Krishna, Sheila, Hugh, Fabrice:
We have been chasing large amplifications at low frequencies (in the range of 10mHz to 30mHz) caused by the Z zensor correction of HEPI, which is necessary to reduce the Z to RZ coupling on Stage 1. It looks like the Z HEPI inertial isolation is causing rotations (RX, RY), that are causing tilt signal in the Stage 1 horizontal seismometers, that couple to X and Y as we blend at 45 mHZ, and then shows up into the cavity signal.
The problem was mostly visible on the BS unit. We convinced ourself that Z to tilt was the problem by moving Stage 1 in high blend, which very significantly reduced the Mich amplification around 20 mHz (which exist only when the Z sensor correction is ON)
It seems that the excessive Z to tilt coupling in the BS was caused by off centered vertical position sensors (up to 24000counts). We recentered them by applying a HEPI vertical force. The Z to Mich coupling is now much lower. So I guess that the gain of the sensors was affected by the large offsets and thus creating excessive Z to RX and RY couplings.
Comparison with high blend configurations show that there is probably room to further reduce this coupling. We need the measure the Z to RX and RY coupling and apply corrections.
Dave, Jim
To perform some further HWWD testing at ETMY, we have temporarily connected the HWWD RMS analog monitor (BNC output on rear of unit) to a spare PEM channel on its AA chassis. We are reading the channel using the slow IOP DAQ channel. The plot shows 1 minute of full 16Hz data during the recent ETMY in-chamber work. The 1.6 second integration time can be easily seen in the data.
We are using the next to last channel on the PEM AA chassis, this is ADC3, Ch30 (counting from zero).
Recent in-chamber work on ETMY has provided a OSEM PD RMS trip level comparison between the Hardware watchdog (HWWD) and the IOP software watchdog (SWWD).
We have verified that the two systems have very similar trip levels. The SWWD trip level is hardcoded at 110mV. The HWWD level can be modified, but it is running at its default of 110mV.
The plot shows: the PD RMS status for the SWWD (Ch 4, GOOD=1, BAD=0); and the HWWD (Ch 2, GOOD=0, BAD=1). Also shown is the SWWD RMS monitor for the ETMY top stage OSEM, SIDE channel, which is the channel driving the RMS errors.
As can be seen, Ch4 and Ch 2 match each other very well.
Note that the maximum time in error in this plot is only 3 minutes. If the excitation were to extend to 10 minutes the SWWD would have tripped the SEI DACs, and if it were to extend to 20 minutes the HWWD would power down the ISI coil drivers.
Approved work to be done:
After Chamber Vent Following E1400430
1: Disconnect external (air side) ESD cable from feed thru.
2: Install flat surface with Vacuum Foil on it near Flange for a work
surface.
3: Remove ESD feed thru flange Gerardo
4: IF enough cable exist cut the cables from the connector on the inside.
IF not enough cable disconnect cable
disassemble connector then remove connector ends.
5: Crimp new pin on to cables.
6: Verify with SUS which pin is connected to which part of ESD.
7: Make up connector with new pins in proper order D1400177 V2
8: Secure new connector to new feed thru.
9: Install Flange on Vacuum chamber. Gerardo
10: Connect new external cabling to feed thru
11: Ensure no one is near optic!!! High Pot connector ~ 700V DC If it
doesn't pass. Stop and investigate.
12: Make up external cables to field current limiting resistor box.
The last step is at the time and discretion of the Vacuum crew.
Decision flow diagram from E1400472, drawing by D. Coyne
Took dust monitor trends from End-X ahead of the vent. Posted are a 7 day trend for both location #1 and #2 and a 72 hour trend. The first cleaning and cleanroom over BSC9 were turned on 12/16/14.
Here is an estimate of the BRDF for ETMx and ETMy from the recent scattering images we took.
