Since error was high for the same input as in my previous elog http://nodus.ligo.caltech.edu:8080/40m/14089
I modified the network topology by tuning the number of nodes, layers and learning rate so that the model fitted the sum of 4 sine waves efficiently, saved weights of the final epoch and then in a different program, loaded saved weights & tested on simulated video that's produced by moving beam spot from the centre of image by sum of 4 sine waves whose frequencies and amplitudes change with time.
Input : Simulated video of beam spot motion in pitch by applying 4 sine waves of frquencies 0.2, 0.4, 0.1, 0.3 Hz and amplitude ratios to frame size to be 0.1, 0.04, 0.05, 0.08. This is divided into train (0.4), validation (0.1) and test (0.5).
Input --> Hidden layer --> Output layer
8 nodes 1 node
Activation function: selu linear
Batch size = 32, Number of epochs = 128, loss function = mean squared error
Optimizer: Nadam ( learning rate = 0.00001, beta_1 = 0.8, beta_2 = 0.85)
Normalized the target sine signal of NN by dividing by its maximum value.
Plot of predicted output by neural network, applied input signal & residual error given in 1st attachment. These weights of the model in the final epoch have been saved to h5 file and then loaded & tested with simulated data of 4 sine waves with amplitudes and frequencies changing with time from their initial values by random uniform noise ranging from 0 to 0.05. Plot of predicted output by neural network, target signal of sine waves & residual error given in 2nd attachment. The actual signal can be got from predicted output of NN by multiplication with normalization constant used before. However, even though network fits training & validation sets efficiently, it gives a comparatively large error on test data of varying amplitude & frequency.
Gautam suggested to try training on this noisy data of varying amplitudes and frequencies. The results using the same model of NN is given in Attachment 3. It was found that tuning the number of nodes, layers or learning rate didn't improve fitting much in this case.
Ah. With MC2 feedback, we have about 3 times smaller "optical gain" for the ASS A2L. We have same dither, same actuator, but we need only 1/3 actuation of the MC2 compared to the ETMY case.
We had to reduce the ASS spot servo from 1 to 0.3 to make is stable, so this means that the ASS is really merginally stable.
We did a quick check of this board today. Main takeaways:
With the correct , we expect 0V from the DAC to result in 0 actuation on the mirror, assuming that an equal 75V goes to 2 PZTs mounted diametrically opposite on the optic. Hopefully, this means we have sufficient range to scan the input pointing into the OMC and get some sort of signal in the REFL signal (while length PZT is being scanned) which indicates a resonance.
We plan to carve out some IFO time for this work next week.
I implemented this today. For now, the LSC output matrix is set to actuate on MC2 for Y arm locking. As expected, the transmission was much more stable, and the PLL control signal RMS was also reduced by factor of ~3. MC2 control signal does pick up a large (~2000 cts) DC component over a few hours, so we need to relieve this periodically.
Now that we have a workable ASS for the Y arm as well, we should be able to have more confidence in returning to the same beam spot position on the ETM and staying there during a scan using this technique.
The main improvement to be trialled next in the scanning is to improve the speed of scanning. As things stand, my script takes ~2.5 seconds per datapoint. If we can cut this in half, that'd be huge. On Wednesday night, we were extraordinarily lucky to avoid MC3 glitching, EPICS/slow machine failures, and GPIB freezes. Today, the latter reared its head. Fortunately, since I'm dumping data to file for each datapoint, this means we at least have data till the GPIB freeze.
For future measurements, we should consider locking the IMC length to the arm cavity - this would eliminate such alignment drifts, and maybe also make the PLL control signal RMS smaller.
Not related to this work: Terra, Sandrine, Keerthana and I cleaned up the lab a bit today, and made better cable labels. Aaron and I have to clean up the OMC area a bit. Huge thanks to Steve for taking care of our mess elsewhere in the lab!
We managed to realize stable ASS configuration for Yarm. The transmission of 1.06~1.07 was recovered by introducing intentional beam spot offset in the horizontal direction towards the opposite side of the elliptic reflector. The end table optics were adjusted to have the spots about the center of the mirrors, lenses, and PDs/QPDs.
- The Y arm was manually aligned with a given input axis. The transmission was ~0.8.
- Then, TT2 was moved in yaw such that it introduced the horizontal beam shift at the end. By moving the spot to the opposite side of the reflector. The transmission ~0.95 was obtained after patient alignment work.
