I ordered the PEEK connector and it should be in by Wed. Worst case, we can set up the ESD and weld the sample without having the connector, and then hook it up once it's in.
I went for the 4-terminal one (instead of the 8-terminal one linked below), since it was half the price and we shouldn't need more than 2 terminals anyway. I chose 4 instead of 2 in case we wanted to expand later.
The ESDs have shipped, though for some reason I chose UPS Ground and so they are scheduled to get here on Wednesday. Oops.
I spoke with Margot today and she thinks she can provide us with as much in-vac solder as we're likely to need for them. She was going to stop by the SUS lab to check out the setup this afternoon, but I think we were both too busy. I'll see if she can come by on Monday.
I also spoke with Rich about the in-vac HV connectors. The idea is that we want to be able to quickly attach and detach the ESD from the HV supply wires so that we can get the drive and the sample set up nicely before lowering it back into the vacuum chamber. For illustration, I think Giordon should make a little diagram of what the in-vac setup looks like (i.e., suspended from the stainless steel disk at the top, with metal-bar-and-teflon-block frame hanging downward. The ESD will be screwed into one of the teflon blocks.
Rich pointed me to this relatively cheap PEEK in-vac connector. We can attach this to the chamber and have the +HV and ground connected to it semi-permanently. Then, we can have long cables soldered to the ESD on one end that we screw into this connector each time we re-lower the suspension into the chamber. Voilà.
I'll order this ASAP and ship it quickly.
I received the 4-terminal PEEK in-vac connector for the high voltage today. It looks like it will work just fine.
I'm glad I didn't spring for the next-day delivery!
I received the ESDs just now. They look pretty good, although the silver finish seems a bit worn in places.
I will drill the beam holes today, and with any luck get one mounted into our setup with Giordon.
This eLog will serve as a compendium of numbers I've found for helping with the calculation.
We're dealing with birefringence here, which means that our light splits into two rays, each of which experiences an index of refraction. The overall outcome we expect is that the net rotation involved is linearly dependent on the amount of birefringence / stress induced in the mirror.
We know from the stress-optic law (Copied the image from wikipedia)
Note: C is measured in Brewsters, retardation is defined as a difference in indices of refractions divided by the thickness. Cite (http://goo.gl/X6sae).
From conversations with Zach, our calculations comes in two parts.
Here are numbers we've found. I've attached PDFs where relevant and otherwise linked to websites.
I wrote a good paragraph or so and for some reason, page auto-refreshed. I don't feel like typing it up again... so I'll summarize.
Got ESD in today, Zach bought both down to lab. He wants to use Silver Epoxy to make a connection to the HV wire. Neither of us are sure about how effective this is. We'll probably use it, and if something's wrong with the connection - we'll blame the epoxy (maybe).
Alastair came down and helped us dismantle the vacuum chamber AND take out the suspension mounts. Currently, the suspension mount is on a box - making it top heavy and liable to fall if bumped into. BE CAREFUL.
We rotated and shifted the chamber so that the beam shines a lot better and is lower. There's already a suspension set up so we'll use what already exists since that will make our lives so much easier in the long run. We also made the beam lower in the chamber to give us as much thread length possible (which helps with absorption of noise).
We should have soldering done tomorrow, and everything ready on Friday to pump down and make a measurement perhaps. I'll need to clarify with Zach.
Just to clarify/add to what Giordon said:
I went to the supply shop in CES and bought some more Teflon that we'll need for supporting the ESD within the vacuum chamber. The leftmost one is one that Alastair had from before, and the other 4 are the ones I got today. They are a tad wider, but that's not a problem.
Here is a picture of the suspension setup as it exists today (you can see the intermediate mass and then some fiber hanging below it if you look real close), and then a diagram showing a bird's eye view of how the ESD will be supported (assuming the bird is actually a bat hanging upside down from the steel disc at the top):
Things to do before mounting the sample, etc.:
The teflon bars were drilled out at the far ends to fit 1/4-20 screws. In two of them (which are attached to the ESD support bar) - there's a 1/4-20 hole drilled lengthwise in the middle of the bar to allow for fine adjustments.
These were all smoothed down and cleaned with acetone and methanol. They're currently wrapped in UHV foil in the lab right now.
The laser is leveled out to be 44.5 cm from the top of the suspension. The ESD will be placed so that the laser passes through the vertical middle of it.
Each teflon bar has a length (vertical) of roughly 1.5 centimeters. The ESD itself is 7.5 cm long, so half of that is 3.75cm. This means the top of the ESD needs to be at 44.5 - 3.75 = 40.75cm. The teflon bars at 3cm more, so they were placed such that the bottom of the bottom teflon bar is at 43.75cm. Which places the top of the ESD at 40.75cm. Which places the vertical center of the ESD at roughly 44.5 cm (where the laser is).
All parts were cleaned, the foil was placed back on the suspension mount. And we're done.
Here are some notes and pictures of what was done yesterday for the in-vac setup:
Giordon has already reported on the ESD support. To make this, I drilled the Teflon bars I had made up in the appropriate way. The two side bars have oblong holes near the center (but shifted aft a bit), so that the crossbar holding the ESD can be adjusted finely in depth. The crossbar has several holes across the middle (perpendicular to the holes that mount it to the side bars), spaced so that the ESD can be moved left or right a bit as necessary to combat cancellation from symmetry. See his entry for a good picture.
