume reported by the solidworks model with the density of aluminum, the empty weight of the cryostat is around 82 lbs. All the parts that I plan to put in are rated well above that, and the steel construction will easily be sturdy enough to support the air filter and lifting the loaded cryostat.

I'm also thinking about putting another level of side beams at the bottom and creating some underneath-shelving for equipment, but that can be added later if needed.

I already put an order in for the parts with lead time (2 weeks). Those are the dual vertical posts (custom order), the double-beams on the upper ends, and the side rails for the pulley cross-beam. Everything else is stocked and can be obtained overnight, with a few exceptions that are shipping from Chicago instead of LA.

As we discussed last week, I sent the copper ferrule used in the aLIGO OMC and SiFi suspension designs (drawing attached) to MillItNow to see if they can make them. Their initial reply was that the central hole was too long for how narrow it was. I am waiting ot hear back about how much wider it must be, or if making it a little bit shorter can allow for similar hole size. Stay tuned...

(Johannes, Chris)

We received an anti-alias chassis and an anti-image chassis from Rich, and installed them in the cymac rack. We also cleaned up the cabling and adapters that go between the ADC/DAC and the new chassis: they're now tucked away inside the cymac for the most part, instead of hanging out loose behind the rack. To check the basic functionality of the system, we made a quick loopback test. So far so good.

To reach this point we had to take a circuitous detour into troubleshooting the hardware of the cymac. After the cleanup, the main issue we observed was the failure of the DAC card to output anything. This problem was flagged on the GDS screen: fourth bit of DAC stateword X1:FEC-5_DAC_STAT_0 went red. It turns out the fourth bit goes red when the DAC's FIFO buffer either overflows or underflows. In our case the buffer was overflowing, as if the DAC never got a clock signal to trigger it to generate outputs and drain the buffer.

By trending the stateword we could see the DAC had been working properly before we started. Then we systematically swapped out the clock source (DS345 sync port), the cables, the DAC adapter board (D060061), and the DAC card itself -- to no avail.

Finally we traced the clock signal, and found out we had two faulty DAC adapter boards on our hands. Both of them had no continuity between the clock input connector and any other pins. (Also, both were labeled D060061 "rev B1", whereas a working board that we got from Rich was labeled "rev B". This rev B may have been the board that was originally installed in our system before the cleanup initiative.)

There were a few other, intermittent issues we ran into in the course of dismantling/reassembling/rebooting the cymac several times:

• No clock signal reaching the ADC card: causes the IOP to hang on startup. The fix was to wiggle/reseat the cable between the ADC and its adapter card.
• Very early boot failures (dead black screen, or screen with vertical stripes). The fix: reseat the cymac's removable power supply.
• Hyperthreading turns itself on, even though it's disabled in the BIOS. This causes lots of jitter in the runtime of the front end code, and 24 cores appear in /proc/cpuinfo instead of 12. To work around: enter BIOS setup, confirm threading is off, save the setting and reboot. It's also possible to shut down the offending HT cores using sysfs (/sys/devices/system/cpu/cpuN/online)

Also, when you probe the clock signal while connected to the ADC/DAC, it displays a lot of ringing. This doesn't happen when the clock is connected only to the scope. Things have been working this way since the beginning, but it looks like we'd benefit from a proper 50 ohm terminated setup here or a timing fanout circuit.

We received our recent order of 10 FZ wafers from UniversityWafer today. The specs are listed below:

Size: 100 mm

Thickness: 500 um

Type: Undoped

Orientation: <100>

Polish: Double-side polish

Resistivity: >20,000 ohm-cm

It's my understanding that some of these wafers will be shipped off to have black coatings applied.

The Cryostat base diameter is 19 inches.
 