- Went to the end table and checked the spots. The beam was not at the center of the last 1" lens for the Trans PDs. The beam steering was adjusted to have the spot nicely going through the lens and the mirrors. This made the transmission level to be ~1.05.
- The beam centering on the Trans PD was checked and adjusted.
- The beam centering on the RF BBPD for the arm scan was checked. The spot was too big for that PD. The lens was slightly moved away from the PD to make the spot on the BBPD small. Now the PD saw the plateu when the steering was scanned (i.e. the spot is small enough).
- With the Y arm locked with MC2, the servo gain needs to be 0.012 instead of nominal 0.015 with ETMY to prevent from servo oscilating.
- First of all, only the bottom 4 loops out of total 8 loops were tuned. They are the servos for the beam alignment with regard to the caivty. The linearity and the zero crossings were checked with regard to the reference alignment. All of these 4 showed offsets that causes the servo running away. Don't know the reason of this offset, but it is freq dependent. Therefore the dither freqs were tuned to make the offset zeroed, and tuned the demod phases there. This kept the transmission as high as the reference (~1.05)
- This allowed us to play with the spot position a bit by tuning the caivty alignment. In the end, the transmission of ~1.08 was obtained. Using this alignment, A2L offset for ETMY Yaw was determined to be +17 (to make the error signal -17). This offset produces almost a beam radius (5mm) shifted on the end mirror towards the opposite direction of the reflector.
- The nominal servo setting made the spot servo running away. Gautam pointed out that this could be a gain hierarchy problem (i.e. the spot servos are too fast). We ended up reducing the gain of the servo from 1.0 to 0.3 to make the spot servo stable.
- All the ASS setting was stored in a new snap file "script/ASS/ASS-DITEHR_ON.snap". The previous snap was saved to "script/ASS/ASS_DITHER_ON_preVent201807.snap". This did not save the exc gains of the oscillators. Therefore "DITHER_ASS_ON.py" was modified to have the new exc gains (CLKGAIN). The old values are stored in the comments in this script.
Overall this is not an ideal situation as we don't know what is the actually cause of the offsets in the dither error signals. We expect to correct the beam clipping and the suspension sooner or later. Therefore, we will come back to the ASS again once the other issues are corrected.
IMC WFS tuning
- IMC was aligned manually to have maximum output and also spot at the center of the end QPD.
- The IMC WFS spots were aligned to be the center of the WFS QPDs.
- With the good alignment, WFS RF offset and MC2 QPD offsets were tuned via the scripts.
I recently realized that the PLL is only using about 20% of the available actuation range of the AUX PZT. The +/-10 V control signal from the LB1005 is being directly inputted into the fast AUX PZT channel, which has an input range of +/-50 V.
I recommend to install a PZT driver (amplifier) between the controller and laser to use the full available actuator range. For cavity scans, this will increase the available sweep range from +/-50 MHz to +/-250MHz. This has a unique advantage even if slow temperature feedback is also implemented. To sample faster than the timescale of most of the angular noise, scans generally need to be made with a total sweep time <1 sec. This is faster than the PLL offset can be offloaded via the slow temperature control, so the only way to scan more than 100 MHz in one measurement is with a larger dynamic range.
Input : Simulated video of beam spot motion in pitch by applying 4 sine waves of frquencies 0.2, 0.4, 0.1, 0.3 Hz and amplitude ratios to frame size to be 0.1, 0.04, 0.05, 0.08
The data has been split into train, validation and test datasets and I tried training on neural network with the same model topology & parameters as in my previous elog (https://nodus.ligo.caltech.edu:8081/40m/14070)
The output of NN and residual error have been shown in Attachment 1. This NN model gives a large error for this. So I think we have to increase the number of nodes and learning rate so that we get a lower error value with a single sine wave simulated video ( but not overfitting) and then try training on linear combination of sine waves.
Case 2 :
Normalized the target sine signal of NN so that it ranges from -1 to 1 and then trained on the same neural network as in my previous elog with simulated video created using single sine wave. This gave comparatively lower error (shown in Attachment 2). But if we train using this network, we can get only the frequency of test mass motion but we can't resolve the amount by which test mass moves. So I'm unclear about whether we can use this.
Roughing down the annuloses required closing V1 for 13 minutes
IFO is 2.2e-5 Torr
Cold cathode gauge just turned on.
Annalisa, Gautum, Koji, Terra
Summary: with the reflector setup, we measured a frequency shift of the first and second order modes! First looks of shifts show 1st HOM shift ~-10 kHz, 2nd HOM shift ~-18 kHz (carrier ~4 kHz). We saw no shift with the cylinder/lenses set up.
