Aaron's past experience at Cryo_Lab/2563 is familiar to other LIGO users of the Lista tool chests. Occaisionally the mechanism which keeps the remaining drawers closed while a single drawer is open can lock all drawers. Dean brought to my attention that the Downs 228 chest was apparently locked. The key was nowhere to be found. Here's how we solved the problem:
The key was found in the upper drawer and placed atop the chest.
Ref. D1400331-v8 Item 48 - new high voltage feedthrough D2000585 protrudes along axis and requires protection.
Constructed simple connector guard for a 6" conflat on TMDS, with the following materials:
- Stainless 304 sheet, 4" x 36" cut to length (a little long since I initially thought the conflat was 6.5" OD, but intended length is 18.7" for the 18.8" diameter) - p/n McMaster 1421T63
- Stainless worm drive clamp - p/n McMaster 5682K22
Ref. T2100351 instructions, plus email exchange indicating available media sizes.
Ref. FRS Ticket 20562 for shipment details.
Simple machining operation setup and executed in Downs 228. Taking notes for future replication, and to drop the Photo Album somewhere useful. Also thought this log would be helpful to communicate that we are able to quickly knock out simple mounts and mods like this in house.
- Used Rotary Table for uniform circular cutting path; bolted down with 1/2" T-nuts and translated mill table to center ID of Rotary Table with edge finder.
--> 8" diameter Rotaty Table was a little too big for a standard end mill as the mill spindle is only offset from the turret by about 5 inches. There was not enough travel range in the direction toward the turret to reach the 40 mm radius from the Rotaty Table center. Could have cut in a different direction, but I didn't want to be caught later so I shifted to a larger radius cutter. While I was figuring this out, I managed to shear off one of the handles for locking down the Rotary Table during cutting - oops!
--> Originally intended cutter was a .5" end mill, but I shifted to a 90 degree face mill for the extended radius given the above issue. I didn't feel too guilty about cutting the corners with the face mill, as the optic will have bevels and the acetal/delrin was soft.
- I attempted to measure the 75mm diameter bore, and found the radius was larger by ~1mm. I then attempted to make a minimum 81mm diameter bore to host the 80mm optic, and overshot by a bit ( < 1mm ).
- I elected to go for a greedy cut, creating a counterbore of the 75mm bore at an intermediate bore depth. This was an attempt to retain the 75mm bore function along with the new 80mm bore, and to fix either bore with the existing set screw. Liyuan verified that the new 80mm bore still behaves. So far, so good.
Imaging effort of coupons collected from LMA coating chamber panels.
uncoated_A_01 = area where coating delaminated during coupon cutting, near angled edge.
SN1535 was annealed between 07 and 08 September 2021 - used the large furnace in Downs 221, atmosphere was air, used glass petri dishes to protect optic (from Gabriele and CRiMe).
Ramp up rate = 100 °C per hour
Hold temp = 300 °C
Hold time = 10 hours
Ramp down rate = 100 °C per hour
Witness RTD observed a ~21 °C overshoot and ~14 °C offset from controller RTD. These parameters were consistent with the elog entry ENG_Labs/253. See attached image of the temperature profile (.xlsx file is source document).
Optic is now with Liyuan in RTS for recharacterization after anneal.
SN0932 was annealed between 02 and 03 September 2021 - used the large furnace in Downs 221, atmosphere was air, used glass petri dishes to protect optic (from Gabriele and CRiMe).
Witness RTD observed a ~22 °C overshoot and ~14 °C offset from controller RTD. See attached image (.xlsx file is source document).
Luis, Chub, Calum, Jordan, Dean, Rich
Today we filled neon into the new enclosure (Dxxx) that's destined for use with the A+ filter cavity length and alignment detector, plus the new DCPD preamplifiers. The goal is to do a neon accumulation test over in the 40m lab.
Here is some related information:
We are chalking our inconclusive results up to experience, and starting again tomorrow with a fresh gasket. We will be sure to account for the boyancy of neon in our fill method, and to rig a better way of flushing the interior of the enclosure with neon.
StephenA, 20 April 2021
Finished integrating the RTS Microscope Mount Assembly for the Keyence VH-Z250T lens (D2000085 WIP).
