This is an update of my previous reports on applications of deep learning to the reconstruction of PRMI degrees of freedom (MICH/PRCL) from real free swinging data. The results shown here are improved with respect to elog 13274 and 13294. The training is performed in two steps, the first one using simulated data, and the second one fine tuning the parameters on real data.
This step is exactly the same already described in the previous entries and in my talks at the CSWG and LVC. For details on the DNN architecture please refer to G1701455 or G1701589. Or if you really want all the details you can look at the code. I used the following signals as input to the DNN: POPDC, POP22_Q, ASDC, REFL11_I/Q, REFL55_I/Q, AS55_I/Q. The network is trained using linear trajectories in the PRCL/MICH space, and signals obtained from a model that simulates the PRMI behavior in the plane wave approximation. A total of 150000 trajectories are used. The model includes uncertainties in all the optical parameters of the 40m PRMI configuration, so that the optical signals for each trajectory are actually computed using random optical parameteres, drwn from gaussian distributions with proper mean and width. Also, white random gaussian sensing noise is added to all signals with levels comparable to the measured sensing noise.
The typical performance on real data of a network pre-trained in this way was already described in elog 13274, and although being reasoble, it was not too good.
Real free swinging data is used in this step. I fine tuned the demodulation phases of the real signals. Please note that due to an old mistake, my convention for phases is 90 degrees off, so for example REFL11 is tuned such that PRCL is maximized in Q instead of I. Regardless of this convention confusion, here's how I tuned the phases:
Then I built the following training architecture. The neural network takes the real signals and produces estimates of PRCL and MICH for each time sample. Those estimates are used as the input for the PRMI model, to produce the corresponding simulated optical signals. My cost function is the squared difference of the simulated versus real signals. The training data is generated from the real signals, by selection 100000 random 0.25s long chunks: the history of real signal over the whole 0.25s is used as input, and only the last sample is used for the cost function computation. The weights and biases of the neural network, as well as the model parameters are allowed to change during the learning process. The model parameters are regularized to suppress large deviations from the nominal values.
One side note here. At first sight it might seems weird that I'm actually fedding as input the last sample and at the same time using it as the reference for the loss function. However, you have to remember that there is no "direct" path from input to output: instead all goes through the estimated MICH/PRCL degrees of freedom, and the optical model. So this actually forces the network to tune the reconstruction to the model. This approach is very similar to the auto-encoder architectures used in unsupervised feature learning in image recognition.
After trainng the network with the two previous steps, I can produce time domain plots like the one below, which show MICH and PRCL signals behaving reasonably well:
To get a feeling of how good the reconstruction is, I produced the 2d maps shown below. I divided the MICH/PRCL plane in 51x51 bins, and averaged the real optical signals with binning determined by the reconstructed MICH and PRCL degrees of freedom. For comparison the expected simulation results are shown. I would say that reconstructed and simulated results match quite well. It looks like MICH reconstruction is still a bit "compressed", but this should not be a big issue, since it should still work for lock acquisition.
There a few things that can be done to futher tune the network. Those are mostly details, and I don't expect significant improvements. However, I think the results are good enough to move on to the next step, which is the on-line implementation of the neural network in the real time system.
The weight of SS can with copper liner is 12.2 kg
Is 1 Amp for the heating jacket going to be enough? We should have some headroom.
to get the sensors to read the same values they have to be in direct thermal contact with the metal block - there can't be any adapter board in-between
for the 2nd attempt, I also recommend encasing it in a metal block rather than just one side. You can drill some 7-10 mm diameter holes in an aluminum or copper block. Then put the sensors in there and plug it up with some thermal paste.
Nothing tripped. No obvious damage.
I brought the DEI Pulser unit and a suitable Pockels cell over from Bridge today (I also found an identical Pockels cell already at the 40m on the SP table, now that I knew what to look for).
I also brought the 200MHz AOM (Crystal Technology 3200-1113) along which can achieve rise times of 10 ns(!). Before I start setting up the Pockels cell I wanted to try this different AOM and look at its switching behavior. It asks for a much smaller beam (<65 um diam.) than what's currently in the path to the fiber (500 um diam.), although it's clear aperture is technically big enough (~1mm diam.). So I still tried, and the result was a somewhat elliptical deflected beam, and the slower decay was again visible after switching the RF input.
I was using the big Fluke function generator for the 200MHz seed signal, a Mini Circuits ZASWA-2-50 switch and a Mini Circuits ZHL-5W-1 amplifier. For the last two I moved two power supplies (+/-5V for the switch and +24V for the amplifier) into the PSL enclosure. I started at low seed power on the Fluke, routing the amplified signal into a 20dB attenuator before measuring it with an RF power meter. The AOM saturates at 2.5W (34 dBm), which I determined is achieved with a power setting on the Fluke of -4 dBm. As expected, this AOM performed faster (~80ns fall time) but I again observed the slower decay.