While he worked on actually attaching it to the suspension, I was working on the HV connection. I decided the best way to do it was to mount the PEEK connector to a Teflon spacer, which is mounted to the base of the chamber via 1/4-20 vented and silvered cap screw. I countersunk the hole in the Teflon for the PEEK-mounting 8-32 screw so that I could fit the nut underneath. Here's a shot:
On one side of the connector, we have the HV supply provided by cable from the SHV connection on the side of the chamber, and the ground is provided by a cable that is screwed directly into the chamber base via 1/4-20 vented and silvered cap screw. Of course, I verified that the entire tank is indeed grounded with respect to the HV amp. Here is the whole assembly in the chamber:
I'm not crazy about the angles of the cables coming out of the connector, so I may choose to shorten them.
I also connected the other cables (which will go into the PEEK connector from the other side) to the real ESD---which I also drilled beam holes into---with silver epoxy. The stuff is a bit of a pain in the ass, and I had to apply a good bit of it just to make sure the wire would be (mostly) submerged. I left it drying under foil.
With this, we should be pretty much ready to weld in the sample and pump everything down on Monday.
We've finished up the HV Cables. The time drain here was pretty much involved in trying to strip down the outer insulation on these wires - we do not have any device that does this consistently other than pure luck and experience. So I spent about 30 minutes test-stripping a bunch of scrap wires until I was confident in my newly-developed abilities to strip wires without breaking it apart.
On the past week (last Thursday area), Zach had cut together appropriate lengths for the wire and attached it with the connector. He mentioned that these wires were slightly long and "didn't like the amount they bent" and so had me cut them to length and strip them. They are currently situated in the vacuum chamber (pictures below) and look good. We have also attached the soldered-ESD onto our suspension mount and double-checked the heights. We also added enough slack so that we can connect the ESD in the vacuum chamber before lowering the suspension mount completely and allowing us to adjust the horizontal-deflection of the laser beam.
I am working on writing up an experimental plan and running some COMSOL simulations on a substrate (3" by 1/8") to get a rough idea of the general level of eigenfrequencies we expect to have in our system. There will be some simulated variations modeled in the environment such as the ambient temperature and absolute dimensions of the substrate to see the relative propagation in variation in the eigenfrequencies reported.
[Alastair, Zach, Giordon]
I was inhibited by a lack of a computer for most of this week. Long story.
On Wednesday, we met Alastair and started work on the welding. First, we stretched a piece of silica so that we have a thread to replace the lost/broken one. This seems to have gone off without a hitch. We also attached this thread back on to the suspension glass - so all that was left was to weld the substrate we're studying onto our suspension mount at the right height. I'm working on the experimental set-up.
[Alastair, Zach, Giordon]
I don't know why I have "eLog"-ophobia. Maybe this will go away soon.
Friday - we achieved awesomeness. Alastair came in earlier to finish up the welding work for us. We placed the ESD back on the suspension mount, and the substrate seems to be perfectly situated
Over the course of 20-30 minutes, we slowly moved our double-pendulum and suspension mount into the vacuum chamber (probably the scariest moment of our lives). Once we lowered it into the vacuum chamber, Zach went ahead and connected the HV cables.
A big issue we noted was the presence of static electricity (static charge) built up on the substrate. This causes it to rotate into the ESD and artificially "stick". We've tried discharging it via various methods (read: grounding wires). By dumb luck, we seem to have solved this and then placed the vacuum chamber hood on top and started pumping it down. We discussed all the specificities of the vacuums in a healthy discourse with Alastair and agreed on me coming in to lab Saturday morning to turn on the UHV.
Saturday Morning: UHV turned on.
Saturday afternoon: We also have sample calculations completed for the substrate and this is in our PDF attached here as well.
Our measured dimensions using a micrometer caliper (with Zach giving me the numbers) for diameter and thickness are:
[Giordon, Alastair, Zach]
Alastair dropped by today and showed us how to use the HV amplifier.
When we got to the lab, we turned on the ion gauge and the pressure reading in the tank was ~10-7 Torr, which Alastair reckoned was pretty good for one day of pumping with the large tank-mounted turbo (Giordon thought he turned this on over the weekend but it looks like a cable was unplugged). Alastair instructed us on how to rig things up and then safely turn things on and off. We tested the ESD with a maximum DC offset of 3 kV and AC amplitude of 1.5 kV @ 1 kHz, and nothing seemed to spark or explode.
Giordon and I then did some initial sweep testing by driving the amp with an Agilent function generator. Using the frequency estimates he posted before, we set up some narrow-band, slow sweeps across some of the modes and then monitored the spectrum of the PD difference signal on a spectrum analyzer in FFT mode. Alastair recommended doing it this way instead of taking a swept sine with the analyzer alone in order to better distinguish between a real signal and a spurious EM coupling.
All in all, we weren't that successful. We did see some cases of what appeared to be modes, roughly where his COMSOL model predicted, but they each had their own problems. The differences in measured vs. predicted eigenfrequencies were all at the high end or slightly beyond the bounds that he put on them by varying sample dimensions. It could be that some other material properties are off.
Here is a list of a few modes we sought to measure (if Giordon sees this perhaps he can upload his fancy animations so we can see what the modes physically look like):
Lots more to come on Friday.
Giordon did see eLog post. Here are fancy gifs. Click an image to see it in a new tab/window for animation to happen. I've attached a zip file that contains all gifs (for properly downloading). Images shown below are in order of increasing eigenfrequency (from the first "non"-translational mode [re: no fixed point]) to the 9th mode as referenced based on the bolded line in the PDF linked to the comment I've replied to.