Things we need (to get/buy/do) for the lab:

• New test cavities (we're only days away from a quote for the mirrors, have spacer quote)
• New cryostat windows
• RIO Planex laser for auxiliary bench
• Temp Controller for auxiliary bench laser
• New PDH controllers
• Current driver to laser cabling for all three Planex lasers
• 9-pin D-SUB twisted pair cables (for DAQ)
• 9-pin D-SUB to 4xBNC breakouts for optical tables
• New D/A card for Cymac1?
• Free-space Faraday isolator for aux laser? (we're using a fiber-coupled one for now, may be sufficient)
• AOM for aux laser? (optional, on the list for completeness)
• Rotational mounts for waveplates. We do have enough waveplates, just no mounts for them.
• Optical posts as needed
• Cut the G11 plate into shape to go under the cryostat permanently
• Additional rack shelving
• New beefy power supply to help power all the new electronics (USOs, laser current drivers, AA/AI chassis', PDH will all be powered by distribution system)
• Fuses for the power distribution system
• 3x KVM switch (for Gaston, Cymac1, Cryoaux)
• Swiffer charges
• Some shelving solution for underneath the now raised tables
• Tape measure
• Probably many more things I forgot to put on the list

Other thoughts:

I would like to acquire one of these for the lab: http://www.edmundoptics.com/cameras/nir-uv-cameras/1500-1600nm-nir-ccd-usb-2-0-camera/3599/ so we can get a better idea of the modes coming out of the collimators. The razor edge method is arguably the most precise for Gaussian beams, but with fibers and collimators it's possible to have slighly elliptical, rotated or otherwise misshapen beams. I have used the plain CCD version of it for 1064 nm before, without the phosphor coating, with good results. The driver (same one for this camera, I found out) is easy to install and even lets you load frames directly into Matlab, from where you have full control over exposure time, gain, and other things. At $2,600 it's maybe a little expensive for just a camera, but still way cheaper than complete beam profiling solutions.

Will the remodeled AA chassis also have 9-pin D-SUB inputs? If not, we need to get corresponding breakouts for its connectors. Will there be a remodeled AI Chassis as well?

I placed an order for 4 of these from MillItNow this afternoon.

The system has not been operated in several weeks, so I checked last night that everything was still locking properly and that the noise level was similar. It was.

The signal chain for this test was:

Laser -> Cavity -> GYRO RFPD S/N 2 -> -3dB -> uPDH demod -> SR560, zpk([],10,50) -> Laser

Notes

• The system still locks reliable with a UGF of 100 kHz
• The loop shape currently prevents locking above this frequency, but it should be possible to push it further
• The noise is similar to how it was recently.
  • Recall that the noise above 100 Hz can be anomalously high due to upconversion from a weak lock---the lock was rather stable last night.
  • The noise between a few Hz and 100 Hz is dominated by a scattering shelf, and this should not be expected to look the same from day to day depending on the excitation of the suspension.
  • It's unclear how the current noise level can dip below the expected laser frequency noise. The calibration last night (and every recent time) was done directly by sweeping the laser current and looking at the sidebands in transmission---this should not be off by factors anywhere near unity. It could be that a.) the original calibration of the laser frequency noise is off by a bigger factor, b.) the laser frequency noise is actually lower at this current operating point, or both.

For a while now, I've been meaning to make a new version of the blade stops for the SiFi suspension. The problem with the first version is that the stop contacts the blade at point too far away from the tip, so the tip remains hyperextended when the stop is engaged. Below is a photo demonstrating this during the original suspension checkout. In the photo, the blade is actually resting against the edge of the blade stop, as opposed to the blade stop screw (which wasn't yet installed); the hyperextension is actually worse when the contact is made with the screw as designed, though I don't have a photo of this.

So as not to make modifications to the main suspension assembly block, I wanted to make a V2 blade stop that would still mount to the original holes on the block, but allow for engaging the blade stop screw much closer to the tip of the blade. Below is what I've come up with:

Here is a wider shot of one within the overall SiFi assembly. The older V1 design can be seen towards the top of the image.

If there are no objections to this design, I will get some made up by MillItNow shortly.

Week 2 :-

I have mapped the beam profile of the laser to be used it the reference bench. The width in the x-axis was found to be 0.81mm at a distance of 92.53cm behind the collimator and the width in the y-axis was found to be 0.925mm at a distance of 89.23cm behind the collimator.

I used a razor blade to eclipse the beam and ploted the photodiode output vs razor depth to get the integratd intensity curve. From this I derived the beam spot size at that position. I repeated this step at different distances from the collimator and found the spot size at all these distances. I then plotted the beam spot size vs distance from collimator and extended the plot to find the beam waist. All fitting were done using Matlab. The plots are attached below. 

I have taken similar data for the remaining two lasers to be used in the test bench, but I am yet to analyse them. I'll do it this week.