Overview of procedure for assembling the mount:
Overview of procedure for installing the microscope:
Images of mount in various states:
Regina's problem statement:
I attached examples of two measurements I got and wanted to know if these look reasonable. I took 6 measurements total, and I attached the first and last measurements. The graph on the last page is a picture of the weighting step for reference.
I didn't see significant ring down in all my measurements. Is this to be expected? I thought since the baffle is now rigidly mounted, the vibration should go almost to zero. I was also getting some low amplitude noise throughout the entire frequency domain that didn't show up in the first couple measurements I took. I tried to reset the vibrometer like you mentioned but they were still present. Is this a problem?
1) Vibrometer may need a higher diffusivity surface to improve signal level.
Replying to your absence of ring down in your measurements - I agree that it looks like the vibrometer output is not behaving well, for one reason or another. In the freely suspended case, I was thinking this was due to large yaw and pitch motion causing high signal variation. Given that the symptom occurs when the baffle is fixed, I think the likeliest reason is the low signal, due to the low scatter and highly specular surface finish of this baffle (aka shiny). One way to troubleshoot would be to attach a compliant, diffusely reflecting material to the surface - think a small square of the adhesive-coated part of a Post-It note, for example - then tune the focus of the vibrometer and see if the vibrometer's signal level bar improves. If the signal level improves, take a hammer-excited measurement, and see if you see any ring down. If this behaves as you might expect, you could generate mode shape data with your excitation roving around the surface of the baffle while your response is fixed - just one baffle point would need your diffuse Post-It square (I might go with a central, or near central, location).
2) Test Article is not well understood, so try measuring something that has been characterized before.
If you try playing with the diffusivity and focus but the signal level doesn't improve, or if you don't see any ring down still, try pointing the vibrometerat the suspension cage and exciting the suspension with the hammer - that should give some real signal regardless of the precise setup, and if that gives similarly mystifying results, let me know and we can think a little harder about what might be going on. I feel pretty confident that between these two tests, you will find an answer.
3) Add another transducer (ie. a witness accelerometer) for comparison.
Another way to support your understanding of your setup (and a good practice) would be to mount the accelerometer to the suspension cage, adjacent to the baffle mounting brackets, or even to the mounting brackets themselves. This accelerometer would supply a witness to the low frequency resonances of the cage, which you may excite during your measurements, and might also provide some insights to the baffle panel resonances (rigid coupling with lower modal mass = smaller vibrations, but likely still above the noise floor of the accel) supporting your eventual successful vibrometer measurements.
4) Notes about mounting an accelerometer.
Mounting would involve collecting a ~[1mm x 1mm x 1mm] chunk of beeswax, spreading the beeswax onto the face of the accelerometer opposite the cable (I like to press it with the outer, flat-ish surface of my thumb nail to spread), and pressing the beeswaxed face onto a flat surface - think a 5 second push with all of your arm strength, which should create a thin layer with plenty of tackiness to hold the accel in any orientation. If it doesn't seem like it could hold for a day, then you might need more beeswax, or more force to create that thin layer. Note: the main way that an accel can go from useful to not is to experience a shock event, so I would recommend that you use some kapton tape to affix the accelerometer cable to the suspension cage - this will strain relieve your cabling and provide a fall restraint, and the potential for the cabling to influence the measurement is minimal here because it is remote from the baffle.
I didn't observe frequency drifts during the month of assembly and monitoring in the Optics Lab. This is not expected from our experiments in the Modal Lab, but it makes the preparation for the sites BRDs easier.
BOUNCE: In anticipation to the frequency drifts, I had tuned the BRDs on the lower side of the target frequency. But the tuning didn't drift so I changed the masses last week for the bounce mode in order to be in the 1% target. The jumps in the curves below are due to retuning (on November 5th for BRD1, on November 20th for the other ones).
Roll: I didn't retune the Roll modes after assembly on October 20th. In the last days, I was experimenting with different ways to excite them (see pictures attached), so this is probably the cause of the slight drifts that we see.