To make a long story (which you can read on the gyro log) short:
I needed a pump for the gyro chamber, and we decided it was best to use the same one we've always used with it, which happened to be the one that was on the coating Q setup bell jar. I replaced it with the (new) one that is for the cryo setup. Eventually, Frank and Dmass will need it back, at which point we'll re-swap.
I vented the chamber per Alastair's instructions, allowing the large flange-mounted turbo to spin down by adding trace amounts of air before shutting the external pump down and equalizing to atmosphere. This went without incident.
I swapped the pumps out and then went to the ATF for a while to set the gyro pumping up. Around 30 mins later, Dmass informed me that the pump had stalled and was spinning down. I re-checked the valves and reengaged the pump. About a half hour later, the same thing happened again. Frank troubleshooted that it was the auto-shutoff feature of the pump that was killing it (it is designed to self-kill if it hasn't reached 1/2 the target speed within half an hour). The problem was that the smaller hose on this pump has a hard time evacuating the large bell jar within this time. We found a way to increase the shut-off interval by enough that it spun up in time.
By the time I was leaving, the coarse vacuum gauge was reading a few mTorr. I checked with the large turbopump's manual that it was fine to engage it at these pressures, which it was. So, I engaged the large turbo and have left it running. So, we should expect to be back at ~10-7 Torr by tomorrow afternoon.
NB: The vacuum system was back at 10-7 Torr when we started work on Friday.
We began using the lock in to monitor demodulated time series while driving modes and then allowing them to ring down. We used two different driving methods:
Both used a DC bias of 3 kV. For the readout, we set the internal SR830 lock in reference signal to ~10-100 Hz offset from the measured mode peak, and the time constant such that there was a clear signal at these beat frequencies without excess high-frequency fuzz. The lock in input was of course the differential PD signal, with the difference taken by an SR560. It was basically like an SRS/Agilent product demo down there.
The noise method was obviously just not strong enough somehow. Looking at the AC drive signal, we increased the power until there was a rough peak-to-peak level of ~2 kV, above which I did not dare go. Looking at the readout, there was basically no difference at all with the noise drive on or off.
The coherent drive seemed to do something, but as expected it both excited and damped the mode as an unpredictable function of the mode drift and the sine sweep. That said, when it excited the mode, there was a notable increase in the signal. When the drive was turned off while the output was at relatively high amplitude, the signal immediately went down to its unexcited level. The timescale was well under a second.
There are at least two possibilities:
It seems odd that we have a hard time ringing the mode up substantially, since the DC bias visibly excites the pendulum mode upon turn-on (so the electrostatic force is high enough to do something). It could be that the vibrational modes' admittance is just much lower than that of the pendulum, or that there is a great deal of cancellation from the particular symmetry. Another thing is that the sample might be getting subtly tapped by the ESD assembly every once in a while, and this destroys the coherence of the measurement.
We might consider slightly adjusting the arrangement (e.g. by translating it) to see if there is any improvement.
We tried some more today to actually get the 25-kHz mode to ring up significantly. We were unsuccessful.
We also toyed with the idea of using the differential output (band-pass filtered) to feed back to the ESD and actively damp the low-frequency pendular motion. This, too, was unsuccessful. I think the problem here is that we only have one actuation mode, whereas the sample is moving at low frequencies in three different ways (pendulum, twisting, and tilting). So, no matter what we set the DC bias and AC gain to, it's impossible to make the thing stable; it is either too weak to combat gravity or it just lists until it physically contacts the ESD. We actually saw some arcing from when the mirror shorts between two ESD electrodes. The HV amp current-limits itself, but we were able to see some slight browning of the ESD substrate in the location where this happened. All still seems to work fine .
We resolved to vent the chamber and translate the ESD sideways with respect to the sample. The idea is that since the modes we are looking at have some radial---and thus bilateral---symmetry, then applying a uniform force across the X and Y axes does no good. So, we moved the ESD over ~1-2 cm so that there is a force imbalance from side to side. Here is the best shot I could get of the new configuration:
An added benefit to the change is that one of the beam holes is now nearer to the center of the sample. This is good because, as you can see in Giordon's nice COMSOL plots, most of the modes we are examining have antinodes at the center.
We re-sealed the chamber and began pumpdown around 2pm. I came back down to the lab around 5 or 6pm and the pressure was ~1 mTorr, so I engaged the chamber-mount turbo. By the time I went home around midnight, the pressure was back down to 2 x 10-7 Torr.
I did a few minutes of playing around with it in the new configuration, to see if it had any real effect on the measurement. It didn't seem to. The SNR of the peak on the analyzer was roughly the same, and I measured roughly the same ~2-Hz linewidth. I was still unable to close a stable damping loop, and I also discovered that the apparent driving-up we have seen using a swept-sine drive is in fact just EM coupling . For that, I just turned off the HeNe and turned on the HV drive, and I demodulated signal at ~100 Hz that was just as strong as the one we see with the laser on.
In conclusion, we are still completely unable to drive the mode up to an acceptable level. Considering how reliably the sample will twist and contact the ESD on one end, it doesn't seem practical to try and move it any closer. Thinking cap time...
Short update. Rana mentioned that he would like a simulink model of the setup we're doing. So on Friday, I tried to do it, but it seemed to crash really unexpected on my Mac in the first 10 minutes. I sent an email to tech support, but had to wait through the weekend for a response. After a few days of debugging - it turns out that the COMSOL update for Matlab somehow screwed up with a specific file for Simulink - so tech support gave a fix, I applied it, and now everything appears to be working.