This struck me as weird and I started swapping components other than the AOM, which I probably should have done before. It turned out that it was the PD I was using (the same PDA10CF Gautam had used for his MC ringdown investigations). When I changed it to a PDA10A (Si diode, 150MHz bandwidth) the slow decay vanished! One last round of crappy screenshots:
Rather than proceeding with the Pockels cell, tomorrow I will make the beam in the AOM smaller and hope that that takes care of the ellipticity. If it does: the AOM can theoretically switch on ~10ns timescale, same for the switch (5-15ns typical), and the amplifier is non-resonant and works up to 500MHz, so it shouldn't be a limiting factor either. If this doesn't work out, we can still have ~100ns switching times with the other AOMs.
So it would seem that there is some other noise which has a 1/f^2 shape and is at the same level we expected the DAC noise to be at. Rana suggested checking coherence with MC transmission to see if this could be laser intensity noise.
I also want to re-do the actuator calibrations for the vertex optics again before re-posting the revised noise budget.
I stumbled upon a faster way to stream data from the TDS3014 oscilloscopes to disk, which speeds the loss measurements up by a lot: ftp://sprite.ssl.berkeley.edu/pub/sharris/MAVEN_LPW_Preamp/109_TDS3014B_control/tds3014b.py
This convenient(!) set of scripts contains a function that parses the scope's native binary format, for which the acquisition of 1 screenful of data takes <1s as opposed to ~20s, into readable data. I tested it for a bit and concluded that it does what it actually claims to do, but there's one weirdness: It get's the channel offset wrong. However this doesn't matter in our measurement because we're subtracting the dark level, which sees the same (wrong) offset. Other than that it seems okay.
So I started a new set of armloss measurements, and since the data acquisition is now much faster, I was able to squeeze a set of 20 individual measurements for each arm into ~30 minutes. This is the procedure I follow when I take these measurements for the XARM (symmetric under XARM <-> YARM):
I will write a more comprehensive post describing the data acquisition and processing, let's just look at the results for now: The "uncertainties" reported by the individual measurements are on the order of 1-2 ppm (~1.9 for the XARM, ~1.3 for the YARM). This accounts for fluctuations of the data read from the scope and uncertainties in mode-matching and modulation depths in the EOM. I made histograms for the 20 datapoints taken for each arm: the standard deviation of the spread is a little over 2ppm. We end up with something like:
XARM: 49.3 +/- 2.1 ppm
YARM: 20.3 +/- 2.3 ppm
I used Gautam's mode measurement of the auxiliary NPRO (w=127.3um, z=82mm) for the spacing of the optics on the PSL table for the fiber injection and light modulation. As mentioned in previous posts, for the time being there is no Faraday isolator and no broadband EOM installed, but they're accounted for in the mode propagation and they have space reserved if desired/required/available.
The coupler used for the injection is a Thorlabs F220APC-1064, which allegedly collimates the beam from the fiber type we use to 2.4mm diameter, which I used as the target for the mode calculations. I coupled the first order diffracted beam to a ~60m fiber, which is a tad long but the only fiber I could locate that was long enough. The coupling efficiency from free-space to fiber is 47.5%, and we can currently get up to 63 mW out of the fiber.
Tomorrow Steve and I are going to pull the fiber through protective tubing and bring it to the AS port. The next step is then characterizing the beam out of the collimator to match it into the interferometer.
As far as the switching itself is concerned: I confirmed that the exponential decay is still present when looking at the fiber output. I located the DEI Pulser unit in the QIL lab, and also found several more AOMs, including a 200MHz Crystal Technologies one, same brand that the PSL has, where the ringdown was not observed. According to past elogs, with good polarizers we can expect an extinction ratio of ~200 from the Pockels cell, which should be fine, but it's going to be tradeoff switching speed <-> extinction (if the alternate AOM doesn't show this ringdown behavior).
MC autolocker was not working - PCdrive was railed at its upper rail for ~2 hours judging by the wall StripTool trace. I tried restarting the init processes on megatron, but that didn't fix the problem. The reason seems to have been related to c1iool0 failing - after keying the crate, autolocker came back fine and MC caught lock almost immediately.
Additionally, c1susaux, c1auxex,c1auxey and c1iscaux are also down. I'm not planning on using the IFO tonight so I am not going to reboot these now.
