40m QIL Cryo_Lab CTN SUS_Lab TCS_Lab OMC_Lab CRIME_Lab FEA ENG_Labs OptContFac Mariner WBEEShop
  SUS Lab eLog, Page 35 of 37  Not logged in ELOG logo
ID Date Author Type Categorydown Subject
  491   Tue May 1 15:57:30 2012 ericqMiscCoating Qre-running

Quote:

nothing visible, restarted sweep around 6 kHz and started up the Turbo at 13:58.

I also noticed that whoever is setting up the 'Remote' data taking is changing the input configuration of the channels of the SR785 into something non-sensical. Check your settings.

 Nothing visible in the 6.07-6.17kHz range either. I'm starting one around the 10.6kHz range

  492   Tue May 1 17:16:26 2012 ZachMiscCoating Qre-running

To: Eric, Rana, Giordon

Why are we doing 100-Hz sweeps again? I thought we decided to focus on narrower sweeps about the frequencies that Rana & Alastair measured with the "screwdriver method". 800 pts over 100 Hz near 6 kHz is not enough resolution to see the modes we want.

Quote:

Quote:

nothing visible, restarted sweep around 6 kHz and started up the Turbo at 13:58.

I also noticed that whoever is setting up the 'Remote' data taking is changing the input configuration of the channels of the SR785 into something non-sensical. Check your settings.

 Nothing visible in the 6.07-6.17kHz range either. I'm starting one around the 10.6kHz range

 

  493   Tue May 1 17:34:15 2012 Giordon StarkMiscCoating Qre-running

I asked Eric about this - and I'm not sure I was convinced. The explanation was something like "even if we don't see the total width of the peak, we'll still see some sharp peak somewhere". I even explained that the line width (gamma) of the peaks are around 0.03 and 300-800 points over 100Hz only gives a resolution of 0.3ish (we're still an order of magnitude off).

Quote:

To: Eric, Rana, Giordon

Why are we doing 100-Hz sweeps again? I thought we decided to focus on narrower sweeps about the frequencies that Rana & Alastair measured with the "screwdriver method". 800 pts over 100 Hz near 6 kHz is not enough resolution to see the modes we want.

Quote:

Quote:

nothing visible, restarted sweep around 6 kHz and started up the Turbo at 13:58.

I also noticed that whoever is setting up the 'Remote' data taking is changing the input configuration of the channels of the SR785 into something non-sensical. Check your settings.

 Nothing visible in the 6.07-6.17kHz range either. I'm starting one around the 10.6kHz range

 

 

  494   Tue May 1 17:37:38 2012 ranaMiscCoating Qre-running

Appears to be fixed now - issue was the channel 1 input was set to be a differential "A-B" later on in the remote script - I removed that line and fixed it.

Quote:

I also noticed that whoever is setting up the 'Remote' data taking is changing the input configuration of the channels of the SR785 into something non-sensical. Check your settings.

 

  496   Tue May 1 17:44:58 2012 ZachMiscCoating Qre-running

If you were just measuring the power in some wide band (e.g., measuring a PSD with a bandwidth that is wider than the resonance), then I think what you're saying is true. That is, you'd see some frequency bin with a higher signal than its neighbors', and you could then choose to make a finer sweep around it.

Since you are measuring a transfer function, that is not true. You are trying to excite a narrow resonance with too coarse a sweep. Since the excitation signal is only being stepped in frequency by the same intervals as your resolution, the best you can hope for is that by some chance it happens to choose a point near the resonance for that one drive cycle. The odds of this are quite small.

Use finer resolution and look for the peaks Rana saw.

Quote:

I asked Eric about this - and I'm not sure I was convinced. The explanation was something like "even if we don't see the total width of the peak, we'll still see some sharp peak somewhere". I even explained that the line width (gamma) of the peaks are around 0.03 and 300-800 points over 100Hz only gives a resolution of 0.3ish (we're still an order of magnitude off).

 

 

 

  501   Tue May 1 20:45:17 2012 Giordon StarkDailyProgressCoating QSweeps Done

Rana swept around 3kHz and 6kHz and found nothing. Eric swept around 10kHz and found nothing.