There should be a simulink model made up shortly and I'll throw it over to Zach and see what he thinks.
We initially found a mode that we've been looking at the past week at roughly 25kHz. This happens to be, not a resonance mode, but a mode from the UHV pumping down. Zach thinks he found a mode at 24.2 kHz and we've been trying to excite it. It appears that hitting the lab table affects it partially (we were exciting it literally; which is a reason for looking at this particular mode). We are still not able to excite it fully (we only see it when the drive is on, but not when it's off). We think the drive is not strong enough.
We pumped up the vacuum chamber and repositioned the ESD because the substrate was touching it / hitting it / being pulled in by the attractive-ness of it's square design. It's currently being pumped down for more measurements and identification this Friday.
In under 20 minutes record time, we've identified a problem with exciting the substrate. Our ESD is broken. We're pumping up the vacuum and plan to switch it out with another spare and then pump it down for more measurements!
For theoretical calculations and numbers to help us - we'll use COMSOL, export some information about the principal stresses of the system at a particular mode; as well as the displacement (in the z-direction, perpendicular to the face of the substrate). From this, we'll interpolate by linear approximation to determine the "local slope" or "local angle" of the surface at a specific r-point. We'll also figure out the average stress through the material [averaging over z] at a specific r-point. Combined, this should allow us to determine the stress-optic coefficient of a material given that we know the retardation involved from measuring the differential output of the photodiodes.
EDIT (ZK): Just to clarify, Giordon is talking about this idea Rana gave us to directly calibrate the signal instead of trying to estimate the stress-optic coupling a priori. Assuming linearity, we will calculate the expected deflection of the beam reflected from the front surface of the sample at the point of reflection for a given internal stress (averaged through the sample at that point). We'll then measure the deflection angle of the beam using a QPD (demodulated at the mode frequency) and the polarization rotation signal, and use the COMSOL results to directly obtain the coupling constant. This method will be limited by the uncertainty in our sample dimensions and material parameters, as well as in the location of the reflection on the sample.
Zach will probably spend time today with the silver epoxy to prep a new ESD to put in on Monday [vacuum is pumping up for the rest of today]. On Monday, we'll re-set for the next measurements and start pumping down the vacuum so we're ready for measurements on Wednesday. We also have issues with Simulink in trying to figure out exactly what blocks we need for modeling our experimental set-up.
As we were testing the newly adjusted setup yesterday, we noticed a faint blue glow within part of the ESD upon application of the DC bias voltage of ~2-3 kV. This are bad.
I removed the ESD from the chamber last night and inspected it. It appears to have some obvious damage in the are where we saw the glow:
It appears that we have exceeded the dielectric strength of the Rogers 4350 substrate. I (somewhat conservatively) calculated a maximum voltage of ~5 kV DC between the electrodes, and I think this could still be accurate. The problem is that we have inadvertently exceeded this on at least one occasion while I was trying to set up the feedback damping (i.e., I set the gain too high and the control + bias signal together exceeded 5 kV.
I am going to prep another ESD (which must be silver epoxied to the HV wires) ASAP so that we can get back down to business.
Alastair removed the tank on Wednesday afternoon. I went ahead and grabbed the ESD and it seemed fine - and screwed it completely into the vacuum chamber, plugged into the HV. For some reason, the joint that's being soldered keeps breaking off as the wire is super fragile - even if the cold solder is also coating the wire shell. I'm not totally sure why it's breaking off after a while, since I put a test pair of wires and soldered those together at the same time - and those didn't break when I tried pulling them apart. I'm pretty sure it has something to do with the thickness of the wire and my caution in not overdoing it (since we don't want to accidentally connect to the front side of the ESD). For now, we will do the pick-up measurements without the ESD in the vacuum chamber.
Alastair put the tank back on and the UHV is currently on. It'll be set for tomorrow morning when I go in to do measurements. Here's the tentative plan for the measurements:
Get a spectrum analyzer, and look at the noise before any changes, and after I make changes to figure out the optimal set up. Changes include:
- grounding and fixing ground loops
- wrapping each cable around each other
- move everything away from the HV supply (last resort)
- faraday cage things with the foil
Responses/Comments from Zach:
Measurements done - here are the general results. Note to Zach: I could not find the floppy-to-USB converter, so I was not able to grab the data I've saved from the spectrum analyzer (including pictures) to put in here. Do you know where it is?
I also prepped another ESD which should be ready to put in the chamber on Monday, and pumped down and ready for measurements on Tuesday (I pumped up the chamber after the tests were done).
Pick-up Measurement Notes
The following measurements were done: [....] contains data filename on floppy (for Zach's reference and my reference)
At this point, I made changes to the physical set-up attempting to reduce the pick-up noise
Conclusion: for the attempts to reduce pick-up noise [GHS4,5,6], I saw no actual change to the noise level in general. It definitely stayed above 3 microvolts/Sqrt(Hz). Only for [GHS7] did I notice that the noise level decreased to roughly 2.8 microvolts/Sqrt(Hz) - but I did not think it was significantly lower to believe that anything happened [although repeated measurements showed that it was consistently below the noise level of the previous tests...]
I don't think I had any success at all with the pick-up here. The only thing I did not do is move the lock-in amplifier away from the HV on the rack.