Today, I stuck on the sensors to a metal block using a flag, rubber bands, and some thermal paste (1st attachment). I then wrapped the whole thing in about 4 layers of insulation and a lot of tape (2nd attachment). The only things leading out of the box were the three connections to the sensors and a thermometer. I then connected the wires to their respective places on the board of the sensor. To get the readings out we would need to use an ADC. Gautam and I checked to make sure the ADC we have inside the lab goes from -10V to 10V so that it would be able to measure the 3V value the sensor typically measures. We then tried to connect all three sensors to a DC source simultaneously, but unfortunately one of them seems to have disconnected somewhere during the process, as it only showed 1.2V instead of 3V. I plan to fix this tomorrow morning so that we can hopefully set this up soon.
I took off the AD590 and attached it to two long wires leading out from the board. This will allow us to attach the sensor to a metal block and not have to stick the whole board to it. I have also completed three identical copies of this and it's pretty much ready to be tested. According to Craig and Andrew's elog here, the sensor is very noisy and they added in a low pass filter to fix that, so that's something to consider for the final version of the circuit. I'll test what I have so far and see how that goes. We still need to figure out how to get readings from the sensors.
To attach the sensor to the metal block, I'll use some thermal paste and fasteners. I'll also put a thermometer on the block to record the actual temperature. I'll then wrap it in some insulation we have in the lab and have only some wires leading out of it to make measurements. I'll leave this setup overnight and record the outputs for about a full day. The fluctuations between the sensors will then indicate the noise of each individual sensor.
This is an update on the results already presented earlier (refer to elog 13274 for more introductory details). I improved significantly the results with the following tricks:
An example of time domain reconstruction is visible below. It already looks better than the old results:
As before, to better evaluate the performance I plotted averaged values of the real signals as a function of the reconstructed MICH and PRCL positions. The results are compared with simulation below. They match quite well (left real data, right simualtion expectation)
One thing to better understand is that MICH seems to be somewhat compressed: most of the output values are between -100 and +100 nm, instead of the expected -lambda/4, lambda/4. The reason is still unclear to me. It might be a bug that I haven't been able to track down yet.
I was unable to download data using nds2. Gabriele had reported similar problems a week ago but I hadn't followed up on this.
I repeated steps 5-7 from elog 13161, and now it seems that I can get data from the nds2 servers again. Unclear why the nds2 server had to be restarted. I wonder if this is somehow related to the mysterious acromag EPICS server tmux session dropout.
I decided to calculate the fluctuation in power that we will have in the heater circuit. The resistors we ordered have 50 ppm/C and it would be useful to know what kind of fluctuation we would expect. For this, I assumed that the heater itself is an ideal resistor that has no temperature variation. The circuit diagram is found in Kevin's elog here. At saturation, the total resistance (we will have a resistor instead of for our new design) will be . Therefore, with a 24V input, the saturation current should be . Therefore, the power in the heater should be (in the ideal case)
Now, in the case where the resistor is not ideal, let's assume the temperature of the resistor changes by 10C (which is about how much we would like to heat the whole thing). Therefore, the resistor will have a new value of . The new current will then be and the new power will be . So the difference in power going through the heater is about 0.00088W.
We can use this power difference to calculate how much the temperature of the metal can we wish to heat up will change. where is the thermal conductivity and x is the thickness of the material. For our seismometer, I calculated it to be 0.012K.
Tonight, I was able to lock the DRMI, turn on the whitening filters for the sensing PDs, and also turn on the coil de-whitening filters for ITMX, ITMY and BS. However, I didn't see the expected improvement in the MICH spectrum between ~50-300 Hz . Sad.
I basically went through the list of tasks I made in the previous elog. Some notes:
Attachment #1: Comparison of MICH_ERR with and without the BS de-whitening. Note that the two ITMs have their coils de-whitened in both sets of traces.
Attachment #2: Spectra of MICH output and one of the BS coil outputs in both states. The DAC RMS increases by ~30x when the de-whitening is engaged, but is still well within limits.
So it looks like the switching of paths is happening correctly. The "CDS BIO STATUS" MEDM screen also shows the appropriate bits toggling when I turn the de-whitening on/off. There is no broadband coherence with MCF between 50-300 Hz so it seems unlikely that this could be frequency noise.
Clearly I am missing something. But anyways I have a good amount of data, may be useful to put together the post CDS/electronics modification DRMI noise budget. More analysis to follow.
not immediately necessary, since you have already got it sort of working, but one of these days we should optimize this for real. In the past, we used to do this by putting a o'scope on the coil Vmon during the switching to catch the transient w/ triggering. We download the data/picture via ethernet. Run for loop on tolerance to see what's what.
Now that the DRMI locking seems to be repeatable again, I want to see if I can improve the measured MICH noise. Recall that the two dominant sources of noise were
In preparation for some locking attempts today evening, I did the following:
Hopefully, I can successfully engage a similar transition tonight with the DRMI locked. The main difference compared to this daytime test is going to be that the MICH control signal is also going to be routed to the BS.
Tasks for tonight, if all goes well:
Unrelated to this work: the PMC was locked near the upper rail of the PZT, so I re-locked it closer to the middle of the range.