Sweeps Ran
Sweep 1 2 3 4 5
Start Freq (kHz) 2.686 4.670 37.062 41.558 43.974
End Freq (kHz) 2.706 4.690 37.082 41.578 43.994
Num Points 2048 1536 1024 1024 1024
Resolution 0.009 0.013 0.02 0.02 0.02
Int Time (s) 1 1 1 1 1
Settle Time (s) 1 1 1 1 1

Note: the resolution varies with the frequency because we're trying to run these sweeps as fast as possible with the largest-possible resolution.

In the meantime, I'll be digitally filtering the ringdown measurements and see if we can't get something better (using Matlab + butter).

 2.696kHz.png4.68kHz.png

37.072kHz.png41.568kHz.png

43.984kHz.png

Conclusion

We see one of the same modes from before - but also a new one about 5Hz higher as well. We'll measure ringdowns at both of these points.

  502   Tue May 1 20:53:09 2012 Giordon StarkDailyProgressCoating QRingdowns of peaks around 37kHz

 The first peak is at 37.0698kHz. The second is at 37.0762kHz. (See fits at the end).

Ringdown measurements will be done now. Results as soon as I get any.

Observations: the two modes we see here make the substrate swing strongly like a pendulum (parallel to the ESD).

Here are the attempts at ringdowns - seems like they exist for the first mode:

tek00000.pngtek00001.pngtek00002.png

37.0698kHz_fit.png37.0762kHz_fit.png

 

  503   Tue May 1 23:03:04 2012 Giordon StarkDailyProgressCoating QFiner Sweeps being run

Talked to Eric - the second plot of the ringdown measurement looks super cool, but it decays fast - so we're going to look at it more in the morning. I've got sweeps running with a better resolution because fast sweeps with 300 points - 800 points over 100 Hz isn't good enough.

 

  504   Wed May 2 01:06:24 2012 ZachDailyProgressCoating QRingdowns of peaks around 37kHz

What is are the differences between the three measurements? What did you change in between? The first and last look dominated by noise, but somehow the SNR is much higher in the second.

In the case of the second, it appears that the beat frequency is ~2 Hz, which I think is too low. You should really have several-to-many cycles over one ringdown time.

 

Quote:

 The first peak is at 37.0698kHz. The second is at 37.0762kHz. (See fits at the end).

Ringdown measurements will be done now. Results as soon as I get any.

Observations: the two modes we see here make the substrate swing strongly like a pendulum (parallel to the ESD).

Here are the attempts at ringdowns - seems like they exist for the first mode:

tek00000.pngtek00001.pngtek00002.png

 

 

  505   Wed May 2 02:38:07 2012 Giordon StarkDailyProgressCoating QRingdowns of peaks around 37kHz

 T'm going to look over this with Eric in the morning - but the _major_ difference (since I maximized SNR as best as I could) was how far away the lock-in amplifier was from the drive frequency

  1. 25 Hz away
  2. ~1 Hz away
  3. 100 Hz away

The thing that bugged me was that I could see a ring-down for 1 Hz away, and "slightly" for 100 Hz away. I know I'm probably doing something wrong.

Quote:

What is are the differences between the three measurements? What did you change in between? The first and last look dominated by noise, but somehow the SNR is much higher in the second.

In the case of the second, it appears that the beat frequency is ~2 Hz, which I think is too low. You should really have several-to-many cycles over one ringdown time.

 

Quote:

 The first peak is at 37.0698kHz. The second is at 37.0762kHz. (See fits at the end).

Ringdown measurements will be done now. Results as soon as I get any.

Observations: the two modes we see here make the substrate swing strongly like a pendulum (parallel to the ESD).