BRDF | Ps (W) | Pinc(W) | |
ETMy | 2.3e-02 | 8.4e-07 | 21 |
ETMx | 2.5e-03 | 8.0e-08 | 19 |
BRDF=Ps/solidAngle/Pinc/cos(theta_s), in W/steradian , as quoted by Hunter Rew in P1400197. Ps is power incident on photodiode, solidAngle is that subtended by the photodiode, Pinc=incident power (here the IR power inside the arm), and theta_s is the scattering angle, which is the angle of the photodiode from normal incidence. SolidAngle-A/L where A is the area of the photo diode and L is its distance from the test mass.
Ps=calibration_factor*Intensity/Exposure, where calibration factor is in (W*microseconds/counts) and is defined by David Feldbaum in LLO alog 12627, intensity is the total pixel counts on the CCD, and exposure is the camera exposure in microseconds.
Pinc is calculated using Dan's calibration of the TMS photodiodes outlined in alog 15431.
A is as given by the camera specs = 1.0209e-05 m^2.
Rick Savage and Joe Gleason helped me get the distance and scattering angle of the camera from the mirror. L is 6.0452m and theta_s is 4 deg for the ETMS. .
The other paramters used are:
UTC Measurement time | Exposure (microsec) | L ((m) | theta_s (deg) | |
ETMy | 14-12-11-03-00-00 | 10000 | 6.0452 | 1 |
ETMx | 14-12-12-02-00-00 | 500000 | 6.0452 | 1 |
Because these images are taken at small angles, its not valid to assume the BRDF is constant, so its not meaningful to integrate the BRDF to get an estimate on total scatter in ppm. I wanted to compare numbers to the ITMs, however I can't because the analogue gain setting was different (1023 not 100) for the ITM pictures. Once ETMy is clean and we have the Y-arm back I will re-measure ETMy and also add numbers for the ITMs.
ITM03 - had a full sheet of FC applied with NO IAS windows, see attached. ITM03 was installed in BSC3.
Thomas V, Elli
The ITMy and ITMx Hartmann wavefront sensors are now aligned to their respective SLEDS.
We took the HWS plates out and centered the SLED beam on the camera using the periscope mirrors and the final mirror before the HWS (pictures of the final alignment on the camera are included).
The ITMx image is a lot dimmer than the ITMy image, however the H1:TCS-ITMX_HWS_SLEDPOWERMON said 5mW whereas H1:TCS-ITMY_HWS_SLEDPOWERMON says 1.8 mW. We were watching to maximise image brightness as we made the adjustments so I don't know why ITMx image is so much dimmer.
This morning, travis and I pulled the FC sheet that we applied to the ETMy yesterday. The patchy smudge mark and the large gooeyish particle that we induced yesterday were apparently picked up by this FC sheet. While there, we tested using the cotton tipped Q-tip on a portion of the ring around the edge of the optic (~1Inch in from bevel edge near the bottom). We used both DI water and also acetone, both chased with ~5 spot drag wiped of acetone using the pressure technique. Upon evaluating neither q-tip+ acetone drags seemed to improve the area since both swipes added to the streaking which the chasing acetone could not fully remove with 5 wipes.
Our cleaning efforts at EndY have focused on the three larger scattering objects on the surace (which were apparently removed) and the ~3" diameter "stain" associated with the thinner central window in a previous layer of First Contact (which we have not been very successful in removing. ) The composite image below includes six images. All were taken by illuminating the lower edge of the central "stain" feature with the green LED flashlight. The two on the left were taken before applying First Contact (note the large scattering objects are still on the surface). The middle two were taken after the first application and removal of First Contact. The two on the right were taken today, after the second application and removal of First Contact. It doesn't appear that we have significantly improved the "stain" feature.
End-Y: The faulty ETM-Y ESD connector was repaired (see aLOG #15656). The first cleaning of the ETM-Y optic with First Contact was completed with some success. There are still several areas of concern left on the optic. A second First Contact application will be removed today. End-X: Preparations (cleaning, cleanrooms, garb, etc.) for entry into End-X are underway in case it is decided to clean the End-X optics. HAM1: The work in HAM1 is finished. The door will be reinstalled today. EE: PEM cabling work in the LVEA. Seismic: Hugh adding CPS ground straps to the HAMs in the LVEA.