Surprisingly, there was no evidence of REFL55 behaving weirdly tonight, and I was able to easily lock the DRMI on 1f error signals using the recipe I've been using in the last few months.
I did some work today to see if I could use the IR beat for ALS control. Initial tests were encouraging.
I will now embark on the noise budgeting.
I am leaving the green beat electronics on the PSL table in the switched state for further testing...
I haven't done an exhaustive check just yet, but I have loaded a few testpoints in dataviewer, and ran a script that use testpoint channels (specifically the ALS phase tracker UGF setting script), all seems good.
So if I remember correctly, the major CDS fix now required is to solve the model unloading issue.
Thanks to Jamie/Jonathan Hanks/KT for getting us back to this point! Here are the details:
After reading logs and code, it was a simple daqdrc config change.
The daqdrc should read something like this:
configure channels begin end;
What had happened was tpconfig was put before the configure channels
begin end. So when daqd_rcv went to configure its test points it did
not have the channel list configured and could not match test points to
the right model & machine. Dave and I suspect that this is so that it
can do an request directly to the correct front end instead of a general
broadcast to all awgtpman instances.
Simply reordering the config fixes it.
I tested by opening a test point in dataviewer and verifiying that
testpoints had opened/closed by using diag -l. Xmgr/grace didn't seem
to be able to keep up with the test point data over a remote connection.
You can find this in the logs by looking for entries like the following
while the daqd is starting up. When we looked we saw that there was an
entry for every model.
Unable to find GDS node 35 system c1daf in INI fiels
I re-enabled the MC SUS damping and IMC locking for some IFO work just now.
MC1, MC2 and MC3 damping turned off to see glitching action at 9:57am
There was a pretty large glitch in MC1 about an hour ago. The misalignment was so large that the autolocker wasn't able to lock the IMC. I manually re-aligned MC1 using the bias sliders, and now IMC locks fine. Attached is a 90 second plot of 2K data from the OSEMs showing the glitch. Judging from the wall StripTool, the IMC was well behaved for ~4 hours before this glitch - there is no evidence of any sort of misalignment building up, judging from the WFS control signals.
After our Demod/Whitening electronics investigations suggested nothing obviously wrong, I decided to give DRMI locking another go tonight.
Not sure what to make of all this .
I got in a ~15 minute lock, but I wasn't prepared to do any sort of characterization/ sensing / attempt to turn on coil-dewhitening, and I'm too tired to try again tonight. I was however able to whiten the error signals, as I have been able to do in the past. There is a ~45Hz bump in MICH that I haven't seen in the past.
I'll try and do some characterization tomorrow eve, but it's encouraging to at least get back to the pre-FB-failure state of locking.
We did an ingenious checkup of the whitening board tonight.
I've restored all connections at that we messed with at the LSC rack to their original positions.
The TT alignment seems to be drifting around more than usual after we disconnected one of the channels - when I came in today afternoon, the spot on the AS camera had drifted by ~1 spot diameter so I had to manually re-align TT1.
Based on my tests, everything on the Demod board seems to work as expected. I need to think more about what else could be happening here - specifically do a more direct test on the whitening board.
sudo dd if=SL-7.3-x86_64-2017-01-20-LiveCD.iso of=/dev/sdf
Debian doesn't like EPICS. Or our XY plots of beam spots...Sad!
No, not confused on that point. We just will not be testing OS versions at the 40m or running multiple OS's on our workstations. As I've said before, we will only move to so-called 'reference' systems once they've been in use for a long time.
Ubuntu16 is not to my knowledge used for any CDS system anywhere. I'm not sure how you expect to have better support for that. There are no pre-compiled packages of any kind available for Ubuntu16. Good luck, you big smelly doofuses. Nyah, nyah, nyah.
K Thorne recommends that we use SL7.3 with the 'xfce' window manager instead of the Debian family of products, so we'll try it out on allegra and rossa to see how it works for us. Hopefully the LLO CDS team will be the tip of the spear on solving the usual software problems we have when we "~up" grade.
nominal changed from 22 to 23 dB to minimize PC drive RMS
previous loop gain measurement is sort of bogus (made on SR785); need some 4395 loop measurements and checking of crossover and error point spectrum
I have moved the USB flash drives from the electronics bench back into the middle drawer of the cabinet next to the AC which is west of the fridge. Drawer re-enlabeled.
Today I tried debugging the mysterious increase in REFL55 signal levels in the DRMI configuration. I focused on the demod board, because last week, I had tried routing these signals through different channels on the whitening board, and saw the same effect.
I did a quick check by switching the output of the REFL55 demod board to the inputs normally used by AS55 signals on the whitening board. Setting the whitening gain to +18dB for these channels had the same effect - ADC overflow galore. So looks like the whitening board isn't to blame. I will have to check the demod board out.