- - - - -
Tonight we modified the cavity scan setup: the LO is provided by the Marconi which, at the same time, is also used to scan the AUX laser frequency instead of the Agilent. In order to get rid of the free running noise between Marconi and Agilent, the Marconi frequency was scanned and, point by point, the Agilent center frequency was changed accordingly. In order to speed up the process, the whole procedure was automated. The script is called AGfast.py and can be found in /users/annalisa/postVent.
One thing that helped in improving the data quality of the phase information was to set the Agilent IF bandwidth @1kHz. Not yet clear why, but it was better than having a lower bandwidth. To be further investigated.
With this setup, we made some coarse scan of the full FSR and then we "zoomed" around the main peaks in order to increase the resolution and get a more precise information about the peak frequency.
Here are the frequency ranges that we scanned:
We powered the heater of the lenses setup @4:55 am at 14.4V and 0.9A. Then we slightly increased the power @5:05am and the final "hot state" configuration is with heater powered at 16V and 0.9A.
With this setup we couldn't see any frequency shift
Then, at around 6:30 am we turned on the reflector setup and we measured a frequency shift of the first and second order modes. First scans show 1st HOM shift ~10 kHz, 2nd HOM shift ~18 kHz. First attachment shows carrier hot/cold, second attachment shows HOM2 hot/cold. We started to get plauged by high seismic noise. Heaters turned off at 7:45 am. Lots of scans and actual analysis to go.
gautam: about the questionable plotting -
My personal favourite plot is Attachment #3, which is a 5 MHz scan (cold) to identify positions of the various peaks. The power of including phase information in the analysis is clear. The second FSR on the right edge of the plot is not as prominent as the first is because the arm transmission was degrading throughout the measurement. For future measurements, we should consider locking the IMC length to the arm cavity - this would eliminate such alignment drifts, and maybe also make the PLL control signal RMS smaller.
Per Steve's instructions, we did the following:
Is the reflector too close to the beam and causing clipping?
For unknown reasons, the Y arm ASS does not maximize TRY. So we are in the unfortunate situation of neither arm having a working ASS servo. To be worked on later.
IFO P1 6e-4 Torr, manual gate valve is fully open
The annuloses will be pumped down tomorrow.
Valve configuration: vacuum normal, RGA and annuloses are not pumped
The manual gate valve scan was clean. Atm1 TP1 was pumping on it overnight.
Pumpdown continued to hand over the pumping to TP1 maglev turbo
V1 was opened at P1 400 mTorr with manual gate at 3/4 turn open position as Magev at 560 Hz rotation.
This is the first time we pumping down from atm with one small "beer can" turbo and throttled gate to control load on small turbo forepump
The 70 l/s turbo is operating at 50k RPM, 0.7 A and 31 C, pumping speed ~ 44 mTorr/h at 200-400 mTorr range.
Watching foreline pressures and current one can keep opening gate valve little by little the so the load is optimized. It is working but not fast.
Let's keep small turbo at 0.8 Amp and 32 C max at this pumpdown.
The P1 pressure is 380mTorr. I allowed Gautam to use the full PSL power (~1W).
Two aux fans on to hold tubo temps TP1 & TP3 . Atm3
This is the first time we pumping down from atm with ONE small "beer can" turbo and throttled gate valve to control load on small turbo forepump
The 70 l/s turbo is operating at 50k RPM, 0.7 A and 31 C, pumping speed ~ 44 mTorr/h at 200-400 mTorr range with aux drypump in the foreline of TP3
[Gautam, Johannes, Koji, Annalisa]
Tonight we increased the power of the PSL laser and we achieved the lock of both arms with high power.
The AUX beam alignment to the Y arm was recovered and the PLL restored (using the Marconi as LO).
We made a quick measurement of the phase noise and the results will be posted tomorrow.
The beam on the PSL has been blocked, as well as the AUX beam on the AS table. The Marconi has been switched off.
Precondition: 4 days at atm. Atm5
HEPA tent used during the vent at ETMY It reduced partical count 10 fold of 0.5 and 0.3 micron particals. Atm6
New items in vacuum: Clean manual gate valve [Cetec made] from John Worden with 6" id....as it came from Hanford... [ Throttle able gate valve- TGV ] Atm3
( note: we have 3 more identical in the lab. The original intention was to use them for purging gates )
Optiform Au plated reflector , Induceramics heating elements, similar as existing Cooner cables and related lenses, hardwear. see 14078
OMC related item : none......... 14,110
The pumpdown is at 510 mTorr with RP1 & RP3 still pumping. Koji will shut it down the roughing later tonight. Tomorrow morning I will start the pumping by switching over to TP1 maglev.