This morning I handed off the parts for the 5 LHO BRDs to Bob Cottingham at LLO for Clean&Bake (https://ics.ligo-la.caltech.edu/JIRA/browse/clean-10214)
The parts for each BRD are in separate containers (see pictures attached).
Here are the masses of each component in order to reassemble the BRDs after C&B:
Stephen A, 2020 Nov 19
Ordered and received PO 75-S492380 including threaded rod and nut with 2"-12 thread. These items are under consideration for A+ FC Tube Support Stand, particularly for use in weldment D2000445. Some observations:
Also ordered were two McMaster offerings which could be used for rust inhibiting conversion coatings (formulations appear to be based on phosphoric acid - more could be learned by investigating specific products). Potential workflow for experimenting:
This experiment would allow us to learn about how the conversion coatings may work and behave.
See attached photos and video for more insights.
The 5 BS BRDs for installation at LLO are now assembled and tuned in the LLO Optics Lab. The target frequencies are 16.846 Hz and 24.672 Hz (measurements from A. Effler, I will post an alog in the LLO logbook about it).
As per the LHO ones, I tuned them at lower frequencies (-1 or -2%) in order to anticipate the frequency drifts.
For the LHO & LLO bounce modes and for LLO roll modes, I had to put two copper blocks per end at the blade tips in order to reach the target frequency.
There are several factors to take into account:
* Extrapolated from the k CIT experimental k.
I would have expected smaller variations in the spring constant.
The resulting masses are:
Even if the mass kit had the expected weight, I would have not been able to build the BRDs with the masses from the kit (or use several washers of 0.1g).
The EDTpdv framegrabber is unaccessible following a security update to the HWS computer. The lspci command shows the EDT frame grabber card as a PCI device but the initcam code no longer can connect to it.
HWS code was backed up to GIT.
Tried updating EDT software to fix this but didn't work. After exhausting options, I'm going to reinstall the OS and try installing the EDT drivers again. Going to try to install Ubuntu 20.04 first (currently running 14.04).
The 5 BS BRDs for installation at LHO are now assembled and tuned in the LLO Optics Lab. The target frequencies are 17.79 Hz and 26.06 Hz (ref alog 49643).
The masses I had to use for the Bounce mode are slightly heavier than I was expecting (above 10.10g instead of 9.90g). The masses for the Roll mode are in target. I deliberately tuned the BRDs to lower frequencies (between -1% and -2% below the target) to anticipate the drift that shifts the frequencies to higher values. After 2 days, the drifts seem lower than was previoulsy observed at CIT (about 0.1% per day compared to 0.5% per day) with some of the BRDs exhibiting no drift at all. That might be good news? Monitoring will continue over the next weeks.
The Q values fluctuate from measurement to measurement, but they are essentially close to 100.
I am using the LLO B&K vibrometer kit, with an older version of the software that Stuart help me to setup, to measure the resonant frequencies. The air flow from the flow bench is currently off.
Basis structural analysis was completed on models of the V5, V6, and a simple rectangular version of the BS BRD blades. The modesl were simplified to remove the glue layers. The models include stress analysis for individual layers of the blades, stress analysis for the entire blade, and a linearized stress analysis to examine stress discontinuties within the cross-section of the blade. The stresses analyzed include equivalent, maximum principle, and normal stress. The total deformation was also examined in the models.
There is also attached an excel summary of the results of the models.
The initial resuts of these models suggest that there is increased structural weakness in the V6 blade as opposed to the V5 blade. However, the results do not match with theorhetical stress analysis for the blades, so further analysis must be completed to refine the accuracy of the models.
ï»¿<?xml version="1.0" encoding="utf-8"?>
<external-version-string valType="String">2019 R1</external-version-string>
<last-saved-utc valType="String">05/18/2020 15:28:54</last-saved-utc>
<managed-categories valType="String">[report, misc, progress]</managed-categories>
ï»¿<?xml version="1.0" encoding="utf-8"?>
<external-version-string valType="String">2019 R1</external-version-string>
<last-saved-utc valType="String">05/18/2020 15:41:06</last-saved-utc>
<managed-categories valType="String">[report, misc, progress]</managed-categories>
Right after tuning the BRDv5 9 and 10 (see elog 238), I installed them on the BS and measured the BS transfer function.