All connections have been restored untill further debugging later in the evening.
A couple of minutes ago, Larry W swapped the fibers to our 40m Edgeswitch (BROCADE FWS 648G) to a faster connection. This is the switch to which our gateway machine, NODUS, is connected. The actual swap itself happened at the core router in Bridge, and took only a few seconds. After the switch, I double checked that I was able to ssh into nodus from my laptop, and Larry informed me that everything is working as expected on his end.
Larry also tells us that the other edgeswitch at the 40m (Foundry Networks), to which most of our GC network machines are connected, is a 100MBPS switch, and so we should re-route the connections from this switch to the BROCADE switch at our convenience to take advantage of the faster connection.
I trained a deep neural network (DNN) to reconstruct MICH and PRCL degrees of freedom in the PRMI configuration. For details on the DNN architecture please refer to G1701455 or G1701589. Or if you really want all the details you can look at the code. I used the following signals as input to the DNN: POPDC, POP22_Q, ASDC, REFL11_I/Q, REFL55_I/Q, AS55_I/Q.
Gautam took some PRMI data in free swinging and driven configuration:
In contrast to the Fabry-Perot cavity case, we don't have a direct measurement of the real PRCL/MICH degrees of freedom, so it's more difficult to assess if the DNN is working well.
All MICH and PRCL values are wrapped into the unique region [-lambda/4, lambda/4]^2. It's even a bit more complicated than simpling wrapping. Indeed, MICH is periodic over [-lambda/2, lambda/2]. However, the Michelson interferometer reflectivity (as seen from PRC) in the first half of the segment is the same as in the second half, except for a change in sign. This change of sign in Michelson reflectivity can be compensated by moving PRCL by lambda/4, thus generating a pi phase shift in the PRC round trip propagation that compensate for the MICH sign change. Therefore, the unit cell of unique values for all signals can be taken as [-lambda/4, lambda/4] x [-lambda/4, lambda/4] for MICH x PRCL. But when we hit the border of the MICH region, PRCL is also affected by addtion of lambda/4. Graphically, the square regions A B C below are all equivalent, as well as more that are not highlighted:
This makes it a bit hard to un-wrap the resonstructed signal, especially when you add in the factor that in the reconstruction the wrapping is "soft".
The plot below shows an example of the time domain reconstruction of MICH/PRCL during the free swinging period.
It's hard to tell if the positions look reasonable, with all the wrapping going on.
Here's an attempt at validating the DNN reconstruction. Using the reconstructed MICH/PRCL signal, I can create a 2d map of the values of the optical signals. I binned the reconstructed MICH/PRCL in a 51x51 grid, and computed the mean value of all optical signals for each bin. The result is shown in the plot below, directly compared with the expectation from a simulation.
The power signals (POP_DC, AS_DC, PO22_Q) looks reasonably good. REFL11_I/Q also looks good (please note that due to an early mistake in my code, I reversed the convention for I/Q, so PRCL signal is maximized in Q instead than in I). The 55MHz signals look a bit less clear...
MC autolocker and FSS loops were stuck because c1psl was unresponsive. I rebooted it and did a burtrestore to enable PSL locking. Then the IMC locked fine.
c1susaux and c1iscaux were also unresponsive so I keyed those crates as well, after taking the usual steps to avoid ITMX getting stuck - but it still got stuck when the Sat. Box. connectors were reconnected after the reboot, so I had to shake it loose with bias slider jiggling. This is annoying and also not very robust. I am afraid we are going to knock the ITMX magnets off at some point. Is this problem indicative of the fact that the ITMX magnets were somehow glued on in a skewed way? Or can we make the situation better by just tweaking the OSEM-holding fixtures on the cage?
In any case, I've started listing stuff down here for things we may want to do when we vent next.
I changed the heater circuit described in this elog to a current sink. The new and old circuits are shown in the attachment. The heater is and is currently 24Ω; the sense resistor is currently 6Ω. The op-amp is still an OP27 and the MOSFET is still an IRF630.
The current through the old circuit was saturating because the gate voltage on the MOSFET was saturating at the op-amp supply rails. This is because the source voltage is relatively high: .
In the new circuit the source voltage is lower and the op-amp can thus drive a large enough to draw more current (until the power supply saturates at 25V/30Ω = 0.8A in this case). The source and DAC voltages are equal in this case and so the current is . Since this is the same current through the heater, the drain voltage is . I observed this behavior in this circuit until the power supply saturated at 0.8A. Note that when this happens and the gate voltage saturates at the supply rails in an attempt to supply the necessary current.
there is a large (~1 cm) aperture Pockels cell that Frank Siefert was using for making pulses to damage photo diodes. There is a DEI Pulser unit near the entrance to the QIL in Bridge which can drive it.