Thanks for Koji and Gautam' help of the installation of the manual gate valve. Atm4 This will allow us to control the load on our Varian foreline D70 turbo TP3
We installed two heaters setup on the ETMY bench in order to try inducing some radius of curvature change and therefore HOMs frequency shift.
We installed two heaters setup.
Elliptic reflector setup (H1): heater put in the focus of the elliptical reflector: this will make a heat pattern as descirbed in the elogs #14043 and #14050.
Lenses setup (H2): heater put in a cylndrical reflector (made up with aluminum foil) 1'' diameter, and 2 ZnSe lenses telescope, composed by a 1.5'' and a 1'' diameter respectively, both 3.5'' focal length. The telescope is designed in such a way to focus the heat map on the mirror HR surface. For this latter the schematic was supposed to be the following:
This setup will project on the mirror a heat pattern like this:
which is very convenient if we want to see a different radius of curvature for different HOMs. However, the power that we are supposed to have absorbed by the mirror with this setup is very low (order of 40-ish mW with 18V, 1.2A) which is probably not enough to see an effect. Moreover, mostly for space reasons (post base too big), the distances were not fully kept, and we ended up with the following setup:
In this configuration we won't probably have a perfect focusing of the heat pattern on the mirror.
See Koji's elog #14077 for the final pin connection details. In summary, in vacuum the pins are:
13 to 8 --> cable bunch 0
7 to 2 --> cable bunch 2
25 to 20 --> cable bunch 1
19 to 14 --> cable bunch 3
where Elliptic reflector setup (H1) is connected to cables 0 and 1, and the lenses setup is connected to cables 2 and 3.
This is the installed setup as seen from above:
[Steve, Koji, Gautam]
We started pumping down at ~12:15PM.
Vent finalization ~ YEND
Vent finalization ~ Vertex
For the EY, instead of balancing the table, I just moved the weight approximately so that the ETMY OSEMS were at half light, but didn't check the level since ETMY is the only optic.
Some notes on OMC/AS work (Aaron/Gautam can amend/correct):
- Beam is now well centered in OMC MMT. Hits input coupling mirror and cleanly exits the vacuum to the AS table.
- Didn't see much on OMC trans, but PDs are good based on flashlight test.
- just before closing, re-aligned beam in yaw so that it gets close to the east screw on the input coupler. Aaron and I think we maybe saw a flash there with the OMC length PZT being driven at full range by a triangle wave.
- with OMC Undulators (aka tip/tilt PZT mirrors) energized, the beam was low on PZT1 mirror. We pitched ITMY by ~150 micro-rad and that centered the beam on PZT1 mirror. ITMY-OL is probably not better than 100 urad as a DC reference?
- We checked the range of Undulator 1 and we were getting ~5 mrad of yaw of the beam for the full range, and perhaps half of that in pitch. Rob Ward emailed us from Oz to say that the range is robably 2.7 mrad, so that checks out.
Even if the ITMY has to be in the wrong position to get the beam to the OMC, we can still do the heater tests in one position and then do the OMC checkout stuff in the other position.
Gautam suspects that there is a possible hysterical behaviour in the Undulators which is related to the MC3 glitching and the slow machine hangups and also possibly the illuminati.
-We noticed a ghost beam that from MC REFL (MMT2) that should be dumped during the next vent--it travels parallel to the OMC's long axis and nearly hits one of the steering mirrors for OMC refl.
-We measured the level of the table and found it ~3 divisions off from level, with the south end tilted up
-Gautam rotated and slightly translated OM5 to realign the optic, as expected. No additional optics were added.
-Gautam and I tested the TT piezo driver. We found that 3.6V on the driver's input gave 75V (of 150V) at the output, at least for yaw on piezo 1. However, as Gautam mentioned, during testing it seemed that the other outputs may have different (nonzero) offset voltages, or some hysterisis.
For the heater setup on EY table, I EQ-stopped ETMY. Only the face EQ stops (3 on HR face, 2 on AR face) were engaged. The EY Oplev HeNe was also shutdown during this procedure.
We are in the process of adding a manual gate valve between TP1 (Osaka Maglev) and the other gate valves (I suppose V1 and VM2).