In figure 1, there are two consecutive measurements of the roll transfer function. The measurements are very close in time (~1 hour). However, we can see a large difference in the measured Q (lower than a factor 2). It is due to a change of settings in the power spectrum analyzer. First measurement is with auto range OFF and second measurement is with autorange ON. We can see a spike at 24.64 Hz in the blue curve. That corresponds to the moment I changed the settings. Edit: Rich A. just explained to me that the "autorange ON" setting isadjusting the gain as a function of the input voltage it is seeing. THerefore it must be used at the beginning of the experiment to select the gain and chek the noise floor and be turned OFF during measurements.
However, the BRDs are in tune for the roll mode.
Similarly, I measured the bounce mode right after installation (see figure 2) with the auto settings on. The Q of the peak is high compared to the measurements for the other BRDs. Could it be related to the setting change once again? Yes, I have to remeasure the mode.
For reference, please find the description of the current installation in the drawing attached.
Today I removed the BRD5_v5 and BRD6_v5 from the BS suspension, after their 6 months stay.
I measured their tuning right after. The identification of the BRD is not visible (it is written on the blade but hidden in the mount). So I just assigned a new name to the BRDs.
The measurements below confirm that the BRDs are still within range(+/-1%) after 6 months.
BRD5a_v5 - Bounce Mode
Tuning is 0.1%
BRD5a_v5 - Roll Mode
Tuning is 0.4%
BRD6a_v5 - Bounce Mode
Tuning is 0.8%
BRD6a_v5 - Roll Mode
Tuning is 0.6%
BRD9_v5 and BRD10_v5 were remeasured and tuned yesterday (see elog 235).
I double checked the tuning right before installing the BRDs on the BS. Only BRD10 Roll mode was out of the 1% range (F = 24.05 Hz, too low), so I retuned it.
For reference, the final tuning is:
BRD9_v5 - Bounce Mode
Tuning is 0.2%
BRD9_v5 - Roll Mode
BRD10_v5 - Bounce Mode
BRD10_v5 - Roll Mode
Tuning is 0.3%
====== Roll mode
Here is a summary of the BS roll mode survey with BRD5_v5 and BRD6_v6 attached to the suspension for 6 months.
Details of the measurements are attached in the spreadsheet.
The measured transfer functions over time are shown in figure 1. We observed two peaks in the data. This is in agreement with our model if the tuning if the BRD is within 0.1% of the BS roll mode (see T1900846).
Therefore it seems that the bounce damping is pretty stable over 6 months. We didn't analyze correlations of the variations with environmental factors in the lab.
Measurements after taking off the BRD from BS:
Recall that pre-installation measurements were:
====== Bounce Mode
Here is a summary of the BS bounce mode survey with BRD5_v5 and BRD6_v6 attached to the suspension for 6 months.
Details of the measurements are attached in the spreadsheet.
The measured transfer functions over time are shown in figure 1. We observed a single peak in the data. This is unexpected from our model.
Therefore it seems that the bounce damping is pertty stable over 6 months. We didn't analyze correlations of the variations with environmental factors in the lab.
Recall that pre-installation measurements were:
Constructed and tuned the BRD9 and BRD10 V5. The masses on each BRD were:
In the attached excel spreadsheet are the measured frequencies of the BRDs before and after tuning the dampeners. The BRDs were tuned to be with +/- 1% of the ideal value for the dampeners. These frequencies are 16.69 Hz for the bounce mode and 24.34 Hz for the roll mode. Also, the change in frequency over this week-long process was recorded to determine potential drift of the dampeners.
It appears that Basler offers GigE cameras with over 2x improved sensitivity to the typical LIGO infrared laser wavelength. This camera is the "acA1300-60gmNIR" and it appears to be 2 to 5 times more sensitive at 1064 nm.
We do not have any of these cameras, and might want to consider getting our hands on one to evaluate its utility in lab settings (or perhaps even in site GigE camera installations).
As discussed in the Basler technical documentation, there are 3 different sensors that Basler packages into their cameras.