I'll look for it tomorrow, but I haven't given up on the AOMs yet. I swapped in the ISOMET modulator today and saw the same behavior, both in 0th and 1st order. The fall time is pretty much identical. Gautam saw no such thing in the PSL AOM using the same photodetector.
In the meantime I prepared the fiber mode-matching but realized in the process that I had mixed up some lenses. As a result the beam did not have a waist at the AOM location and thus didn't have the intended size, although I doubt that this would cause the slower decay. I'll fix it tomorrow, along with setting up the fiber injection, beat note with the PSL, and routing the fiber if possible.
I don't understand why the 1st order diffracted beam doesn't go to zero when you shut off the drive. My guess is that the standing acoustic wave in the AO crystal needs some time to decay: f = 40 MHz, tau = 1 usec... Q ~ 100. Perhaps, the crystal is damped by the PZT and ther output impedance of the mini-circuits switch is different from the AO driver.
In any case, if you need a faster shut off, or want something that more cleanly goes to zero, there is a large (~1 cm) aperture Pockels cell that Frank Siefert was using for making pulses to damage photo diodes. There is a DEI Pulser unit near the entrance to the QIL in Bridge which can drive it.
I worked with Kevin and Gautam to create a heater circuit. The first attachment is Kevin's schematic of the circuit. The OP amp connects to the gate of the power MOSFET, and the power supply connects to the drain, while the source goes into the heater. We set the power supply voltage to 22V and varied the voltage of the input to the OP amp. At 6V to the OP amp, we got a current of 0.35A flowing through the heater and resistor. This was the peak current we got due to the OP amp being saturated (an increase in either of the power supplies did not change the current), but when we increased the voltage of the supply rails of the OP amp from 15V to 20V, we got a current of 0.5A. We would want a higher current than this, so we will need to get a different OP amp with a higher max voltage rating, and a resistor that can take more power than this one (it currently takes 5W of power, and is the best one we could find).
Kevin and I created a simulation of this circuit using CircuitLab to understand why the current was so low (second attachment). The horizontal axis is the voltage we supply to the OP amp. The blue line shows the voltage at the point between the output of the OP amp and the gate of the MOSFET. The orange line is the voltage at the point between the source of the MOSFET and the heater. The brown line is the voltage at the point between the heater and resistor. Thus, we can see that saturation occurs at about 2.1V. At that point, the gate-source voltage is the difference between the blue curve and the orange curve, which is about 4V, which is what we measured. Likewise, the voltage across the heater is the difference between the orange curve and the brown curve, which comes out to around 8V, which is also what we measured. Lastly, the voltage across the resistor is the brown curve, which is about 2V, which matches our observations. The circuit works as it should, but saturates too soon to get a high enough current out of it.
Gautam noted that it is important to measure the current correctly. We can't just use an ammeter and place it across the resistor or heater, because the internal resistance of the ammeter (~0.5 ohm) is comparable to the resistance we want to measure, so the current gets split between the circuit and the ammeter and we get an equivalent resistance of 1/R = 1/R0 + 1/Ra, where R0 is the resistance of the part we want to measure the current across, and Ra is the ammeter resistance. Thus, the new resistance will be lower and the ammeter will show a higher current value than what is actually there. So to accurately measure the current, we must place the ammeter in series with the part we want to measure. We initially got a 1A reading on the heater, which was not correct, and our setup did not heat up at all basically. When we placed the ammeter in series with the heater, we got only 0.35A.
The last two images are the setup for testing of the heater. We wrapped it around an aluminum piece and covered it with a few layers of insulating material. We can stick a thermometer in between the insulation and heater to see the temperature change. In later tests, we may insulate the whole piece so that less heat gets dissipated. In addition, we used a heat sink and thermal paste to secure the MOSFET to it, as it got very hot.
Our next steps will be to get a resistor and an OP amp that are better suited for our purposes. We will also run simulations with components that we choose to make sure that it can provide the desired current of 1A (the maximum output of the power supply is 24V, and the heater is 24 ohm, so max current is 1A). Kevin is working on that now.
RP1 and RP3 roughing pump manual of Leybold D30A oily rotory pump
Fore pump of TP2 & TP3 Varian SH-100 Dry Scroll
TP2 and TP3 small turbo drag pump Varian 969-9361
TP2 and TP3 turbo controller Varian 969-9505
TP1 magnetically suspended turbo pump Osaka TG390MCAB, sn360 and controller TC010M and note : this pump running on 208VAC single phaseIt is not on the UPS !
Osaka Maglev Manual and Osaka Controller Communication Wiring
VC1 cryo pump CTI-Cryogenics Cryo Torr 8 sn 8g23925 SAFETY note: compressor single phase 208VAC and the head driver 3 phase 208VAC Compressor and driver have each separate power cord!