The work is still on going and we will continue to work on this tomorrow. Because this section is isolated from the main volume, this work does not hold off the possible rough pumping tomorrow morning.
The motivation of this work is as follows:
- Since TP2 failed, the main vacuum volume has been pumped down by TP1 and TP3. However TP3 is not capable to handle the large pressure difference at the early stage of the turbo pumping. This cause TP3 to have excessive heating or even thermal shutdown.
- The remedy is to put a gate valve between TPs and the main vacuum to limit the amount of gas flowing into the TPs. This indeed slows down the pumping speed of turbo, but this is not the dominant part of the pumping time.
- Comfirmed TP1 is isolated.
- Unscrewed the flange of TP1.
- Remove TP1. This required to lift up TP1 with some shim as the nuts interferes with the TP1 body. (Attachment1, 2, 3)
- Now remove 10inch flange adapter. (Attachment4)
-Attach 10"-8" adapter and 8" rotational sleeve. (Attachment5)
- Attachment1: Removed the thermal cap. Checked the temperature of the oven. It was totally cold.
- Attachment2: Confirmed the RGA section was isolated. The pumps for the RGA was left running.
- Attachment3: Closed the main valve. The pumps for the main volume was left running.
- Attachment4: Started removing the rid. This did not change the gause readings as they were isolated from the venting main volume.
- Attachment5: Opened the rid. Took the components out on a UHV foil bag. The rid was replaced but loosely held by a few screws with the old gasket, just to protect the frange and the volume from rough dusts.
I traced a cable from the OMC electrical feedthrough flanges to find the DCPD/OMMT Satellite Box (D060105). I couldn't find the DCC number or mention of the box anywhere except this old elog.
Gautam and I supplied the box with power and tested what we think is the bias for the PD, but don't read any bias... we tracked down the problem to a suspicious cable, labelled.
We confirmed that the board supplies the +5V bias that Rich told us we should supply to the PDs.
We tested the TFs for the board from the PD input pins to output pins with a 100kHz low pass (attached, sorry no phase plots). The TFs look flat as expected. The unfiltered outputs of the board appear bandpassed; we couldn't identify why this was from the circuit diagram but didn't worry too much about it, as we can plan to use the low passed outputs.
In order to power the heater setup to be installed in the ETMY chamber, we took the Sorensen DSC33-33E power supply from the Xend rack which was supposed to power the heater for the seismometer setup.
We modified the J3 connector behind in such a way to allow a remote control (unsoldered pins 9 and 8).
Now pins 9 and 12 need to be connected to a BNC cable running to the EPICS.
RXA update: the Sorensen's have the capability to be controlled by an external current source, voltage source, or resistive load. We have configured it so that 0-5V moves the output from 0-33 V. There is also the possibility to make it a current source and have the output current (rather than voltage) follow the control voltage. This might be useful since out heater resistance is changing with temperature.
I tried to reduce the overfitting problem in previous neural network by reducing the number of nodes and layers and by varying the learning rate, beta factors (exponential decay rates of moving first and second moments) of Nadam optimizer assuming error of 5% is reasonable.
32 * 32 image frames (converted to 1d array & pixel values of 0 to 255 normalized) of simulated video by applying sine signal to move beam spot in pitch with frequency 0.2Hz and at 10 frames per second.
Total: 300 cycles , Train: 60 cycles, Validation: 90 cycles, Test: 150 cycles
4 nodes 1 node
Learning rate = 0.00001, beta_1 = 0.8 (default value in Keras = 0.9), beta_2 = 0.85 (default value in Keras = 0.999)
Plot of predicted output by neural network, applied input signal & residual error given in 1st attachment.
Changed number of nodes in hidden layer from 4 to 8. All other parameters same.
These plots show that when residual error increases basically the output of neural network has a smaller amplitude compared to the applied signal. This kind of training error is unclear to me.
When beta parameters of optimizer is changed farther from 1, error increases.
I went to the Y-end and took more photos of the cable stand. These revealed that in-vac pin #13 is connected to the shield of the cable (P.2). This in-vac pin #13 corresponds to in-air pin #1. So in the end, we bunch the pins in the following order.
After getting the go ahead from Steve, I removed the physical beam block on the PSL table, sent the beam into the IFO, and re-aligned the MC to lock at low power. I've also revived my low power autolocker (running on megatron), seems to work okay though the gains may not be optimal, but it seems to do the job for now. Nominal transmission when well aligned at low power is ~1200cts. I briefly checked Y arm alignment with the green, seems okay, but didn't try locking the Y arm yet. All doors are still on, and I'm closing the PSL shutter again while Keerthana and Sandrine are working near the AS table.