--> ref. https://www.baslerweb.com/en/vision-campus/camera-technology/nir-cameras/
"The NIR-optimized cameras with NIR-optimized 2 MP (CMV2000) and 4 MP (CMV4000) sensors from CMOSIS, or the 1.3 MP sensor (EV76C661) from e2V, still manage quantum efficiencies close to 40% in the 850nm range. Compared to non-NIR-optimized cameras, this represents a doubling of the sensitivity value at this wavelength."
Examining these sensors closer, the better QE at 1000 nm wavelength is from the 1.3 MP sensor (EV76C661) from e2V. This is notable because LIGO uses 1064 nm wavelength, and the highest QE reported in the datasheets is at 1000 nm.
The camera which uses this sensor is the "acA1300-60gmNIR" which is available in either C-mount lens interface, or CS-mount lens interface.
--> ref. https://www.baslerweb.com/en/products/cameras/area-scan-cameras/ace/aca1300-60gmnir/
Begin at a given camera's Basler webpage. See example below.
--> ref. https://www.baslerweb.com/en/sales-support/downloads/document-downloads/basler-ace-aca2000-50gmnir-emva-data/
Navigate to "Documents" tab and then to "EMVA Data". There are two attachments in this section:
By using the camera-specific pdf, you can identify the quantum efficiency at NIR wavelenghts up to 1000 nm.
Alternatively, if you know you want a specific sensor or specification, you can identify the cameras using the overview pdf, or using the Vision System Configurator
--> ref. https://www.baslerweb.com/en/products/tools/vision-system-configurator/#/selection/camera/
We'll use this log as a starting point to compile the resolutions, sensitivities, and any other parameters for each existing LIGO GigE camera and each possible improved camera.
We will also continue our work in identifying all of the necessary components that could help us construct a mobile setup which enables interchangeability of the various GigE cameras that are at our disposal (WIP).
Data is in attached excel file.
From January 14 to the 21, Bounce BRD 7's frequency drifted -2.68% which differes from the drift logged on January 14 which was 0.15% on the opposite direction. The Roll of BRD 7's frequency drifted -4.41% which is a big change from the previous logged drift of 0.06%.
The Roll side for BRD 7 on January 21 was much closer to the target frequency. It was logged as 24.36 Hz and the target is 24.34 Hz. The Bounce side was also logged as much closer to the target frequncy. The target is 16.69 Hz and the frequency logged was 16.80 Hz.
The Q factors for both the Roll side and the Bounce side have decreased from the previous entry. Roll decreased from 108 Hz to 100 Hz and Bounce decreased from 87 Hz to 79 Hz.
BRD 8's frequency was tested for the first time. Roll side's frequency was 24.57 Hz. Bounce side's frequency was 17.956 Hz. Roll side is close to the target frequency, but Bounce side is off by 1.26 Hz.
The length for the new copper masses will be computed.
To determine the best achievable Q as a function of the mistuning, we studied a simple model with 2 resonant masses (see elog 150). Now we have a model with 3 resonant masses (presented in elog 217, 219, T1900846).
As a reminder, in order to obtain the same transfer function with 2 masses model and the 3 masses model, we need to multiply the mass of the unique BRD by 2 in the model with 2 masses (see figure 1).
Here we assumed the Q of the standalone BRD is 110 and perfect tuning. The resulting Q of the two peaks is around 220. This is in line with the rule of thumb that states "for perfect tuning, the resultant Qs of the bounce and roll modes are approximately 2x the Q of the damper" (Norna's email). However, with the 3 masses model, the resonance of the BS corresponds to the "dip" in the transfer function. At this frequency, the Q value is exactly the Q of the damper. Therefore, there is no factor 2 for the BS mode.
%% Recalling here some results:
Mass ratio is therefore:
And we measured:
In T1500271, Brett derives the best achievable Q of a damper for a given mass ratio:
Q = 1/(2*w1/w2*sqrt(3*mu(ii)/(8*(1+mu(ii))^3))); % from equation 12 in T1500271
Comparing these values to our blade v5 measurements for bounce and roll modes, we can see that the actual Qs for the standalone BRDs are a factor 2.5 higher than this ideal case (see figure 2). We therefore updated the study of elog 150 with the new 3 resonant masses model and the Qs of the standalone BRDs equal to 2.5 times the ideal Qs. The results are summarized in graphs 3 and 4. We can see that:
So we could aim for a tuning of the BRDs within 1% of the BS resonant frequencies.