Here's the tentative plan for the measurements:
Get a spectrum analyzer, and look at the noise before any changes, and after I make changes to figure out the optimal set up. Changes include:
- grounding and fixing ground loops
- wrapping each cable around each other
- move everything away from the HV supply (last resort)
- faraday cage things with the foil
First, what do you mean by 24 dB offset? The offset I was talking about was a frequency offset. You set the lock-in frequency to 24.213 kHz---so, how far was this from the measured mode frequency? Come to think of it, I don't see any explicit mention of the lock-in output here (i.e., did the RMS level appear to go down when you made any of these changes?).
I don't think there is any reasonable measurement to be made with only one PD on. You have the SR560 taking the difference (PD2 - PD1) and then multiplying by a high gain, so it will rail unless PD1 ~ PD2. Check to make sure you don't see any red overload indicators on any of the instruments before trusting any data.
The floppy drive should be in the top drawer of the first workstation desk in the gyro lab. That's actually where it should be kept, but sometimes it stays with the spectrum analyzer.
24 dB offset (although this may not have been set, I forgot to note this)
Also, I noticed that if one photodiode is turned off when we make a measurement - the noise on the spectrum analyzer appears flat yet slanted [I think this is just useful for recognizing this issue later if it occurs]
I believe I mistyped with "offset" when I meant "notch filters that they let you put in to remove oscillator noise coupling".
No, now I realize what you meant. You had selected "24 dB/octave" as the slope of the low-pass filter after the mixer. This decides how sharply signals above the cutoff frequencies are attenuated. I think 24 is the max and is good.
There's an ESD in there now, so I'm going to run the pickup measurements with it. This is going to be a continuously updated eLog post to keep track of the measurements I've made and saved so I can post the data for it. I'll also have the data for the last set of measurements up tonight in a fancy plot.
Below are the data measurements taken with the laser output blocked before the photodiodes!
Overall impression is that nothing is really reducing that peak from the pick-up.
Like, WHERE, SCOOB?
Below are the data measurements taken with the laser output blocked before the photodiodes!
This is the general awesome noise plot you're looking for, zoinks! The baseline noise is hard to see on this plot since it's on the bottom but below everyone.
This is the peak plot, where you can see how the peak increases as I vary the output amplitude of the function generator (put in reverse order so you can see the smaller peaks on top of the larger peaks for effect).
[Rana, Giordon, Zach]
This serves as an update of the last few days of coating Q work. This is terrible elog practice on my part; all underlings should do as I say and not as I do.
Giordon was sick last week, so on Friday Rana and I went in to try and ring up a mode. At first we used a fixed sinusoidal drive out of the SR830 lock-in amplifier to characterize the EM pickup with the laser blocked. It was large, so we took some steps to mitigate it:
Both of these were done so that components mentioned were not so close to the HV amplifier (they previously shared the rack).
This pushed the laser-off EM pickup into the noise. There still seemed to be a pickup on the laser light at the drive frequency (as measured using a single PD), but there was strong enough common-mode rejection at the differential readout the the peak was in still in the noise with the laser on. At this point, we decided we wanted a finer frequency resolution, so we began making transfer function measurements using the internal drive of the Agilent spectrum analyzer (i.e., drive the ESD with the source through the HV amp, then measure the ratio of the SR560 output to the source). The HVDC bias was 2 kV, and the AC amplitude was 1 kVpk after the HV amp.
Using a span of ~70 Hz (24.15-24.22 kHz) and 1600 points (i.e., bandwidth ~0.043 Hz), with 1-s settle and 25-s integration time, we measured a very flat response in magnitude and phase:
Rana recorded this data on Sunday, after which he did something to the averaging settings(?) and left it running for another long measurement, which was also uninteresting:
It appears that the span was changed to 90 Hz (24.16-24.25 kHz), giving df = 0.056 Hz with 1600 pts, but the averaging remained unchanged. I am not sure why the measurement is so much nastier.
Yesterday (Monday), Giordon moved the laser head itself away from the rack with the HV amp in it. It is now next to the vacuum chamber, along the long edge of the table, closer to the PDs:
Following this change, we took another transfer function using the same frequency settings as in the last measurement, but with 5-s settle and 100-s integration time. The data was taken in a more sophisticated way, and it appeared that there might be some sort of resonance feature near 24.19 kHz:
To get a better look, I changed the span to 10 Hz (24.185-24.195 kHz), using the same averaging settings. This sweep shows no obvious features:
The sweeps looked strange to me, as if there was something broken. Zach assured me that the solder/paste is sound. He tested the voltage continuity of the ESD just before pumping down.
I used the HV knob to change the ESD's DC voltage from -2 kV to +2 kV to look for arcing, etc. There is indeed some arcing (the next version should have less pointy edges).
However, the big discovery for me is that any voltage beyond +/- 500 V is enough to stick the optic to the ESD by the electrostatic force. All of our runs for the past few days have been at +1-2 kVdc, so the optic was stuck the whole time.
I have now set the DC voltage to +200 V and confirmed by eye that the optic is swinging (its obvious with a flashlight): the red laser beam swings around in the chamber with a several second period.
I have set up a sweep with a 0.1 Vpp with a 200 Hz span around 5030 Hz and with a 5 sec settling time and a 55 s integration time per point. Also the 'auto resolution' feature of the sweep is on. Let's see what happens.
Next time around, we should set up active damping or make the yaw frequency higher by a factor of 5-10.
I took the Giordon FEM file and changed the material from 'silica glass' to 'Corning 7940' since I think the material parameters more closely resemble the thing we have in the can.
Then I changed the mesh size parameter to 'Extremely Fine'. The eigenfrequencies changed with every resolution, which indicates that we're not yet meshing fine enough...