Steve and Aaron,
6 hrs vent is reaching equlibrium to room air. It took 3 and a half instrument grade air cilynders [ AI UZ300 as labelled ] at 10 psi pressure. Average vent speed ~ 2 Torr/min
Valve configuration: IFO at atm and RGA is pumped through VM2 by TP1 maglev.
Steve gave me a venting tutorial. I'll record this in probably a bit more detail than is strictly necessary, so I can keep track of some of the minor details for future reference.
Here is Steve's checklist:
Gautam already did the pre-vent checks, and Steve took a screenshot of the IFO alignment, IMC alignment, master op lev screen, suspension condition, and shutter status to get a reference point. We later added the TT_CONTROL screen. Steve turned off all op levs.
We then went inside to do the mechanical checks
After completing these checks, we grabbed a nitrogen cylinder and hooked it up to the VV1 filter. Steve gave me a rundown of how the vacuum system works. For my own memory, the oil pumps which provide the first level of roughing backstream below 500mtorr, so we typically turn on the turbo pumps (TP) below that level... just in case there is a calibrated leak to keep the pressure above 350mtorr at the oil pumps. TP2 has broken, so during this vent we'll install a manual valve so we can narrow the aperture that TP1 sees at V1 so we can hand off to the turbo at 500mtorr without overwhelming it. When the turbos have the pressure low enough, we open the mag lev pump. Close V1 if things screw up to protect the IFO. This 6" id manual gatevalve will allow us throttle the load on the small turbo while the maglev is taking over the pumping The missmatch in pumping speed is 390/70 l/s [ maglev/varian D70 ] We need to close down the conductive intake of the TP1 with manual gate valve so the 6x smaller turbo does not get overloaded...
We checked CC1, which read 7.2utorr.
Open the medm c0/ce/VacControl_BAK.adl to control the valves.
Steve tells me we are starting from vacuum normal state, but that some things are broken so it doesn't exactly match the state as described. In particular, VA6 is 'moving' because it has been disconnected and permanently closed to avoid pumping on the annulus. During this v ent, we will also keep pumping on the RGA since it is a short vent; steve logged the RGA yesterday.
We began the vent by following the vacuum normal to chamber open procedure.
Everything looks good, so I'm monitoring the vent and swapping out cylinders.
At 12:08pm, the pressure was at 257 torr and I swapped out in a new cylinder.
Steve: Do not overpressurize the vacuum envelope! Stop around 720 Torr and let lab air do the rest. Our bellows are thin walled for seismic isolation.
@SV, we are ready to vent tomorrow. Aaron is supposed to show up ~830am to assist.
[Annalisa, Terra, Koji, Gautam]
Summary: We find a configuration for arm scans which significantly reduces phase noise. We run several arm scans and we were able to resolve several HOM peaks; analysis to come.
As first, we made a measurement with the already established setup and, as Jon already pointed out, we found lots of phase noise. We hypothesized that it could either come from the PLL or from the motion of the optics between the AUX injection point (AS port) and the Y arm.
In this configuration, we were able to do arm scans where the phase variation at each peak was pretty clear and well defined. We took several 10MHz scan, we also zoomed around some specific HOM peak, and we were able to resolve some frequency split.
We add some pictures of the setup and of the scan.
The data are saved in users/OLD/annalisa/Yscans. More analysis and plots will follow tomorrow.
We only anticipate opening up the IOO chamber and the EY chamber.
Vent preparation: see here.
In preparation for tomorrow's vent, I'm checking some of the OMC-related electronics we plan to use.
(well, technically the first up was the Kepco HV power supply... but I quickly tested that its output works up to 300V on a multimeter. The power supply for OMC-L-PZT is all good!)
According to the DCC, the nominal HV supply for this board is 200V; the board itself is printed with "+400V MAX", and the label on the HV supply says it was run at 250V. For now I'm applying 200V. I'm also supplying +-15V from a Tektronix supply.
I used two DB25 breakout boards to look at the pins for the DC and AC voltage monitors (OMC_Vmon_+/-, pins 1/6, and OMC_Vmon_AC+/-, pins 2 and 7) on a scope. I hooked up a DS345 function generator to the piezo drive inputn (pins 1,6). According to the 2013 diagram from the DCC, there is just one drive input, and an alternative "dither in" BNC that can override the DAC drive signal. I leave the alternative dither floating and am just talking to the DAC pins.