Week 3 :-

The tables have been swapped. I set the two primary lasers on the new table and made them beat with each other. The beat note obtained was quite noisy. A snapshot the the oscillscope i attached. I tried to lock the eat note to a signal generator using a PLL by was unsuccessful. Maybe the beat note is to fast for the SG to latch on. I'll look into it further.

The reference bench mode matching is complete. The cavity requires a beam of waist 0.3482mm, located at the surface of the plane mirror, to resonate. I have obtained a lens solution for this (I had to predict the location of the cavity as the vacuum tank will be set up later.). I have used 4 lenses with focal lenghts 400mm, 200mm, 225mm and 250mm at distances 20cm, 85cm, 160cm and 224.26cm. The 400mm lens is required to squeeze the beam between the 2mm clear diameter of the EOM. The remaining 3 lenses produce a beam of waist 0.3478mm at a distance of 285cm, which is the predicted location of the plane mirror of the cavity (the mode matching screenshot is attached). I have completed the setup. Now I will map the resulting beam again to make sure it is the same as the theoretical predictions.  

The work crew came to the lab today and in a collaborative effort we swapped the table positions and put them on the new legs. The self-leveling legs were installed underneath the the now mostly empty table that will support the cryostat. We will start installing the tubing but hold off on floating the table until we find a good way to connect it to a pressured line/gas cylinder.

This would be a good time to do some cleaning in the lab, as removing the cabling under the tables revealed a plethora of dust bunnies. To not stir dust up unnecessarily, we will start wet-mopping tomorrow with just water and clean compromised surfaces. We can later follow it up with some lab/clean-room approved detergent.

I received a price estimate for the mirrors from Laseroptik today, I added it to the Wiki page. The quote is supposed to follow shortly.

The drawings on there are now current, complete with LIGO document number.

I have received the first spacer quotes, they can be found on the new cavities wiki page, which I expanded with a lot of cavity design blabla.

https://nodus.ligo.caltech.edu:30889/CryoWiki/doku.php?id=documents:2nd_gen_cavities

Some details on the laser:

I connected the West laser's designated TEC controller and turned on the feedback. The thermistor stabilized at the target resistance of ~9.9kOhm.

I turned on one of the new current drivers (S/N: S1500268), but before connecting it to one of the lasers I wanted to check its diagnostic outputs on a scope. I didn't see anyting out of the ordinary, however when I was reaching for the (BNC) cable I had been using, the moment I tried to remove it (from the total laser current monitor channel) I heard something pop inside and the indicator LED for the -15V supply voltage went dark. Sure enough I found that the driver was shorting the negative input of its power supply. I disconnected it. I have no idea what could have caused this.

A little concerned I repeated the same with the second one I had installed in the rack (SN: S1600247), but fortunately it seemed fine. I even manhandled (within reason) the total current monitor connector with a cable attached, but nothing happened. The blown unit had been tested along with all the other ones, so I was a little hesitant to connect the working one to the laser, but since I found nothing wrong I went for it. It worked fine, and the laser (West RIO Planex) turned on. The coarse adjust of the current regulates the laser output power as expected, and all seems in order. I turned the system off and disconnected everything, and will do some proper testing next week.

Subham and I then used the same TEC controller and current driver with the covega fiber laser that will be used in the RefCav setup, to see if it still works. It also turned on, and seemed a little brighter than the west laser at the same current setting. So far so good, we have two spare current drivers, but it would be good to find out if the busted one was a fluke or what else could have happened. I will check with Rich or Todd (who tested them) next week.

Hi!

I am Subham Vidyant and I'll be working as a LIGO SURF student in the Cryo Lab under Johannes this summer.

I started on Monday (May 16) and spent this week familiarizing myself with the lab and the components I'll be using, like network analysers and oscilloscopes. We preped the optical tables for the swap, labeled all the wires, and cleaned up the lab. We fired up the old laser i will be using yesterday and it worked! Phew!! I have worked out a lens solution for the cavity, but can't tell if it will work or not.

Can't do anything much now because the optical tables have not been swapped yet. Hope it gets done soon so I can get started on some REAL work!

I'm curious to see the after bake pump down plot.

Dmass has confirmed that ATF coated these mirrors, so the ATF data we have is for them. The specs are:

HR: 165 ppm

AR: 531 ppm

Likely specs for the substrates and coatings:

https://github.com/CaltechExperimentalGravity/CryoLab/blob/master/docs/datasheets/Coastline_silicon_mirrors/test-data.pdf

https://github.com/CaltechExperimentalGravity/CryoLab/blob/master/docs/datasheets/ATF_coatings/HR1550nm_150ppm_and_AR.pdf

I was interested in turning on the heater to avoid condensation on the upper volume of the turbo pump. This is standard good way to keep your station clean while baking.