Here are the attempts at ringdowns - seems like they exist for the first mode:

tek00000.pngtek00001.pngtek00002.png

 

 

 

  506   Wed May 2 03:04:25 2012 Giordon StarkDailyProgressCoating QFine Sweeps of All Frequencies

 There is an issue: we cannot get more than 1024 points sampled (both the manual and online quote this sampling to be capped at 2048; for us - it appears to be 1024). SR785 Manual (2-51)

Here is an eLog which contains all pictures of the sweeps for quick perusing:

2.686kHz-2.706kHz.png3.446kHz-3.466kHz.png4.670kHz-4.690kHz.png

6.118kHz-6.138kHz.png10.622kHz-10.642kHz.png37.062kHz-37.082kHz.png

41.558kHz-41.578kHz.png43.974kHz-43.994kHz.png

 

 

 

 

 

  507   Wed May 2 13:00:42 2012 Giordon StarkDailyProgressCoating QRingdown Measurements and Fits

[Eric, Giordon]

 We were able to get more beats of the ringdown by moving the drive frequency away from the resonance - I don't understand totally why this makes it work... but it does. And the ringdowns are real.

 

Variable Value Confidence Interval
tau 1.963 (1.955,1.970)

Q = pi*f*tau

f = 37.0685kHz; this gives us Q = 2.286e+05

Best (and prettiest) Fit

RingdownFit_37.0685kHz.png

Fits are below

TEK00000fit.png

 

General model:
     f(t) = amplitude * exp(-(t+t0)/tau) * cos(2*pi*freq*(t+t0) + phi) + v0
Coefficients (with 95% confidence bounds):
       amplitude =      0.3463  (-2.059e+04, 2.059e+04)
       freq =       1.648  (1.647, 1.649)
       phi =      -2.836  (-1.192e+06, 1.192e+06)
       t0 =       2.984  (-1.151e+05, 1.151e+05)
       tau =       1.936  (1.917, 1.955)
       v0 =   -0.009284  (-0.009759, -0.008809)
 
Goodness of fit:
  SSE: 1.759
  R-square: 0.9707
  Adjusted R-square: 0.9706
  RMSE: 0.0179
 

 

TEK00001fit.png 

General model:
     f(t) = amplitude * exp(-(t+t0)/tau) * cos(2*pi*freq*(t+t0) + phi) + v0
Coefficients (with 95% confidence bounds):
       amplitude =      0.3598  (-1.629e+04, 1.629e+04)
       freq =       1.625  (1.624, 1.626)
       phi =      -0.467  (-9.204e+05, 9.204e+05)
       t0 =         2.8  (-9.014e+04, 9.014e+04)
       tau =       1.991  (1.962, 2.02)
       v0 =   -0.009983  (-0.01058, -0.009388)
 
Goodness of fit:
  SSE: 1.674
  R-square: 0.9578
  Adjusted R-square: 0.9577
  RMSE: 0.01974
TEK00002fit.png
General model:
     f(t) = amplitude * exp(-(t+t0)/tau) * cos(2*pi*freq*(t+t0) + phi) + v0
Coefficients (with 95% confidence bounds):
       amplitude =      0.2811  (-5310, 5311)
       freq =       1.619  (1.618, 1.62)
       phi =      -2.035  (-4.048e+05, 4.048e+05)
       t0 =       3.441  (-3.98e+04, 3.98e+04)
       tau =       2.106  (2.081, 2.132)
       v0 =    -0.01416  (-0.01463, -0.0137)
 
Goodness of fit:
  SSE: 1.629
  R-square: 0.9603
  Adjusted R-square: 0.9602
  RMSE: 0.01738
TEK00003fit.png
General model:
     f(t) = amplitude * exp(-(t+t0)/tau) * cos(2*pi*freq*(t+t0) + phi) + v0
Coefficients (with 95% confidence bounds):
       amplitude =      0.3236  (-2889, 2890)
       freq =       1.602  (1.602, 1.603)
       phi =      -3.897  (-1.717e+05, 1.717e+05)
       t0 =       3.437  (-1.705e+04, 1.705e+04)
       tau =       1.909  (1.896, 1.923)
       v0 =    -0.01285  (-0.01318, -0.01253)
 
Goodness of fit:
  SSE: 1.007
  R-square: 0.9817
  Adjusted R-square: 0.9817
  RMSE: 0.01293
TEK00004fit.png 
General model:
     f(t) = amplitude * exp(-(t+t0)/tau) * cos(2*pi*freq*(t+t0) + phi) + v0
Coefficients (with 95% confidence bounds):
       amplitude =      0.2909  (-1.035e+04, 1.035e+04)
       freq =        1.67  (1.669, 1.671)
       phi =     -0.3562  (-6.772e+05, 6.772e+05)
       t0 =       3.152  (-6.452e+04, 6.453e+04)
       tau =       1.814  (1.798, 1.83)
       v0 =    -0.01601  (-0.01655, -0.01548)
 