Installed at 40m wiki also
The V1 gate valve specs installed at 40m wiki page. VAT model number 10846-UE44-0007 Our main volume pumping goes through this 8" id gate valve V1 to Maglev turbo or Cryo pump to VC1
The ion pumps have 6" id gate valves:VAT 10844-UE44-AAY1, Pneumatic actuator with position indicator and double acting solenoid valve 115V 60Hz Purchased 1999 Dec 22
UHV gate valves 2.5" id. VAT 10836-UE44 Pneumatic actuator with position indicator and double acting solenoid valve 115V 60 hz, IFO to RGA VM1 & RGA to Maglev VM2
mini UHV gate valve 1.5" id. VAT 01032-UE01 2016 cataloge page 14, manual - no position indicator, VM4 next to manual adjustable fine leak valve to RGA
UHV angle valve 1.5" id, model VAT 28432-GE41, Viton plate seal, pneumatic actuator with position indicator & solenoid valve 115V & single acting closing spring MEDM screen: VM3,VC2, V3,V4,V5,V6,VA6,V7 & annuloses Each chamber annulos has 2 valves.
UHV angle valve 1.5" id, model VAT 57132-GE05 go page 208, Metal tip seal, manual actuating only with position indicator, MEDM screen: roughing RV1 and venting VV1 hand wheel needed to close to torque spec
UHV angle valve 1.5" id. model VAT 28432-GE01 Viton plate seal, manual operation only at IT gauges Hornet & Super Bee and ion pumps roughing ports. These are not labeled.
The Cryo pump interlock wiring was added too
Note: all moving valve plate seals are single.
Last night, while we were working on the ALS, I noticed the GTRY spot moving around (in PITCH) on the CCD monitor in the control room at ~1-2Hz. The operating condition was that the arm was locked to the IR, and the PSL green shutter was closed, so that only the arm transmissions were visible on the CCD screens. There was no such noticable movement of the GTRX spot. When looking at the out-of-loop ALS nosie in this configuration (but now with the PSL green shutter open of course), the Y arm ALS noise at low frequencies was much worse than the X arm.
Today, I looked into this a little more. I first checked that the Y-endtable enclosure was closed off as usual (as I had done some tweaking to the green input pointing some days ago). There are various green ghost beams on the Y-endtable. When time permits, we should make an effort to cleanly dump these. But the enclosure was closed as usual.
Then I looked at the in-loop Oplev error signal spectra for the ITMY and ETMY Oplev loops. There was high coherence between ETMYP Oplev error signal and GTRY. So I took a loop transfer function measurement - the upper UGF was around 3.5Hz. I increased the loop gain such that the upper UGF was around 4.5Hz, with phase margin ~30degrees. Doing so visibly reduced the angular movement of the GTRY spot on the CCD. Attachment #1 shows the Oplev loop TF after the gain increase, while Attachment #2 compares the GTRX and GTRY spectra (DC value is approximately the same for both, around 0.4). GTRY still seems a bit noisier at low frequencies, but the out-of-loop ALS noise for the Y arm now lines up much more closely with its reference trace from a known good time.
Y-arm ALS wasn't so stellar tonight, especially at low frequencies. I can see the GTRY spot moving on the CCD monitor, so something is wonky. To be investigated.
I was having a chat with EricQ about this today, just noting some points from our discussion down here so that I remember to look into this tomorrow.
Can we make use of the Jetstor raid array for some kind of consolidated 40m CDS backup system? Once we've gotten everything of interest out of it...
It is unclear when this was last done, and since I modified the coil driver electronics for the ITMs and BS recently, I figured it would be useful to get this calibration done. The primary motivation was to see if we could resolve the discrepancy between the current ALS noise (using POX as a sensor) compared to the Izumi et. al. plot.
Because we are planning to change the coil driver electronics further soon anyways, we decided to do the calibration at a single frequency for tonight. For future reference, the extension of this method to calibrate the actuator over a wider range of frequencies is here. The procedure followed, and the relevant numbers from tonight, are as follows.
Once these calibrations were updated, we decided to control the arms with ALS, and look at the POX spectrum. Y-arm ALS wasn't so stellar tonight, especially at low frequencies. I can see the GTRY spot moving on the CCD monitor, so something is wonky. To be investigated. But the X arm ALS noise looked pretty good.
Seems like updating the calibration did the job; see the attached comparison plot.
I moved the axuiliary NPRO to the PSL table today and started setting up the optics.
The Faraday Isolator was showing a pretty unclean mode at the output so I took the polarizers off to take a look through them, and found that the front polarizer is either out of place or damaged (there is a straight edge visible right in the middle of the aperture, but the way the polarizer is packaged prevents me from inspecting it closer). I proceeded without it but left space so an FI can be added in the future. The same goes for the broadband EOM.