Aspects of the system seem to work. For example, I can apply a sine wave at the input, and watch on the AC monitor FFT as I shift the frequency. However, anything I do at DC seems to be filtered out. The DC output is always 150V (as long as 200V comes from the supply). I also notice that the sign of the DC mon is negative (when the Vmon_+ pin is kept high on the scope), even though when I measure the voltage directly with a multimeter the voltage has the expected (+) polarity.
A few things to try:
On further investigation this was the key clue. I had the wrong DCC document, this is an old version of this board, the actual board we are using is version A1 of D060283-x0 (one of the "other files")
Gautam and Koji returned at this point and we started going through the testpoints of the board, before quickly realizing that the DC voltage wasn't making it to the board. Turns out the cable was a "NULL" cable, so indeed the AC wasn't passing. We swapped out the cable, and tested the circuit with 30V from the HV supply to trim the voltage reference at U14. The minimum voltage we could get is 5V, due to the voltage divider to ground made by R39. We confirmed that the board, powered with 200V, can drive a sine wave and the DC and AC mons behave as expected.
We are getting ready to vent.
We calculated the expected power of the beat note for Annalisa's Y arm cavity scans.
Beat Note Measurement
We began by calculating the transmitted power of the PSL and AUX. We assumed that the input power of the PSL was 25 mW and the input power of the AUX was 250 uW. We also assumed a loss of 25 ppm for the ITM and ETM. We used T1 = 0.0138 and T2 = 25 x 10-6.
The transmitted power of the PSL is approximately 100 uW, and the transmitted power of the AUX is approximately 0.974 uW.
The beat note was calculated with the following:
The expected beat note should be approximately 20 uW.
I tried to compare the cavity scan data we get from the Finesse simulation and that we expect from the Analytical solution. The diagram of the cavity I defined in Finesse is given below along with the values of different quantities I used. For the analytical solution I have used two different equations and they are listed below.
Analytical 1 - Blue Graph
Analytical 2 - Red Graph
The graph obtained from both these solutions completely matches with each other.
The cavity which I defined in Finesse is shown below. The solution from Finesse and the Analytical solution also matches with each other. Another plot is made by taking the difference between Finesse solution and Analytical solution. The difference seems to be of the order of .
The Difference plot is also attached below.
We found a diagram describing the DC Readout wiring scheme on the wiki page for DC readout (THIS DIAGRAM LIED TO US). The wiring scheme is in D060096 on the old DCC.
Following this scheme for the OMC PZT Driver, we measured the capacitance across pins 1 and 14 on the driver end of the cable nominally going to the PZT (so we measured the capacitance of the cable and PZT) at 0.5nF. Gautam thought this seemed a bit low, and indeed a back of the envelope calculation says that the cable capacitance is enough to explain this entire capacitance.
Gautam has gone in to open up the HV driver box and check that the pinout diagram was correct. We could identify the PZT from Gautam's photos from vent 79, but couldn't tell if the wires were connected, so this may be something to check during the vent.
Turns out the output was pins 13 and 25, we measured the capacitance again and got 209nF, which makes a lot more sense.
Aaron and I are going to do the checkout of the OMC electronics outside vacuum today. At some point, we will also want to run a c1omc model to integrate with rtcds. Barring objections, I will set up this model on one of the spare cores on the physical machine c1ioo tomorrow.
From the Measurement Jon made, FSR is 3.967 MHz and the Gouy phase is 52 degrees. From this, the length of the Y-arm cavity seems to be 37.78 m and the radius of curvature of the mirror seems to be 60.85 m.
FSR = Free spectral Range
L = Lenth of the arm
R = Radius of curvature of the mirror (R1 = , R2= unknown)
This note reports analysis of cavity scans made by directly sweeping the AUX laser carrier frequency (no sidebands). The measurement is made by sweeping the RF offset of the AUX-PSL phase-locked loop and demodulating the cavity reflection/transmission signal at the offset frequency.
Due to the simplicity of its expected response, the Y-arm cavity was scanned first as a test of the AUX hardware and the sensitivity of the technique. Attachment 1 shows the measured cavity transmission with respect to RF drive signal.
The AUX laser launch setup is capable of injecting up to 9.3 mW into the AS port. This high-power measurement is shown by the black trace. The same measurement is repeated for a realistic SQZ injection power, 70 uW, indicated by the red curve. At low power, the technique still clearly resolves the FSR and six HOM resonances. From the identified mode resonance frequencies the following cavity parameters are directly extracted.