RF Input Power | Peak Height measured with OSA relative to noise floor | Modulation Depth | |
45 MHz | 21.5dBm | 79.2mV | 0.284 |
9 MHz | 17.27dBm | 41.2mV * | 0.205 |
RFAM | RFAM/PM | |
45 MHz | 1.146e-4 | 4.0199e-4 |
9 MHz | 1.486e-3 | 7.233e-3 |
I annotated the scope screenshot from alog 15639 with carrier and sideband frequencies. Just reading off the plot we have
Frequency | Amplitude | Unit | Ratio | Γ |
---|---|---|---|---|
Carrier | –22.0 | dBm | 1 | — |
–9 MHz sideband | –41.0 | dBm | 0.112 | 0.226 |
+9 MHz sideband | –41.5 | dBm | 0.106 | 0.213 |
–45 MHz sideband | –39.0 | dBm | 0.141 | 0.285 |
+45 MHz sideband | –39.5 | dBm | 0.133 | 0.269 |
Taking the average and assuming a 0.5 dB reading error we have Γ = 0.219(12) for the 9 MHz and Γ = 0.277(16) for the 45 MHz sidebands, respectively.
Paul, Mackenzie, Alexa
We measured the RFAM again; this time after the MC (with the PRM misaligned). Paul and Mackenzie had already aligned the beam onto the AC coupled 1811 NewFocus PD on IOT2R (same specs as before). We blocked the auxiliary laser. Again we used a TPS 30324 to measure the DC signal at high impedance, and the Agilent 43958 Spectrum Analyzer to measure the AC signal. We measured 210mV at DC, and
RFAM | RFAM/PM | |
45 MHz | 2.245e-4 | 6.762e-4 |
9 MHz | 1.121e-3 | 5.63e-3 |
The PM represents the modulation depth. We repeated Daniel's calculation but wanted to collect the numbers with the Spectrum Analyzer to reduce the error. So again, we had the aux beam phase locked to the beam on IOT2R. We measured the following:
Freq | Amplitude | Ratio | Mod Depth | |
Beat | 60 MHz | -22.8dBm | 1 | |
+ 45 MHz | 105 MHz | -37.96dBm | 0.174 | 0.348 |
+ 9 MHz | 69 MHz | -43.2dBm | 0.095 | 0.19 |
-9 MHz | 51 MHz | -42.50dBm | 0.104 | 0.208 |
-45 MHz | 14.5 MHz | -38.82dBm | 0.158 | 0.316 |
Taking the average we find. Gamma_9MHz = 0.199 and Gamma_45MHz = 0.332. This seems a bit high given our previous two measurements. I have used RFAM = 2 * (Vsb/Vcar). There is no sqrt this time because we are measuring the amplitudes; whereas the OSA measured the power. The PD has a BW of 200 MHz, but maybe we are approacing the roll-off. We tried locking the beatnote at a lower frequency so that we would not need to worry about the PD's roll off; however, we had trouble getting a clean lock even at 20 MHz which seemed to be fine yesterday. Paul will attach the raw data.
..and here are the data. It seems like we can even pick out sidebands of sidebands with this measurement technique.
Betsy, Travis, Rick
How today went:
We spent a few hours this morning working on the in-vac side of the new ESD connector termination checkouts.
At ~2pm we pulled the FC sheet off of the ETMy-HR (painted on last night). We immediately saw that the 3 larger FC remnants that were causing all of the glint in the cavity arm were removed. Yay! However, it was also immediately clear (with the green flashlight) that the 3" ring feature and mottled haze were still there.