Goodness of fit:
  SSE: 1.908
  R-square: 0.9777
  Adjusted R-square: 0.9777
  RMSE: 0.01936
TEK00005fit.png
General model:
     f(t) = amplitude * exp(-(t+t0)/tau) * cos(2*pi*freq*(t+t0) + phi) + v0
Coefficients (with 95% confidence bounds):
       amplitude =      0.3266  (-7595, 7596)
       freq =       1.665  (1.664, 1.666)
       phi =      -2.655  (-4.552e+05, 4.552e+05)
       t0 =        3.73  (-4.35e+04, 4.351e+04)
       tau =       1.871  (1.852, 1.889)
       v0 =    -0.01725  (-0.01785, -0.01665)
 
Goodness of fit:
  SSE: 2.644
  R-square: 0.9695
  Adjusted R-square: 0.9695
  RMSE: 0.02224
TEK00006fit.png
General model:
     f(t) = amplitude * exp(-(t+t0)/tau) * cos(2*pi*freq*(t+t0) + phi) + v0
Coefficients (with 95% confidence bounds):
       amplitude =      0.3356  (-3844, 3844)
       freq =       1.658  (1.657, 1.658)
       phi =      -1.919  (-2.519e+05, 2.519e+05)
       t0 =       3.856  (-2.418e+04, 2.418e+04)
       tau =       2.111  (2.091, 2.132)
       v0 =    -0.01635  (-0.01694, -0.01575)
 
Goodness of fit:
  SSE: 2.916
  R-square: 0.972
  Adjusted R-square: 0.9719
  RMSE: 0.02274
 

 

  508   Wed May 2 14:01:46 2012 QDailyProgressCoating QRingdown Measurements and Fits

Nice work James!

Screen_Shot_2012-05-02_at_2.00.33_PM.png

Quote:

[Eric, Giordon]

 We were able to get more beats of the ringdown by moving the drive frequency away from the resonance - I don't understand totally why this makes it work... but it does. And the ringdowns are real.

 

 

  509   Wed May 2 16:14:28 2012 GSDailyProgressCoating QMoved ESD, pumping down now

 We moved the ESD to a new position (IE: centered horizontally wrt to substrate first) which is vertically above. This allows us to sweep horizontally with the laser physically so we can find more modes by just moving the laser around YAAAAAY

 

photo.JPG

The only "slightly" annoying thing is the lack of vertical flexibility - but that's fine for now. Also, the output glass port is SUPER dirty. Maybe after I turn in the thesis, I can punish myself by cleaning it.

  510   Thu May 3 08:16:18 2012 GSDailyProgressCoating QFine Sweeps Ran - Found new peak(s)

 Here are pictures of the sweeps - total time is roughly 10 hours.

Interesting Sweeps: 2.7kHz, 37kHz, 44kHz

2.686kHz-2.706kHz.png3.446kHz-3.466kHz.png

4.670kHz-4.690kHz.png6.118kHz-6.138kHz.png

10.622kHz-10.642kHz.png37.062kHz-37.082kHz.png

41.558kHz-41.578kHz.png43.974kHz-43.994kHz.png

Peak: 2.69828 kHz

2.686kHz-2.706kHz_fit.png

  511   Thu May 3 10:08:18 2012 GSDailyProgressCoating QRingdown of 2.698kHz Successful

 We've got some good measurements of the ringdown (at least 2 cycles) of the decay sign for the 2.698kHz peak.

  • drive: 2.698130kHz, 500mVpk (5th measurement was 1Vpk)
  • Lock-In Amplifier: 300ms, 2mV, 2.698kHz

How was this found? First the lock in amplifier was set for high SNR, and then we set the drive at the peak frequency we found (see previous eLog post) 2.69828kHz. Then we slowly lowered the drive frequency until the signal was maximized on the oscilloscope (readily found). The reported drive frequency above is what is displayed on the screen. Note that a FFT of this signal will show a drive frequency higher than actually is reported. This seems to be doing the trick as I was able to find these ringdowns within minutes.