There are two spare AOMs (ISOMET and Intraaction, both resonant at 40MHz) available before we have to resort to the one currently installed in the PSL.
I installed the Intraaction AOM first and looked at the switching speed of its first order diffracted beam using both its commercial driver and a combination of minicircuits components. Both show similar behavior. The fall time of the initial step is ~110ns in both cases, but it doesn't decay rapidly no light but a slower exponential. Need to check the 0 order beam and also the other AOM.
In addition to bootable full disk backups, it would be wise to make sure the important service configuration files from each machine are version controlled in the 40m SVN. Things like apache files on nodus, martian hosts and DHCP files on chiara, nds2 configuration and init scripts on megatron, etc. This can make future OS/hardware upgrades easier too.
What are the critical filesystems? I've also indicated the size of these disks and the volume currently used, and the current backup situation.
Not backed up
LDAS pulls files from nodus daily via rsync, so there's no cron job for us to manage. We just allow incoming rsync.
Local backup on /media/40mBackup on chiara via daily cronjob
Remote backup to ldas-cit.ligo.caltech.edu::40m/cvs via daily cronjob on nodus
Currently mounted on Megatron, not backed up.
Then there is Optimus, but I don't think there is anything critical on it.
So, based on my understanding, we need to back up a whole bunch of stuff, particularly the boot disks and root filesystems for Chiara, Megatron and Nodus. We should also test that the backups we make are useful (i.e. we can recover current operating state in the event of a disk failure).
Please edit this elog if I have made a mistake. I also don't have any idea about whether there is any sort of backup for the slow computing system code.
The RGA was turned on 7 days ago. It's 46 C now. The X-arm room tem ~20 C
IFO pressure 6.5e-6 Torr at IT-Hornet gauge. Valve configuration vacuum normal.
It seems like the main contribution to the RMS comes from the high frequency bump. When using the ALS loop to lock the arm to the beat, only the stuff below ~100 Hz will matter. Interesting to see what that noise budget will show. Perhaps the discrepancy between inloop and out of loop will go down.
I included the 55 MHz sideband and higher order modes in my training examples. To keep things simple, I just assumed there are higher order modes up to n+m=4 in the input beam. The power in each HOM is randomly chosen from a random gaussian distribution with width determined from experimental cavity scans. I used a value of 0.913+-0.01 rad for the Gouy phase (again estimated from cavity scans, but in reasonable agreement with the nominal radius of curvature of ETMX)
Results are improved. The plot belows show the performance of the neural network on 100s of experimental data
For reference, the plots below show the performance of the same network on simulated data (that includes sensing noise but no higher order modes)
Is it better to mount the box in the PSL under the existing shelf, or in a nearby PSL rack?
Further characterization needs to be done, but the results of this test are encouraging. If we are able to get this kind of out of loop ALS noise with the IR beat, perhaps we can avoid having to frequently fine-tune the green beat alignment on the PSL table. It would also be ideal to mount this whole 1U setup in an electronics rack instead of leaving it on the PSL table
Attachment #1 - Photo of the revamped beat setup. The top panel has to be installed. New features include:
Attachment #2 - Power budget inside the box. Some of these FC/APC connectors seem to not offer good coupling between the two fibers. Specifically, the one on the front panel meant to accept the PSL light input fiber seems particularly bad. Right now, the PSL light is entering the box through one of the front panel connectors marked "PSL + X out". I've also indicated the beat amplitude measured with an RF analyzer. Need to do the math now to confirm if these match the expected amplitudes based on the power levels measured.
Attachment #3 - We repeated the measurement detailed here. The X arm (locked to IR) was used for this test. The "X" delay line electronics were connected to the X green beat PD, while the "Y" delay line electronics were connected to the X IR beat PD. I divided the phase tracker Hz calibration factor by 2 to get IR Hz for the Y arm channels. IR beat was at ~38MHz, green beat was at ~76MHz. The broadband excess noise seen in the previous test is no longer present. Indeed, below ~20Hz, the IR beat seems less noisy. So seems like the cleaning / electronics revamp did some good.
Further characterization needs to be done, but the results of this test are encouraging. If we are able to get this kind of out of loop ALS noise with the IR beat, perhaps we can avoid having to frequently fine-tune the green beat alignment on the PSL table. It would also be ideal to mount this whole 1U setup in an electronics rack instead of leaving it on the PSL table.
Photos + power budget + plan of action for using this box to characterize the green PDH locking to follow.
GV Edit: I've added better photos to the 40m Google Photos page. I've also started a wiki page for this box / the proposed IR ALS system. For the moment, all that is there is the datasheet to the Fiber Couplers used, I will populate this more as I further characterize the setup.