We took a break at 3pm and spoke to a wider SYS+ audience where it was determined we would attempt to do a test drag wipe on the mottling nearer the edge of the optic. We worked on getting pictures of numerous mottled areas in the haze of the donut shape on the ETMy-HR before doing some test drag wipes. Then we chose an area toward the top of the optic where there was another relic IAS window circle print near the 12 o'clock position to test drag wiping. After numerous failed attempts at performing the light-duty friction-only drag wiping technique on this area I resorted to folding the lens wipe a few times and applying pressure during the drag wipe. This made a smudge where I wiped which I then had to spend another ~5 wipes removing. We then reevaluated the mottling and found that we might have improved it in one small place, but not the entire area I had been working on. We then redirected a streaky area at the bottom of the optic to see if I could get a better technique down. No such luck and I again had to resort to applying pressure numerous times in order to see an improvement. Since we had better pictures of the 12 o'clock position mottling area that I had been working on earlier, we decided to revisit that area. I again made an attempt at friction-only drag wiping of this top area. In the process I must have lightly swiped the optic with my glove because when inspecting the optic after the drag wipe, we found a ~2mmx2mm patch of particulate just outside of the 3" ring near the center of the optic. Brilliant. We also found more particulate on the optic from the waving lens tissues. We tried to blow them off with a few minutes of N2. This did not seem to work. One of the particles was quite large and even a light dab with an swab did not move it (in fact, ~5 attempts to snatch it off via dabbing failed). Rick captured a few pictures of the "new" features.
At this point we aborted the drag wipe testing and carefully repainted FC back on. I was very careful to not brush across the large particle nor the "patch" area very much. I applied very thick ~1 inch long strokes which at first were more like dabs near these areas. (All other saturated sloppy strokes were ~1-2" in length, repeated.)
I do not think we can drag wipe the full surface (or even the central ~8") of the ETMy-HR. We were not sucessful in ~25 attempts to get a good pull on the optic except for in the localized areas I mentioned above when I used finger pressue. And after these attempts we added contaminant to the surface by accident.
The optic size obviously makes it hard to work with. The working area is too small for 2 hands, elbows, flashlights, your head, etc. Then, the task is too hampered by suspension brackets and braces to get good surface tension with the flat lens tissue. The wipe wants to continue to pull off the surface and ripples easily. As well (or worse), we had a hard time getting the wetted wipe to the surface before the acetone dried.
Rick plans to attach some pictures to this alog so check back later or tommorrow.
Some images to supplement Betsy's narrative. Hopefully the file names are sufficiently descriptive. For visualizing some of the features, zooming helps a lot. Look in the elliptical area illuminated by the green flashlight beam. Filenames, in order, are: LHOyTopLineBeforeCleaning176.jpg LHOyTopLineBeforeCleaning185.jpg LHOyBirdsHeadBrightEdge180.jpg LHOyDarkBirdsBeak181.jpg LHOyLongVerticalBrightPatch182.jpg LHOyGloveTouchPatch186.jpg LHOyTopLineAfterCleaning190.jpg
[Mackenzie, Paul]
Yesterday we had trouble getting a large enough beat signal between the aux laser and the main PSL. After discussions with Dave O., Stefan and Daniel, we were convinced that the aux laser was not operating monochromatically. This can apparently occur if the aux laser is close to a "mode hop" region when we bring its frequency close to the PSL frequency using the temperature control. We then end up observing the beat frequency between the PSL and one of the "parasitic" modes of the aux laser, which has a much smaller amplitude than the main mode.
On Daniel's advice, we adjusted the diode input current as well as temperature, in order to give us an additional frequency control option that affected the proximity to the mode hopping region differently to the temperature. After some searching around, we suddenly observed a forest of peaks, with one especially large peak.
This forest appears to have been a combination of a) saturation of the 1811 AC output exciting harmonics of the base beat frequency and b) contributions from RF sidebands of the PSL light. We reduced the power on the 1811 using the filter wheel (we ended up using the OD 2.5 filter), resulting in the spectrum shown in the attached figure. The VCO offset frequency was 20MHz. Interestingly, the beat between aux laser and 9, 45MHz sidebands are quite visible.