The station DCU offers option: heater on/off but in reality it does not have a heater.

With regards to the heating jacket (actually a “band”), please note the following:

1. Heating can only be used on SS body turbos – these are the CF flanged turbos only   ISO flanged turbos have Al bodies!
2. Water cooling is required when using heater
3. Heater operates in 80-90C range
4. Max temp for CF flanged turbo housing is 120C, max 90C rotor temp

There is a cache of 1" silicon optics in the blue Lista cabinet in 050. They are from Coastline Optics and I believe they are from some version of the CryoCav experiment.

There are 3 different labels. They are listed below, followed by the serial numbers of found optics bearing each.

"Coastline Optics, Inc. --- 1 Meter/Plano Silicon Substrate --- S/N: 1.0-SI-1.0M-XXXX": 0004, 0019, 0016, 0021, 0027, 0029, 0033, 0034, 0036*, 0040*

"1 Meter Concave / Plano --- Silicon Substrate w/Annulus --- S/N: XXXX": 1, 2, 3, 4, 5, 6, 7, 8*, 9*

"Silicon Substrate, Coated --- 1 Meter CC w/Annulus - Plano --- S/N: XXXX": 0025, 0028, 0032, 0035*, 0039*,

(Key: GREEN - Apparently coated on both sides; RED - Optic missing; Starred* - Found in a smaller inner bag)

Notes:

• The coated optics with the third "Annulus" label have a larger uncoated annular ring on the less beveled side relative to the coated optics with the first label.

I need to figure out what coatings are on these mirrors

The heat is on at 8:35am Tuesday 22,282 hrs. The temp controller is ramping up to 100 C in 4 hours, it will hold for 48 hrs and turn heat off.

Turbo motor temp 48 C   

Cryostat control point K1 termocouple ~55C,  K2 (Fluke 52) monitor 49C at 10:15am

13:10      K1  87.1 C,  K2  70.3 C,  Turbo  49 C

14:28      K1  91.5 C,  K2  74.2 C,  Turbo  49 C

18:06      K1  98.1 C.  K2  79.3 C.  Turbo  49 C     

8:00 W    K1   93 C,    K2   77 C,     Turbo  49 C,     more Al foil insulation added......Omega controller "CNi8" stoked somehow .....may be I bumped it as foiling.......it was restarted

9:30        K1  90.7 C.  K2  70.6 C,  Turbo  49 C

10:27      K1  99.8 C,  K2 78 C,       Turbo  49 C

13:13      K1 100.8 C, K2 80.1 C,    Turbo 49 C     Temperature underneath G11 plate is 37.3 C, measured with hand-held thermocouple from below through the hole in the table, 42.4 C from below the G11 overhang on the pump side

14:38     K1  100.8 C,  K2 77.5 C,   Turbo 49 C,  Turbo CF  28 C,  room temp 19 C It is time to think of an RGA scan.    That's the way to know how clean is the green monster is right after the bake.

17:36     K1  98.3 C    K2  79.6 C

7:55 Thr   K1  98 C.   K2 79 C,   Turbo 49 C,  CF 28 C

8:10     K1 100.0 C,   K2 80.7,    Turbo pumping  22,330 hrs, 1.56A 23.8V, actual rotation 90,029 RPM

8:25     K1 94.5 C,      K2 72.1 C     I think it is cooling down  now. 48 hrs of heating completed. This would be a good time to reconnect "Fischer plug" to see how warm it is in bottom inside and start plotting it.

17:00    K1 97.6 C    K2  77.7 C    Turbo 48 C    Not so much cooled down. Did the heating cycle restart?

7:50F   K1 91.4 C,   K2  69.3 C      Turbo 49,  room temp 20C   Heaters are disconnected at 71 hrs

8:08     K1  84.2 C,  K2  63.9 C      cross connected K1 into Fluke monitor, it red 4 C lower

                                                      Fisher plug connected to see inner temp

8:15   K1 82.1 C,   K2  62.3 C

10:33   K1 52.5 C,  K2  42.2 C

Johann, Chris and Steve,

2 sets of 4 tapes ( 0.5"x48" ) are ready for bake in series connectons ~95 ohm each, covered with multy layers of Al foil insulation.