I've uploaded a new version to the SVN: its in COMSOL/CoatingQ/ where we can put all of our coating Q matlab codes, documents, models, etc.
Frank and Dmass needed their turbopump back, so I brought back the original one that has been on the gyro lately. So that we don't have problems in the future, I added an extra valve I found in the ATF to the setup. This way, the pump can be shared between the gyro and the coating Q setup without having to vent each time:
While I was doing this, I noticed that the output-side viewport of the tank is totally ass-looking. It is covered in some kind of translucent goop on what appears to be both sides. The input side viewport, by comparison, is quite clean.
RA: I wiped the outside and the scum remains. I think its all on the inside. Lets wipe it next time we open.
I had set everything up and the chamber was pumped down, but Rana suggested that we not waste this pump cycle and instead adjust the ESD before re-pumping. So, we vented the tank and I left him tinkering with the ESD. I think the basic plan is to move it back to the center of the sample (horizontally), so that there is less of a yaw component to the net force.
[Rana, Zach, Giordon, and more Giordon]
Using all changes from before - I know that the pick-up noise is drastically reduced when I put the photodetectors on a 20 dB gain (gain of 10 compared to 0 dB gain) and the preamp gain is set to 100. After I spent time watching Rana's magic brown fingers on the spectrum analyzer and the sine sweep - I went ahead and did one (at a higher resolution). Let me slightly elaborate on the set-up:
The laser is on the side of the vacuum chamber, the wobbly periscopes are being used, both photodiodes (20dB gain) are unblocked, using a differential output from a preamplifier with a gain of 100, bandwidth of 1kHz to 100kHz. 20 minutes into the scan - I notice a really strange feature:
When I mean strange, I mean interesting strange - not... creepy or weird. I think it's interesting because it's not a "peak" but more broad as if someone flipped a switch and the transfer function increased for a short period, and then flipped the switch off and it slowly went away... This isn't due to any dancing antics of yours truly - as I took this picture as soon as I walked over there to check it after like 30 minutes after starting it... I might feign a guess and say, that given that it's a broad peak - it may just be something like the vacuum being crazy... So, after this scan is done, I'll save it and run it one more time before actually calling it a night.
Update: it appears that this peak shows up again with the second scan and the same conditions duplicated. It is definitely not time-dependent (because something happened during the time period it scanned it) and definitely exists in our system. More analysis needs to be done to figure out what it is. A suggestion would be to turn off the vacuum, run the scan, and see if it still persists (and since we're sealing the chamber up tomorrow - that'll be convenient).
Update: the plots are made - for both sweeps (they are identical, just taken one right after the other):
Again, most of the commentary about the plots is within the obvious peak in the magnitude of the transfer function while the phase appears to be "noisy" . Peak is centered at 4997 Hz with a 4 Hz span (roughly).
In the morning, I'll swap out the periscopes and sweep at 21.408 kHz.
To be updated...
As Zach writes, after we took the can off the top, I repositioned the ESD by one screw position so as to center it on the optic. The idea is that then the DC bias will generate less DC yaw torque.
After we got to ~! torr, we took the bias up to 100 V and the optic still seems to be parallel to the ESD. The dominant signal in the PD signals seems to be the pendulum mode (at ~1 Hz) rather than the yaw. The presence of the pendular modes indicates that the thing is not touching.
After the pressure goes below 1 mTorr, we can up the bias and try again.
We also measured the pickup into the individual PDs and the difference signal. Seems to be much less now; Giordon has the notes which give the quantitative numbers.
I've uploaded photos of the optic + ESD before movement to our shared Picasa; forgot to take photos afterwards.
Since we care about STRESS induced bi-refringence, I've plotted here the radial and azimuthal stress components for the modes instead of displacement.
[Giordon, Rana, Eric, Zach]
The vacuum chamber was sealed off today at 11:25am (Pressure ~<= 1mTorr) with the HV at 1kV bias - observing that there were no arcs on the ESD. From this point on, detailed observation of the pressure and time were recorded to make the vacuum depressurization plot seen below - it's basically a constant rate of roughly 1 mTorr every 30 minutes (this means that, if we run overnight - we can reasonably expect the chamber to reach 20 mTorr if it is pumped down to 1 mTorr initially).
Below are the transfer functions: all were sampled using a 20 second integration time with a 3 second settling time, with 1000 points. The title of each plot should be mostly descriptive of what the data is.
These seem pretty boring - there might be some peaks (and the SNR is pretty low) or there aren't any peaks and we are either scanning the wrong regions, in the right region but need better resolution, or we can't excite our modes strongly enough.
This last one was to try and find the mode that Rana pointed out at 46.880 kHz (see a previous eLog entry for a picture of the radial and azimuthal stresses from COMSOL). Again, it doesn't seem very interesting.
I'll be in tomorrow and I want to try and higher resolution scan around 24.7 kHz - it's a gut feeling. I'll keep the chamber at 1 mTorr, run something for a good 5-8 hours, and pump it back down again. The HV bias would be at 1 kV with a 0.5V pk-pk sinusoidal drive.
There was some issue with the pumps apparently. Giordon will detail this in the elog very soon.
This evening, around 7:30, I restarted the turbo. The coarse gauge was bottomed out at 1 mTorr.
I turned on the laser and the HV. Then re-aligned the laser and the input and output optics and rebalanced the detectors.
Wondering why we don't see any signal. Is the stress induced polarization modulation too small? Seems unlikely since we're almost shot noise limited in the readout. Perhaps the ESD force pattern is too weak or perhaps the ESD is broken (open rather than short) ?