The dominant beat frequency now registers a -22dBm signal at the AC PD output on the spectrum analyzer, even with the power on the PD attenuated such that the DC output from the PD produced by the PSL beam is less than 1mV (the PSL beam is the weaker of the two beams at the PD). This may still seem a little weak, but it is much better than yesterday, when we had ~-50dBm AC on a ~100mV DC signal.
After some adjustment of the servo box gains, we were able to achieve stable PLL locks. There is no slow temperature servo path included yet, so the PZT on the aux laser will go out of range rapidly if the temperature dial is not adjusted by hand. By adjusting the temperature dial, it was possible to keep the PLL locked for tens of minutes.
Next, we switched the frequency reference for the PLL offset from the Marconi VCO to the swept sine output of the network analyzer. This presents a new challenge, because now not only does the aux laser have to phase lock to the PSL, but it also has to track the offset frequency generated by the NA. After playing around with IF BW, sweep ranges, source power etc., we were eventually able to get the aux laser to reliably phase lock to the PSL and track the NA frequency through a ~2MHz frequency range (almost enough for a full PRC sweep). We convinced ourselves that the PLL was locked and tracking the NA frequency by observing the TF from NA drive signal to 1811 AC output. When the PLL is locked, the TF is flat, high, and smooth, since the 1811 AC output is coherent with the NA signal at the same frequency. When the PLL is unlocked, the same TF is many decades smaller.
Each time the NA reaches the end of a sweep and flips back to the start frequency, the lock drops. I'm not sure there's much we can do about that. It would help to have the slow path enabled though, in order to compensate for slow drifts of the aux laser frequency away from the PSL frequency.
Next we plan to try some PRC sweeps if there is PRMI locked time available. For this, we will take the TF from NA drive signal to one of the REFL AIR broadband detectors, with the PRMI locked and the PLL locked. Depending on how that goes, we may pursue the slow temperature feedback development.
With a subcarrier you can measure the strength of each individual RF sideband---with upper and lower sidebands separated around the VCO frequency. The ratio you see here between RF sideband amplitude and carrier is a direct measurement of the modulation depth.
Great point, what a convenient way to measure the modulation depth! It's also probably a lot more precise that the OSA.
When we get the PLL going again later today I'll save the data from the spectrum analyzer so we can do a more careful analysis.
The plot attached shows the Mich Out signal:
- in the first box, HEPI Z sensor correction is ON, Stage 1 X abd Y are in low blend, the IPS are off centered. The low frequency amplication is huge.
- in the second box, HEPI Z sensor correction is ON, Stage 1 X abd Y are in high blend, the IPS are still off centered. The low frequency amplication is gone.
- in the third box, HEPI Z sensor correction is ON, Stage 1 X abd Y are in low blend, the IPS are re-centered. Our current understanding/aseumption is that the Z to RX and RY coupling on HEPI has been well reduced.
The latest configuration is likely the best compromize:
- good micro-seism atenuation thanks to the low blend on X and Y
- low vertical to pitch coupling thanks to the Z feedback
- little RZ amplification at the micro-seism thanks to the Z sensor correction to HEPI (that offloads Stage 1 Z drive at the micro-seism)
- amplification acceptable at very low frequency, now that the IPS have been re-centered. We'll try to further improve it.
An ASD plot of the MICH_OUT channel is attached under different configurations. The first (RED) is with no sensor correction on BS, ITMX and ITMY. The second (BLUE) is with X, Y sensor correction signals to all three BSCs. The third (GREEN) is with Z sensor correction to HEPI for the BS chamber, showing the large low frequency amplification. The fourth (BROWN) shows the MICH_OUT with the IPS recentered and same configuration as the third. Tilt-decoupling on HEPI ought to reduce the amplification further.
Z sensor correction has been turned OFF on BS and ITMY. X and Y sensor corrections seem to be working fine and can be left ON.
Could be that I'm missing something but it sounds to me like at least one of the IPS is not working properly (ie broken). They are supposed to be linear to within 0.1% over the full range (+/- 0.05 inches)