Overnight the Bounce mode of BRD7_v5 drifted 0.16% which is surprising if we compare it to the drift from the previous month which was lower and in the other direction. The Roll mode of BRD7_v5 drifted by 0.06% and also in the positive direction. As mentioned in elog 154 we want to tune the BRD to 16.69 Hz for Bounce and 24.34 Hz for Roll.
We tuned the bounce side at 0.25% from the target at 16.731 Hz. We noticed that the Q factor increased from 60 to 90 during the tuning. Could it be related to the tightness of the screw?
We tuned the Roll side 0.07% from the target at 24.347 Hz. Again we see an increase in the Q factor to over 100.
Today we added the masses for the Roll of BRD 8:
We built 2 BRDs with version 5 of the blade before the break to be used as witnesses for the drift over the next weeks. BRD 8 is made with a baked blade from last summer. We didn't have time to properly tune the BRDs, but they can still be used as witnesses.
Here are the measurment results for BRD 7 before and after the break. We see a negative drift for both modes. For the roll mode the drift is -0.1% and the bounce mode is -0.3%.
We built 2 BRDs with version 5 of the blade before the break to be used as witnesses for the drift over the next weeks. BRD 8 v5 is made with a baked blade from last summer. We didn't have time to properly tune the BRDs, but they can still be used as witnesses.
Here are the measurment results for BRD 7 v5 before and after the break. We see a negative drift for both modes. For the roll mode the drift is -0.1% and the bounce mode is -0.3%.
CIT Modal lab is currently experimenting using a mirror relay to measure the VMD at a 45 degree angle to the mounting plate. The rational behind this is the desire to measure the approximate resonance frequency of both the first and second mode in one experiment. Future work will include validating this measurement with tuning of the VMD to make sure we are correctly identifying the modes.
Below are the temperature v. time plots from the new Platinum series contoller/software. The profile ran was a 2.5 hour ramp to 94°C, 2.5 hour soak, and 2.5 hour ramp down to room temp. The controller had a maximum overshoot of ~4°C. The Extech datalogger was also used to track temperature, with a maximum overshoot of ~ 5°C.
Attached are the measurements of the bounce roll transfer function of the dummy BS with BRD5 and BRD6 (v5 blades) fastened to the pum.
BRD5_v5 was built on July 16 (elog 165) and BRD6_v5 was built on August 21 (elog 198) after baking (elog 204). It was found that BRD5_v5 stopped drifting at the end of August (elog 204). They were attached on the suspension on August 28th after tuning (elog 202). The date of the measurements is recalled in the table attached, with the latests ones from December 4th, 2019. There are some measurements missing from early November unfortunately (issues with the spectrum analyzer saving data/USB key problems).
Nevertheless, we can clearly see that the BS resonances are still damped over the time period of 3 months, meaning that BRDs are still well in tune with the BS resonances. It means that they stopped drifting or that the drift is now negligeable.
Note: the measurements 206, 207, 208, 212 were taken with a lower amplitude of excitation (250 mV instead of 1V) to check if a high amplitude excitation was not introducing non-linear effects. It might be that the coherence is lower for these measurements and there are therefore less reliable.
StephenA, various others :)
Sharing a link to photos of SR3 ROC Actuator efforts in various labs at CIT - others involved in various measurements are welcome to add their own photos here.
Notes about the lab setups:
See attached file for measurements.
RichA, StephenA, LizN 12 November 2019 (catching up after-the-fact)
A new simple oven has been procured which will be dedicated to dirty operations like the "Wait Test" (in other words, elevated temperature checks that proportions and procedures of epoxy mixing have been adequate mixing of epoxy, such as EP30-2 - ref. T1300322 section I.5).