 

Variable Value Confidence Interval
tau 13.540 (13.482,13.597)

Q = pi*f*tau

f = 2.69828kHz; this gives us Q = 1.148e+05

Pictures (See last plot for actual title and labels - it's the prettiest one!)

TEK0fit.png

 

General model:
     f(t) = amplitude*exp(-(t+t0)/tau)*cos(2*pi*freq*(t+t0)+phi)+v0
Coefficients (with 95% confidence bounds):
       amplitude =         1.5  (fixed at bound)
       freq =       0.126  (0.1259, 0.1261)
       phi =      -9.545  (-9.584, -9.506)
       t0 =       11.35  (11.3, 11.4)
       tau =       13.02  (12.91, 13.12)
       v0 =     -0.1671  (-0.1697, -0.1646)
 
Goodness of fit:
  SSE: 146.8
  R-square: 0.9674
  Adjusted R-square: 0.9674
  RMSE: 0.1253
 

 

 

TEK1fit.png

 

General model:
     f(t) = amplitude*exp(-(t+t0)/tau)*cos(2*pi*freq*(t+t0)+phi)+v0
Coefficients (with 95% confidence bounds):
       amplitude =      0.5171  (-3672, 3673)
       freq =      0.1257  (0.1256, 0.1258)
       phi =     -0.2989  (-8.571e+04, 8.571e+04)
       t0 =      0.6966  (-1.085e+05, 1.085e+05)
       tau =       15.28  (15.1, 15.46)
       v0 =     -0.1102  (-0.1132, -0.1071)
 
Goodness of fit:
  SSE: 234.2
  R-square: 0.9301
  Adjusted R-square: 0.9301
  RMSE: 0.1531
 

 

TEK2fit.png

 

General model:
     f(t) = amplitude*exp(-(t+t0)/tau)*cos(2*pi*freq*(t+t0)+phi)+v0
Coefficients (with 95% confidence bounds):
       amplitude =      0.4999  (-2439, 2440)
       freq =      0.1275  (0.1274, 0.1276)
       phi =      -1.805  (-5.305e+04, 5.304e+04)
       t0 =     0.07889  (-6.622e+04, 6.622e+04)
       tau =       13.57  (13.44, 13.69)
       v0 =     -0.1144  (-0.117, -0.1118)
 
Goodness of fit:
  SSE: 163
  R-square: 0.9564
  Adjusted R-square: 0.9564
  RMSE: 0.129
 

 

 

TEK3fit.png

 

General model:
     f(t) = amplitude*exp(-(t+t0)/tau)*cos(2*pi*freq*(t+t0)+phi)+v0
Coefficients (with 95% confidence bounds):
       amplitude =      0.4712  (-3027, 3028)
       freq =      0.1291  (0.1289, 0.1292)
       phi =     -0.8274  (-7.143e+04, 7.143e+04)
       t0 =      0.4611  (-8.809e+04, 8.809e+04)
       tau =       13.71  (13.58, 13.85)
       v0 =     -0.1166  (-0.1194, -0.1138)
 
Goodness of fit:
  SSE: 197.6
  R-square: 0.9468
  Adjusted R-square: 0.9468
  RMSE: 0.1406
 

 

 

TEK4fit.png

 

General model:
     f(t) = amplitude*exp(-(t+t0)/tau)*cos(2*pi*freq*(t+t0)+phi)+v0
Coefficients (with 95% confidence bounds):
       amplitude =      0.8299  (-9267, 9269)
       freq =       0.131  (0.1309, 0.1311)
       phi =     -0.7241  (-1.114e+05, 1.114e+05)
       t0 =      0.5943  (-1.353e+05, 1.353e+05)
       tau =       12.12  (12.02, 12.22)
       v0 =     -0.1372  (-0.1408, -0.1335)
 
Goodness of fit:
  SSE: 212.2
  R-square: 0.9723
  Adjusted R-square: 0.9723
  RMSE: 0.1629
 

 

 

 

  512   Thu May 3 12:15:52 2012 Giordon StarkDailyProgressCoating QMore measurements of ringdown

 [Eric, Giordon (E.G.)]