Discussing with Calum and Alastair during Friday donuts, we thought we could possibly use the laser vibrometer.
Triggered by that, I wondered if we could just make a Michelson with the disk making the reflection for one arm.
Then this evening, I noticed that the reflection from the disk already has a fringing pattern in it; my guess is that this is from the two surfaces. Maybe this fringe signal already has the mode vibration in it?
The issue with the pumps was basically my fault. I misinterpreted what was meant by turning on the pump and had forgotten to turn on the backing pump first before the UHV pump. Alastair went ahead and fixed this and reset the error as well (thanks!)
I'll be making a few mechanical adjustments to the set-up today and making it less sloppy. The sweep that Rana did from 20-22 kHz doesn't reveal any modes:
As a reminder - we predict a mode around 21kHz and allowed for up to 5% error in frequency (5% of 21kHz is roughly 1kHz).
Edit: currently have a sweep going 1kHz span centered at 21.4kHz.
Do you mean short rather than open?
We were considering the interferometric readout from the start, but the feeling was that we would just get killed by the CoM motion (pendulum, torsion, tilt, etc.). That was supposed to be one advantage of a transmission measurement.
I agree that the fringe we see is probably from the parallel disk surfaces. I'm having a hard time seeing how we could use this to our advantage, though. Judging by your COMSOL stuff, I wouldn't think there is a first-order displacement of the disk surfaces relative to each other for these modes. Also, if there was, we would have to be very careful about the yawing of the disk, and we'd also be sort of at the whim of the system in terms of whether the fringe is dark or bright or whatever without any stress.
At 11:53am - I turned off the turbo pump (it was at 833 Hz) and let it spin down significantly before sealing the chamber. It basically spun down to 0 Hz at 1:34pm (roughly 1.5 hours). I sealed the chamber and then turned off the backing pump. The interesting thing is that the pressure seems to be holding steady at the moment - I recorded it as 2 mTorr at 3pm, and it is currently at 2 mTorr at 4:15pm (at this point, it should around 5 mTorr by now - so somehow, the leaks are probably smaller than before!) I've also replaced the back periscope (closest to the PDs and spectrum analyzer) with a much heavier one.
Given that we're not seeing any signs at all of a mode around this area - I'm running two more good-quality scans; one with the HV on and one with the HV off. Just to convince myself that we can't find it, I'll compare the two scans over this region just to make sure there's no pattern or significant differences.
Scan with the HV Off and HV On separately:
Nothing super interesting - you do notice some differences, but it seems like these are random (vary by roughly a factor of sqrt(2) - which I assume is probably something like shot noise?). Given the raw data (nohv, hv) - these are computed as follows
There are some strong peaks - but nothing really catches my eye much. I'm going to conclude that we are not seeing the mode predicted at 21.4kHz - so we should look into other ways of detecting these modes.
I have also attached the data files (21400NOH == HV Off, 21400HV == HV On) which are column-separated to be imported into matlab or whatever.
I don't see that it matters for the internal reflection.
We're at the whim, but mostly there seems to be a fringe. Since the disk surfaces are not parallel, there is a first order term (although small). Might as well try it since the other method gives squat so far.
There is, indeed, a giant resistor inside of that cast aluminum box which converts the output of the HV driver into a SHV connector.
Its a thick film, Ohmite, MOX-1125-23E. Which Google tells me may be somewhere from 1-1000000 kOhm. Need a Ohm-meter to measure it, but there's not one in the lab.
However, it could be that this is why we can't ring up any modes.
Pictures are in our Picasa account.
Our HV driver is a Matsusas AMT-5B20 (http://www.matsusada.com/high-voltage/ams/AMS.pdf). According to the data sheet (attached) it has these specs:
Vdc = -5000 to + 5000 V
Imax = 20 mA
Slew = 360 V/uS
-3 dB BW = 20 kHz (full scale) or 40 kHz (10% scale)
So its not bad; we should still be able to ring up the modes up to 100 kHz, although we just drive a little harder.
(Also, remember to electrically isolate the PDs from the table)
However, I will be betting $5 that this is why we can't ring up any modes. Need to figure out a more clever way to current limit so that it doesn't completely filter out our drive.
How is the resistor involved in the SHV connector affecting the actual signal being driven? Aren't SHV connectors good for up to like 2 Amps of current? I'm assuming you're talking about the resistance here because the change in impedance at the connection between the output of the HV driver and the SHV connector is modulating our signal somewhat significantly (0 < T < 1).
Where's the Picasa account (or how do I find it?)
Wasn't the limit on our spectrum analyzer below 100kHz?
I came into lab last night around midnight or so wanting to run some more measurements - but it seems that when Rana took apart the HV (see previous post) - it wasn't put back together, so it wasn't functional.
Came into lab today, and spent a good amount of time trying to get the GPIB set-up to work (https://wiki-40m.ligo.caltech.edu/GPIB). In the end - it seems like only one of the ethernet hubs in the lab actually works - and it's being taken up by the other group there. So after my attempt to put it online - I tried instead connecting it directly to my Mac. I went into bootcamp, loaded up Windows, ran the exe, set-up an ip address, and then booted back up to my Mac and tried accessing it via telnet -- but that didn't seem to work. I'm not sure what the problem here is.
Apart from that, the lab was rather hot/warm again, and while the air conditioner appeared to be blowing "cold" air - I don't think it was strong enough to cool down the lab at any rate - so I've sorta thrown in the towel for now until I hear back from other people (mostly Rana) about the GPIB and the HV.