I had created a python code to find the combination of hyperparameters that trains the neural network. The code (nn_hyperparam_opt.py) is present in the github repo. It's running in cluster since a few days. In the meanwhile I had just tried some combination of hyperparameters.
These give a low loss value of approximately 1e-5 but there is a large error bar for loss value since it fluctuates a lot even after 1500 epochs. This is unclear.
Input: 64*64 image frames of simulated video by applying beam motion sine wave of frequency 0.2Hz and at 10 frames per sec. This input data is given as an hdf5 file.
Train : 100 cycles, Test: 300 cycles, Optimizer = Nadam (learning rate = 0.001)
256 -> 128 -> 1
Activation : selu selu linear
Case 1: batch size = 48, epochs = 1000, loss function = mean squared error
Plots of output predicted by neural network (NN) & input signal has been shown in 1st graph & variation in loss value with epochs in 2nd graph.
Case 2: batch size = 32, epochs = 1500, loss function = mean squared logarithmic error
Plots of output predicted by neural network (NN) & input signal has been shown in 3rd graph & variation in loss value with epochs in 4th graph.
I started this document on my own with notes as I was tracing the beam path through the output optics, as well as some notes as I started digging through the elogs. Let's just put it here instead....
Notes during reading elog
As of at least Nov 2009, the .par file for the OMC was located at /cvs/cds/gds/param/tpchn_C2 (see elog 2316)
Need to check:
Today both the heater and the reflector were delivered, and we set down the setup to make some first test.
The schematic is the usual: the rod heater (30mm long, 3.8 mm diameter) is set inside the elliptical reflector, as close as possible to the first focus. In the second focus we put the power meter in order to measure the radiated power. The broadband power meter wavelength calibration has been set at 4µm: indeed, the heater emits all over the spectrum with the Black Body radiation distribution, and the broadband power meter measures all of them, but only starting from 4µm they will be actually absorbed my the mirror, that's why that calibration was chosen.
We measured the cold resistance of the heater, and it was about 3.5 Ohm. The heater was powered with the BK precision DC power supply 1735, and we took measurements at different input current.
We also aimed at measuring the heater temperature at each step, but the Fluke thermal camera is sensitive up to 300°C and also the FLIR seems to have a very limited temperature range (150°C?). We thought about using a thermocouple, but we tested its response and it seems definitely too slow.
Some pictures of the setup are shown in figures 1 and 6.
Then we put an absorbing screen in the suspension mount to see the heat pattern, in such a way to get an idea of the heat spot position and size on the ETMY. (figure 2)
The projected pattern is shown in figures 3-4-5
The optimal position of the heater which minimizes the heat beam spot seems when the heater inserted by 2/3 in the reflector (1/3 out). However, this is just a qualitative evaluation.
Finally, two more pictures showing the DB connector on the flange and the in-vacuum cables.
I copied the netgpibdata folder onto rossa (under the directory ~/Agilent/), which contains all the necessary scripts and templates you'll need to remotely set up, run, and download the results of measurements taken on the AG4395A network analyzer. The computer will communicate with the network analyzer through the GPIB device (plugged into the back of the Agilent, and whose communication protocol is found in the AG4395A.py file in the directory ~/Agilent/netgpibdata/).
The parameter template file you'll be concerned with is TFAG4395Atemplate.yml (again, under ~/Agilent/netgpibdata/), which you can edit to fit your measurement needs. (The parameters you can change are all helpfully commented, so it's pretty straightforward to use! Note: this template file should remain in the same directory as AGmeasure, which is the executable python script you'll be using). Then, to actually set up, run, and download your measurement, you'll want to navigate to the ~/Agilent/netgpibdata/ directory, where you can run on the command line the following: python AGmeasure TFAG4395Atemplate.yml
The above command will run the measurement defined in your template file and then save a .txt file of your measured data points to the directory specified in your parameters. If you set up the template file such that the data is also plotted and saved after the measurement, a .pdf of the plot will be saved along with your .txt file.
Now if you want to just download the data currently on the instrument display, you can run: python AGmeasure -i 192.168.113.105 -a 10 --getdata
Those are the big points, but you can also run python AGmeasure --help to learn about all the other functions of AGmeasure (alternatively, you can read through the actual python script).
Happy remote measuring! :)
Bulb replaced at day 110 We have now spare now.