 

Variable Value Confidence Interval
tau 10.565 (10.464,10.669)

Q = pi*f*tau

f = 2.69828kHz; this gives us Q = 8.956e+04

TEK0fit.pngTEK0fit_residuals.png

 

General model:
     f(t) = amplitude*exp(-(t+t0)/tau)*cos(2*pi*freq*(t+t0) + phi)+v0
Coefficients (with 95% confidence bounds):
       amplitude =       2.712  (-2748, 2754)
       freq =        4.15  (4.15, 4.15)
       phi =      -51.63  (-2.987e+05, 2.986e+05)
       t0 =      -13.15  (-1.146e+04, 1.144e+04)
       tau =        11.3  (11.11, 11.48)
       v0 =     -0.1402  (-0.1446, -0.1358)
 
Goodness of fit:
  SSE: 54.65
  R-square: 0.9505
  Adjusted R-square: 0.9504
  RMSE: 0.1288
 

 

tek1fit.png

 

General model:
     f(t) = amplitude*exp(-(t+t0)/tau)*cos(2*pi*freq*(t+t0)+phi)+v0
Coefficients (with 95% confidence bounds):
       amplitude =      0.6546  (-501.3, 502.6)
       freq =       4.152  (4.152, 4.153)
       phi =     -0.1172  (-1.966e+05, 1.966e+05)
       t0 =      0.2326  (-7536, 7536)
       tau =       9.829  (9.724, 9.933)
       v0 =     -0.1389  (-0.1417, -0.1362)
 
Goodness of fit:
  SSE: 196.7
  R-square: 0.9507
  Adjusted R-square: 0.9506
  RMSE: 0.1403
 

 

  513   Thu May 3 16:26:22 2012 Giordon StarkDailyProgressCoating QSwitching out sample

[Alastair, Zach, Eric, Giordon]

 Quoting Zach: "Given that the Qs are a bit lower than what we had expected, Rana was thinking that we should switch out the sample for one of the ones from Steve Penn ASAP"

We measured some of the ringdowns for the 2kHz mode - didn't get a chance to measure the 37kHz. We had a slight issue with keeping the beam aligned as we move across. This led to the process of rebalancing the setup, which took up time, so we never got to measure the 37kHz in this position.

4:24pm - Alastair is basically ready to weld things. By 5pm, we should have the whole system pumping down - at which point I shall continue the process of rebalancing the laser so it's ready to run sweeps later tonight once I get the turbo pump spinning up. In the meantime, I'll read the presentation Rana sent for the FEM analysis and work on the thesis more so that people can comment.

Why am I still typing this eLog when there's Science to do?

 5:05pm - "OH SHIT". Glass fiber between the test mass and the substrate broke while it was in the chamber. Substrate fell, chipped along the edge at one spot. Zach probably knows better about the plan for this - he's talking to Alastair?

I've realigned and rebalanced the laser in the experimental set-up. I've also added more beam dumps (see picture). The one closest to the QWP is to grab the reflection off of the focusing lens (which is tilted slightly). I've also removed all excess wires and cleaned up the area a little bit more. We had an issue with one of the photodetectors (top right) as it wasn't working (constant 8V output no matter what). I just put it aside for now and replaced it.

beamdumps_backend_expsetup.JPG

 

The fit for the 2kHz that Eric rang down is not going terribly well. I've tried Alastair's idea which was to flip the negative parts of the amplitude up, average it, then take the log and fit a line to it - the slope of the line being -1/tau . This resulted in a line that gave tau ~ 18; but preliminary. I need to do this better, and right.

  514   Fri May 4 15:11:19 2012 Giordon StarkDailyProgressCoating QReplaced substrate, re-welded, and pumped down (now scanning)

 Current: roughing is on, turbo is on and spun up - a sweep was ran across the 37kHz area again but this showed no modes!

A sweep is being run across the 2kHz mode from before - and if this doesn't show anything, we'll run all the scans again and see if there's a mode somewhere in the same frequencies that Rana found (although I'm a little wary that we'll find anything -- perhaps we should pump up, tap, and find the new modes?)

Zach should have better pictures of the fixed substrate (the one that fell and got chipped).