Edit: Rana is still working on the HV.
I measured the resistance of the resistor: ~10 MOhms.
The cable capacitance (measured from the driver side of the SHV cable) is 200 pF. This must include the cable capacitance + the ESD capacitance. I leave it to Giordon to calculate these to see if the measurement makes sense.
Then I took a transfer function from the input (Vcon-in) to the output SHV connector. To make sure it was safe, I first used a DVM to measure the SHV connector and dialed the front panel knob to minimize this voltage. Then I set the spectrum analyzer to have a source output of 2 mVpk and connected it to the HV driver input.
The transfer function shows a rolloff above ~1 kHz. However, I don't think its steep enough to be a big issue for us. I have saved this data on the box as data registers D3 & D4 for Giordon to download and plot.
Since we're having so much trouble finding modes, I decided to check the results so far by doing the FEM in 3D instead of 2D.
There are some things worth noting:
Updated files are on the SVN
I also tried to substitute BK7 for fused silica -> less than 1% change in the frequencies.
This animated GIF shows the modes shapes. The color represents not physical deformation, but the Principal Stress.
Here is a link to some COMSOL example on stress-optic effects
*** Also: the lab is cold again. Looks like the AC got fixed sometime in the afternoon of Thursday.
Plots from Rana's sweeps (transfer function of the input (Vcon-in) to the output SHV connector; wide sweep showing resonances).
The transfer function shows a rolloff above ~1 kHz. However, I don't think its steep enough to be a big issue for us.
The details of the GPIB set-up is below
We're using an SR785 in the lab, the GPIB is connected by ethernet into the Caltech network - this configuration is done and it communicates successfully with the SR785. Transfer functions and grabbing current data off of it work just fine (these were tested and compared properly).
All the scripts we refer to can be found in the SVN under CoatingQ/gpib. I have changed Eric's scripts slightly to be better formatted for MATLAB plotting in the end. For grabbing data that is currently displayed on the spectrum analyzer, please see "Grabbing Data". For running transfer functions and saving the data, see "Running Transfer Functions".
This uses the SR785.py script. From terminal or using the interactive interpreter, run python. The following code will allow you to import the functions contained in this file, and then run the functions appropriately. This saves a file with a filename you specify with 3 columns (frequency [Hz], magnitude [dB], and phase [deg]).
>> from SR785 import *
>> g = getgpibObj("18.104.22.168",6)
>> from SR785 import *
>> g = getgpibObj("22.214.171.124",6)
If you see an error [int() conversion error for example] - this is just a slight bug that appears to be related to the timeout in querying something specific about the SR785 hardware itself. Just run the last line getdata(g,"filename") again. This should usually resolve the problem.
This uses the TFSR785.py script. Edit the script to change the settings to the ones you wish to have for running the transfer function (see below the example code for what these settings might look like; highlighted lines are the generally the ones that matter). After that, simply run it by executing it:
>> python TFSR785.py
If you follow the prompts, you should be on your way. The "Serial Number" prompt is just something appended to the filename before the timestamp. This script will run the transfer function, inform you of its percentage completed, and then generate two files (one *.dat and one *.bod). The *.dat is a full output detailing all the parameters of the transfer function sweep; the *.bod is a MATLAB friendly (3 column) file that you can import directly and use.
# EDIT SETTINGS HERE
#Edit these options carefully
#Undefined items may produce undefined or wrong
ipAddress = "126.96.36.199" # Ip of GPIB Device
gpibAddress = int(6) # port of GPIB device (10) is deafult
filename = filetime # Add time to filename
memo = "" # Use this option to note miscellaneous things.
title = "TSFRSR785" # Title of the measurement.
startFreq = "3.5kHz" #Start frequency. Units can be mHz, Hz or kHz.
stopFreq = "7kHz" #Stop frequency. Units can be mHz, Hz or kHz.
##### changing freq range requires editing the index, summary table, and dataFile.write(freq[...]) below (near bottom)! ######
numOfPoints = int(2047) #Number of frequency points"
sweepType = "Log" #Sweep type: Log or Linear (default: Log)
excAmp = "300mV" #Excitation amplitude
settleCycles = int(44) #Settle cycles
intCycles = int(444) #Integration cycles
inputCoupling1 = "AC" #CH1 input coupling. DC or AC (default:AC)
inputCoupling2 = "AC" #CH2 input coupling. DC or AC (default:AC)
inputGND1 = "Float" #CH1 input grounding. Float or Ground (default:Float)
inputGND2 = "Float" #CH2 input grounding. Float or Ground (default:Float)
arMode = "UpOnly" #Auto range mode: UpOnly or Tracking (default: UpOnly)
correction = float(26.0) #Correction to apply to data in dB (formula: DATA - CORRECTION, default: 26.0)
inputDiff1 = "A-B" #CH1 input Differential A-B, or single ended A"
The transfer function that is run computes Ch2a/Ch1a. The documentation here is very lacking on how to configure it any other way - but I can't imagine that we would ever need to change the way we do it now.
If the GPIB is not functioning correctly - make sure you're able to "ping" it
>> ping 188.8.131.52
If this pinging times out, make sure the GPIB is plugged into the ethernet hub and is powered on. If this persists, unplug both the ethernet and power from the GPIB, wait 10 seconds, and then plug it back in. It takes roughly 5-7 minutes for it to fully boot-up again.
If this pinging does not time out, then verify that the SR785 has the GPIB address set to 6 and not it's default 11.
If all else fails, panic until someone helps you.