Fischer wired- connector "DEEGAOSS-130" and 4 Sapphire windows left in place. Insulation plate G11 was placed under the chamber.

The chamber is being pumped again. Baking starts tomorrow morning.

I jus realized that all plumbing on the top of the cryostat has to be part of the bake so there will be no condensation.

............

Good morning Steve,

Well let,s see.   IMG_0808-----is a 1.0" Aluminum Vacuum Safety Relief valve ass'y. has 2-each   O-Rings on brass plug.
                                                       1.) #2-210  (Viton) 400 deg. F
                                                       2) #2-116 (Viton) 400 deg. F

                          IMG_0819-----Evacuation Valve ---#BAP-25-O Bellows sealed---  has 2-each O-rings ,  (Viton)
                                                                                www.keyhigh.com     or    keysupport@keyhigh.com

                         The Epoxy on NPT thread for BAP-25 valve   Armstrong 12   ---------------- 100 deg. C  bake out is fine.

-----Original Message----- From: Steve Vass
Sent: Friday, May 6, 2016 7:12 PM
To: Precision Cryogenic Systems, Inc.
Subject: Re: Precision Cryogenics - Contact

Hi Dick,

I'm still finding things that I have not checked.

Thanks, Steve

ps: did you guys pump this unit down after final assembly? what pressure
did you get?

We are almost ready for 100 C,  48 hrs vacuum bake starting ~ Monday.

I'd like to replace gauges-feedthrough with blanks, G11 insulation goes under chamber and Al foil wrap insulation on side and top.

From PCS:
"Good morning Steve,

1. Top large O-ring #2-385.
2. Bottom large O-ring #2-386.
3. Small O-ring #2-237   (Housekeeping port).
4. Small O-ring #2-131   (Window ports 4 )
5. Material--------NITRILE    (Buna-N)
6. Max. temperature------250 deg   F .=121C !!

Regards,
Dick"

Nice! There's one more thing that I forgot earlier. Steve ordered an insulating plate to put under the cryostat while baking, to avoid dumping heat into the table. We either wait for its arrival or can scramble for something similar.

The spare O-ring we have is in fact Viton (photo attached). Hard to imagine that the installed O-rings would be anything different. Also we have a bag full of KF25 blanks in the lab. So we should be good to go for baking.

As for optical tables -- the table we've been using is at least 30 years old and has the awkward central through-hole. The one we don't use is a relatively modern Newport RS2000 with some passive tuned damping. Both are the same size (smaller than we'd like).

[Steve, Johannes]

This morning we wrapped the cryostat with heating tape. We used 6 pieces of tape (1x top, 3x upper, 2x lower) after confirming that they all are intact with a resistance of ~24 Ohm. We used Kapton tape in strategic places to have appropriate spacing between the strands, producing the thing of beaty you can see in the picture. Two things are remaining to be looked at before we can start baking: 1. The plan is to bake at 100 C. Depending on the elastomer the O-rings may not be able to sustain this, and there don't seem to be any records indicating their material. Chris mentioned there MAY be a spare somewhere that could help us identify. If not, we could just get Viton replacement rings that we know can endure 100 C. 2. We need some more blanks to take the place of the pressure gauges while we're bake-pumping. Once these two things are sorted out we're set to wrap it up und start baking.

In other news, the new table legs from Newport have arrived. While we're swapping them we COULD as well switch the optical tables (the central one has better internal structure for damping according to Chris), and we need to redo the optical setup to some degree anyways for the new cavities. If we want to switch this would be the time to do it.

[Steve, Johannes]

We concluded the heat gun test, and weren't able to bring the paint to rub off (wiping with Q-tip) or vaporize to the point where it would cause a noticable smell. It seems this far that the heat tolerance of the paint far exceeds 80C.

We attached two thermocouples to the upper plate of the tank. One was to serve as the primary readout point, while the other was to check the heat flow through the plate. For this we one-at-a-time undid two screws of the seal with the crystat body and scraped the paint off in a small area around it with a razor blade to enhance the thermal contact between thermocouple and the plate. We clamped the thermo-couples to the plate using the sealing screws. We used the heat gun on its high setting and measured the air stream to have a temperature of about 250C with a third temperature sensor. We "insulated" the sensitive parts on the top of the tank (e.g. pressure gauges) by generously shielding them with multiple layers of Aluminum foil, and directed the air stream from the gun away from them.