This oven has been property tagged and Liz was planning to help input into the PCS system - C32329 is the property tag but currently there is a conflict in PCS; here is the conflicting entry https://services.ligo-la.caltech.edu/Inventory/history.php?recid=1590&invtype=cit_noncap
Rich has reviewed the electrical characteristics of the toaster (Power = 1150 W per manufacturer spec, so current drawn will be < 10 A, and therefore no issue plugging into standard wall outlet) and has checked dis/continuity of ground and power at the plug. Rich has given the OK to proceed using this oven.
Photos show oven, property tag, and receipt.
Following up elog 217, I want to make sure that the 2 BRD model is correct. Here is a comparison of the model with the 2 identical BRDs compared to the simple model with 1 BRD (the mass is twice the mass of an individual BRD but the Q is unchanged). In both cases the BRDs are tuned at the exact resonant frequency of the BS mode.
We can see that the models are perfectly overlapping. Hence, the 2 BRD model is in agreement with the 1 BRD model (see figure 1)
For the Roll mode, the poles and zeros of the 2 BRD model are:
The poles and zeros of the 1 BRD model are:
If the frequencies of the BRDs are symetrically detuned from the BS resonant frequency in the 2 BRD model, the poles and zeros are:
For 1 % detuning:
For 0.1% detuning :
The resulting transfer function is presented in figure 2.
Updated T1900341 Matlab code with additional comments to allow for more ease in changing the code for individual needs. Also updated VMD document T1800474 to clarify the conclusions in mass shift, and fix some grammatical errors. Continued modal testing of Mirror relay in hopes of determining feasibility of using this mounting set up.
We want to compare the model transfer function with the measurements on BS.
We built a model with two BRDs attached to the main mass. Model is provided in the attachment. Details of the computation will be given in a note later. We implemented the model in Matlab.
Back in July (alog 177), we tuned the BRD frequencies right before mounting them on BS. We installed BRD1 and BRD3 (alog 181) with the following properties (alog 177, alog 164, alog 141):
Plugging these properties in our new 2 BRDs model, we obtain the transfer function in Figure 1 and Figure 2. For comparison, the 1 BRD model with only BRD1(3) is shown on the same graph. The tables below show a summary of the model outputs.
The 2 BRD model generates three poles but in the present case we observe only "2 peaks", as we have seen in the measurements (see here measurements #33 and #34). We need some larger detuning to see "3 peaks". See for example figure 3 with respectively -0.1Hz detuning for BRD1 and +0.1Hz detuning for BRD3. The Q are ~ a factor 2 lower than expected from the model.
%% Use the results from the Model_with_2parallel_BRDs.nb
% based on the model from Norna T1600259
% MK November 2019
clc; clear all; close all;
% BS and damper parameters
% primary oscillator with damping
% BS Bounce mode
I just added a document to DCC the describe the electrical characteristics from the Aplus Satellite Amplififer E1900355. Also a test procedure was preparer to check all connections on the Aplus Satellite Amplifier E1900245.
Finalized revisions for version two of the Matlab script responsible for intaking B&K files and self-normalizing the different tests to allow for direct comparisons. This program's purpose is to allow for comparisons of different tests in which the excitation level is varied, or binning needs to be completed before comparisons can be made. Outputs include the normalized plot, a plot illustrating the effect of binning selection, and then a file output showing details pertaining 3dB and resonance frequency. More details can be found at T1900341, along with a demonstration video to the posted in the next week.
BRDs with the version 4 of the blades have been tested on stand-alone version and in the BS suspension during the summer. Here is a summary of our findings with the references to the corresponding elogs.
Drifts measured on two BRDs before baking over 20 and 40 days (elog 186). The variation of the resonance frequencies are reported in the table below:
Time lapse dt [days]
Drift df [Hz]
Drift df [%]
Drift Rate [mHz/day]
Drift Rate [%/day]
On the dummy BS suspension:
1 - Resonance frequencies
The two BRDs were installed on the dummy BS for a month. Unfortunately we are missing some of the measurements because some data got corrupted (see spreadsheet attached). Therefore the analysis is only performed over 16 days (8 days) for the bounce (roll) mode.
For each mode (bounce and roll), two peaks are observed around the resonance where we expected to resolve three peaks. The three peaks would be the main BS resonance as well as one peak per BRD. We might need to increase the scan resolution (see figure 2 and 3).