We first tried to heat close to the primary sensor, but it responded too strongly just to the heat directly from the gun and didn't give a reliable reading of the local plate temperature, such that we eventually shielded it with several layers of foil as well, to keep the direct exposure to the heat stream away. We spent about 30 minutes in total exposing a small area of the top plate to the heat gun. The insulated sensor T1 went up to ~70C, while the secondary sensor T2 showed a max of 57C. The third, hand-held thermo-couple we brought in direct contact with the heated area. We both used it after taking the heat off (with readings up to 95C) and produced peak readings of up to 130C while simultaneously applying heat. No noticable smell was produced, and no rubbing effort was enough to remove visible amounts of paint.

This strongly suggests that the paint job can actually take temperatures exceeding 80C by quite a bit. To be completely sure, Steve scraped off some more paint, and collected it in a glass beaker to be heated on a hot plate in a more controlled way. The scrapped off paint was baked in a glass beaker at starting at 70 to 115 C and it is still baking. By all means it passed the test of 120 C with a mild smell.

I hope now that the o-rings are Viton.

The braun colored epoxy Armstrong A12 (mixed as 1:1) still has to be tested. According to PCS it can take 100 C

Continuing with the cleanup, I got rid of the old serial-USB adapters that are used to acquire slow data for Epics. They have no unique ID in their firmware, so it's impossible to reliably distinguish which one is which on the USB bus. Works fine when you have only one plugged in, but not so great in our case. They've been replaced with FTDI based adapters, which do supply unique IDs. Now a dedicated /dev/ttyUSB name is assigned to each device by udev as it is connected. This solves 1/2 of the problems from log 1342.

I also cleaned out a large tangled knot of cables between the analog and digital racks. These were overly long, badly made, and entirely unused. Based on labeling, they may have been a relic of some past attempt to involve the digital system in the cavity locking.

This leaves the racks in decent shape, except for a tumorous mass of clip-doodles that we still have in place substituting for an AA chassis.

[Johannes, Steve]

I'm writing this report because Steve's computer has given up on him, basically he found all this out and shared it with me. We are currently planning on getting the baking work started this Monday, but there are a few concerns that we wanted to share for the record and hopefully get a go-ahead for our (his) plan.

Steve contacted the cryostat manufacturer, who was actually very helpful, and through them got in touch with the paint supplier. While they have never really tested the high-temperature limit for their paint 80C was mentioned as a tentative upper limit.[Sherwin Williams Polane-T Polyurethane Enamel ] This is a problem because it would pretty much ruin efforts to bake with heater tape, which is likely to get hotter and directly attaches to the paint. We will do a test with heat gun and temperature sensor on Monday and test this. Given an 80C limit we will have to disassemble the cryostat and take it over to the 40m lab to bake it in a chamber at ATM. 80C will still be the highest temperature, but it will be evenly applied.

Bake prep: remove sapphire windows, remove brass valve and Al foil, disassemble as much as possible,  remove screws from tapped holes and check their cleaniness. Preclean if needed.

There is a hierarchy in possible bake temperatures: 80 C (again, tentative) for the paint job < 100 C for the epoxy used for the G11 < 120 C because the

Aluminum chamber was not stress relieved after welding. It's max baking temp 120 C

Besides having the lowest max temperature, the paint remains a more severe limitation because it would be very close to the heater tape.

So I repeat: Our plan is to check on Monday the max paint temperature with a heat gun. We anticipate that we will have to disassmble the tank and take it over to the 40m for extended baking. If the paint proves to be significantly more heat resistant, we can reconsider if local heat-tape baking without disassembling could suffice. The next upper limitation for the temperature would be the 100C for the epoxy.

Does anyone have any concerns about the suggested procedure?

I sank some time (already way too much) into modeling the photothermal and thermo-optic noise levels in the cavities according to Stephan's 2015 paper (PRD 91, 023010 (2015)) because I'm stigmatized by getting caught by unmodeled noise. While that's still going on, it merely corrects the noise levels above certain characteristic frequencies downwards. Therefore I only used the thin coating approximation from PRD 78, 102003 (2008) for a comparison of intra-cavity noise sources. For the most part I used the coating parameters from the papers and what was on the LIGO Wiki under https://wiki.ligo.org/viewauth/OPT/CoatingProperties . For the cavity gemoetry I used 5 cm ROC mirrors and a 15mm long spacer that is 1.5 inches thick. The seismic noise data comes from https://nodus.ligo.caltech.edu:8081/Cryo_Lab/1299 with an assumed coupling coefficient of 50kHz/g (motivated by a 100 micron tolerance for the graph in https://nodus.ligo.caltech.edu:8081/Cryo_Lab/1388). I assumed a finesse of 100,000, which require a coating thickness of about 8.5 microns, 1mW of incident power, a mode-defect of 0.2, and a modulation index of 0.2. This is all for room temperature.

If these numbers are to believed (I still need to hear back about machinability for the spacer, and also want to look into slight variations of this mounting method that would allow for small adjustments) it would mean that we do not absolutely need to rely too much on common-mode cancellation between the two cavities, which brings an in-cryostat reference cavity back to the table. One thing to keep in mind when going that route is that we only want to pay for one coating run, and a high-finesse longer cavity has somewhat narrow linewidth (10kHz for L=15cm). I also pasted the Brownian noise into the most recent beat noise spectrum that I could find ( https://nodus.ligo.caltech.edu:8081/Cryo_Lab/1099 ). While it is a big step upwards, as you can see we don't even breach the achieved laser noise levels (are there more current plots anywhere?), so I don't think we can compromise towards a pair of slightly longer cavities, we need at least one very short one.

The only company that has gotten back to me saying they can do the coatings is Laseroptik, and their polishers also have agreed. In order to get a quote from them I need to finalize the radius of curvatures. I think I will order the mirrors first, and spend some more time on the spacer design. Since the 5cm did not raise concerns (at least none that were shared with me) I think I want to log at least that one in. This might be too short to assemble a longer cavity though. But with a longer roc we can either build a second short cavity with larger beams, giving different Brownian noise levels with potential common-mode rejection (which I'm quite fond of) or a longer reference cavity. So this still leaves some flexibility for which route to take afterwards, and we can also add some flat substrates to the order. For the second roc I'm thinking something in the range 20-50cm, I will pick a value based on the g-factor that gives for a second short cavity, and also find out what the max batch size is for the coating run.

• SSH security has been tightened somewhat. When you log in to cymac1.caltech.edu, it now takes you to the workstation machine (gaston) instead of the cymac. And it is configured in the /etc/hosts.deny file to only allow connections from caltech.edu addresses. However, you can still log in from off campus by going through nodus or another campus computer.
• The folder on cymac1 where the RCG models are kept (/opt/rtcds) is now being shared over NFS to the other computers.
• All machines are backed up nightly by rsnapshot. cymac1 and cryoaux have spare disks mounted as /backup partitions, and gaston is backed up by ssh to cryoaux.
• In the course of setting up backups on cymac1, I noticed that the raw minute trend files were extremely slow to access -- like 100x slower than normal frame files. Turns out because they are slowly growing files, they get fragmented over time into thousands of chunks. I defragmented by copying them over to another partition and back. Then I used fallocate to reserve 20MB of contiguous space on the disk for each file's future growth, which should be good for at least a couple more years. Now the backup goes at a normal speed, and as a side effect minute trends are a lot quicker to plot in dataviewer. (Maybe this effect also has some bearing on the framebuilder issues at the 40m?)

While waiting for data recovery on cymac1, I took the opportunity to work on the computer formerly known as "fb".

"fb" was a lousy name since this machine hasn't acted as a framebuilder in many years. Instead it runs slow controls and miscellaneous scripts. So I dubbed it "cryoaux".

It was running an ancient and un-upgradable version of CentOS. After making a backup, I gave it a fresh install of Debian 8.

I then installed EPICS and ported over the code for reading out the particle counter, vacuum gauges, and CTC100 temperature controller. These are found under /home/controls/services. It's all set up to be started automatically by systemd when the machine boots.

There was one weird problem with running certain EPICS IOCs under systemd, but I found a hack to work around it. The issue was that some IOCs (including our CTC100 IOC) provide an interactive command line. And when reading commands, they interpret end-of-file (Ctrl-D) as a request to shut down the IOC. Normally when we launch them in the background, the input is disconnected so that nothing can be read. But systemd actually connects the input of the programs it starts to /dev/null -- which returns end-of-file when read. The result was that the IOC would start up, read end-of-file, and immediately shut down again without any error or explanation. To avoid this, the IOC is launched from a bash shell using the "coproc" keyword, causing it to start in the background with its input disconnected.