The frequency of the peaks do not match the frequency of the BRDs measured alone. The frequency of the minima in between the resonance peaks is close to the resonance frequency of the BS for respectively the bounce and the roll modes. The shift in the BS resonance frequency due to the added mass is negligible.
We observed a steady drift of the resonance frequencies over time. The frequencies are increasing by few hundreds of ppm per day, see below and figures 4 and 5.
Summary of BS Bounce frequencies
Initial mistuning [%]
Final mistuning [%]
We are observing a change of 0.2% in the resonance frequency over the 16 days. This is above the requirements that we set at 0.1% tuning. We didn’t observe a stabilization in the drift.
Summary of BS Roll frequencies
We are observing a change of 0.1% in the resonance frequency over the 8 days. We didn’t observe a stabilization in the drift, so this is likely to exceed our requirements.
When remeasured stand-alone after being uninstalled from the BS suspension, we established that the BRDs frequencies drifted of about 0.5% after being for a month on the suspension (elog 203).
2 - Q factor
The resonance amplitudes, corresponding to the quality factor of the modes, fluctuate over time without a distinguishable pattern. However, it seems that for each mode the two resonances vary together, in particular for the roll modes.
The mean of the bounce mode Q is 140 (168) for the 16.66 Hz (16.69 Hz) resonance. The mean of the bounce mode Q is 115 (152) for the 24 Hz (24.5 Hz) resonance. Q are lower than expected according to the model (we expected Q~200).
We observed a steady drift of the resonance frequencies of the BRDs over time, when stand alone or on the dummy BS. The frequencies are increasing by few hundreds of ppm per day. We see a stabilization in the drifts after about a month in the lab. The drift is slightly lower when the BRDs are mounted on the BS suspension compared to the stand alone BRDs in the lab. This could confirm that the excitation measurements cause some of the drift and we need to revise the method. We have no evidence that the baking process reduces the frequency drifts.
The measurements of the quality factor shows that the peaks may not be quite resolved, and that we are underestimating the Qs of the modes on the suspension. However, the value of the Qs in the stand-alone measurements is already promising.
Luis, Rich, Don:
After getting some feedback from Chandra, we agree to revise the documents D1900114 and D1900116. This revision is to allocate the air purge valve on the other side of the chamber, now changing the valve from D6 Flange to D2 Flange. The documents are being modified by Don. As seen in the image, we are also adding the "HDS" on the suspension lines for better identification. The seismic db25 connectors are being allocated to D3 Flange and D3-F11 and D3-F112. The same modifications are being made to document D1900116.
A modification on documents D1900114 and D1900116 is on the works, and the changes will be modified by Don, as soon as he has the time. This new change is required because we need a one port for the Air Purge Valve at HAM7. This valve will help to bring the Chamber to room conditions -not under pressure-. Squeezer Tip-Tiilts that are located on D6 Flange (IC1, IC2) will change their location to D8 Flange, which is located at the top of the chamber. D8 will be a flange with 12 dsubs with 25 pins each.
Just started to test the pcb D1900217-v2, the voltage noise level appears normal and in range, see following charts. I need to take the Instrument level noise and add this to the charts.
I assembled two A-Plus Suspension Satellite chassis, and also I seriliazed the pcb boards D1900217 for these D1900089 units, the numbers are S1900540 and S1900541. In the near future I will add all information on the e-traveler, for now I need to do electrical test on these units.
I tried to generate a Sunstone quote for PUM Driver D070483 pcb board, but for some reason the web site was not generating the quote numbers. I asked Patrick from Sunstone and he helped us to generate the quote, now the ODB++ and gerber files are located in Sunstone server for future reference, quote number SQW-52105.
I serialized the pcb boards for Binary Output chassis, and I completed the electrical assembly of 5 units. I still need to serialized the chassis and add all information to the e-traveler. Another step that these chassis require is to do electrical test, also I need to find or to order some 1U handles, screws, etc.
I modified the Squeezer Wiring diagram by removng the VOPO and ZM Suspension elements and I updated this. I will add the removed sections to the A-plus HAM7 chamber.
Created HDS clasification table, We need to define which units will require dither and where these will be located.