Our cymac suffered a double hard disk failure this week. Two Seagate 1 TB disks were part of a RAID mirror holding the /frames partition. One was completely dead, the other was limping along with errors.

Larry gave me a spare drive and I was able to recover all but about 20 kB of data onto it using ddrescue.

The system is back up and running, with Bruce Allen's smartmontools keeping an eye out for future signs of failure.

Upon request, this is the optics layout I propose to include the ref cavity in the setup, at least at the early stage. We pick off / use a flip mount in one of the cavity paths (can also include in the other one) and have a very similar setup on the center table.

I was thinking we can use the existing electronics depending on which laser we are controlling, with an independently generated error signal.

At todays meeting with Rich and Koji we were wondering what would explain my observation --- enjoy my poor cartoon explaining the physics behind it:

The solid lines represent the height of the (exaggeratedly far off from the mid-plane) contacts. In both cases the sagging causes a tilt of the mirrors. Essentially this: If we choose it high, the sagging will compress the lower spacer regions, shortening the cavity. If we on the other hand choose it low, the top part will sort of spill over, elongating the cavity. We get the best cancellation somewhere inbetween, when the length change combined with the tilt causes a net rotation of the mirrors around the beam spot.

This is actually quite similar to varying the angle of the mounting points from before, the reasoning kind of works out there as well. The good thing about this is that we can get arbitrarily close to the mid-plane, and even go above it (although that' shouldn't be necessary) unlike with the side-contacts.

Now the question is what kind of tolerances we need to meet in order to exploit this effect.

More like weekly progress. Based on Evans past efforts I built a Comsol model for the new silicon cavities over the last two weeks with plenty of geometric parameters for their construction, such that they will represent the final product closely. I've also been experimenting with different ways of mounting them. In particular, there are these two methods that are supposed to mitigate the coupling of vertical acceleration noise:

Mounting solution 1: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6231007

This approach from the people at the PTB was used for the current silicon cavities. I went to high school in Braunschweig and visited the place many years ago. I was initially looking into this method, but I was having trouble with the numerical stability of my results, and I couldn't quite convince myself that I was modeling the mounting points in a physically correct way.

Mounting solution 2: http://journals.aps.org/pra/pdf/10.1103/PhysRevA.75.01180

We were talking a little while ago about concerns about the thermal contact, and in this paper the seismic coupling was minimized a little differently, which gave me the idea to look into this. While they use only point contacts I modeled the entire horizontal section as the contact plane. After solving some numerical problems due to meshing with help from Tara, I have now reasonable confidence in the numerical accuracy of my simulations. This is a model of the cavity spacer, reduced to a quarter for better precision using symmetry conditions:

The construction is entirely parameterized and can easily be adapted to longer and thicker spacers. In this case the spacer is 15 mm long and has a radius of 20 mm. I defined the shelf as cutting along the entire spacer, going in with a width of 3 mm. I also added a height offset parameter for the exact shelf position, and varied it over a small range. I calculated the coupling coefficient for seismic noise along the vertical direction for each configuration by assessing the displacement of a test point in the center of the mirror with the assembly subject only to gravity, and also with an additional 10^-3 m/s^2 in z-direction. I converted the difference in displacement to a shift of the resonance frequency according to 2*(dX_acc-dX_equ))*1e3/L*lambda/c and varied the height offset from -.1 to -3 mm. The graph shows the result of the simulation in kHz/g. While there is some variation in the exact value of the coefficient, there is a clear trend and no funny behavior which gives me confidence in the simulation. For the given configuration there is a zero-crossing at about -1.4mm. I ran these simulations on my laptop, which took about 30 minutes. The next step would be to use a better machine in the 40m network using finer meshing. This looks very promising, and I will get in touch with polishers about what kind of tolerances they can give for shelf height. I don't think it's ridiculous to ask for 100 microns or even 10s of microns from a precision polisher.

Since I had the model ready to go I also ran an eigenmode analysis. I found that the lowest frequency drum-like mode is at 115 kHz:

When I get the chance I will upload the model to the svn, but for the time being I'm still making a lot of changes so I'm holding off on that.