 

 

  515   Fri May 4 15:45:11 2012 Giordon StarkDailyProgressCoating QChamber pumped up

[Giordon] Waiting on someone to lift it off so we can tap to find modes

  516   Fri May 4 16:51:07 2012 Giordon StarkDailyProgressCoating QNew modes found by tapping
  • 2.728kHz
  • 6.44kHz * probably not one
  • 14.056kHz
  • 15.464kHz
  • 37.16kHz
  • 37.416kHz
  • 41.192kHz
  • 41.576kHz
  • 64.448kHz

The ones highlighted green are the strongest modes we found - these will be swept first. If we find that mode, we will do ringdown measurements at that mode before sweeping the next one.

Chamber is being pumped down now.

 

  517   Fri May 4 19:37:22 2012 ZachMiscCoating QOur samples

I was asked to identify our suspects:

  • Initial disc from Gregg Harry
    • Dimensions: 3" diameter x 1/8" thickness, beveled edge
    • Substrate: fused silica
    • Coating: 45-layer Tantala:Alumina

Mug shots:

GHarry_disc.png g_harry_3in_disc_copy.png

 

  • New disc from Steve Penn (there are two, but one is cracked and not shown here)
    • Dimensions: 3" diameter x 1/10" thickness, beveled edge (more than GH's)
    • Substrate: fused silica
    • Coating: None
    • Comments: Broke off of fiber inside chamber during prep and was chipped on the side. Alastair flame polished the chipped portion.

Mug shots: (freshly chipped on left, smoothed on right)

SPenn_disc_chipped.png SPenn_disc_chip_smoothed.png

 

  518   Fri May 4 21:02:19 2012 Giordon StarkDailyProgressCoating QRunning sweeps now

That took a while - but it pumped down to around 30 mTorr so I turned on the turbo and let it spin up. Turned everything on and we're sweeping the green ones now - roughly 2ks per sweep (quick, but fine enough to get the details of each peak).

  • 2.728kHz == nope
  • 37.16kHz == Yes (37.156961kHz)

 

 

Quote:
  • 2.728kHz
  • 6.44kHz * probably not one
  • 14.056kHz
  • 15.464kHz
  • 37.16kHz
  • 37.416kHz
  • 41.192kHz
  • 41.576kHz
  • 64.448kHz

The ones highlighted green are the strongest modes we found - these will be swept first. If we find that mode, we will do ringdown measurements at that mode before sweeping the next one.

Chamber is being pumped down now.

 

 2.728kHz.png37.16kHz.png

37kHz Mode Ringdown

The SNR here is significantly higher than what I've ever gotten
before. On a really interesting note, while it seems modulated like a
beat, it is ringing down, although this ringdown time seems to be
significantly long (IE: order of minutes). I have yet to find the
exact peak, but it is somewhere between:

37.156348kHz - 37.156961kHz

So I'm going to search slower. But it rang up pretty fast and pretty
strong. I have no idea how I'm going to measure the ringdown easily if
it rings down this slowly, but I shall try! See attached video.

  519   Fri May 4 23:51:02 2012 Giordon StarkDailyProgressCoating Q37kHz ringdown?

I talked with Zach and I think he is right - there are two modes in the area and I don't think I can excite one without exciting the other (the resolution in my function generator is not good enough). This one was ran at 37.156170 kHz on the drive, with a lock-in at 37.15.

 37kHzringdown?.png

Beats in my ringdown?

I increased the drive (by rotating the knob) to 37.156570 kHz and we see the following (these are taken while the drive is on):

driving_t0.pngdriving_t1.png

Where those beats we see happen to be the driven signal. My guess of what's going on is that the beat shown is because of the way the lock-in works (it's close to the drive) and we're driving near the resonance, but not right at it. In this plot, I turned off the driving signal around t=25 - and you can see that the beats go away.

driving_to_undriven.png

So overall - we're seeing beats because we're driving pretty much not on resonance - so it rings up a little bit, but as it rings down (fast-ish) the drive rings it up again, and it's sort of a push-pull thing I can imagine going on. At some point, if I increased the drive to 1Vpk (instead of 0.5Vpk), these beats would disappear. My guess here is that the resonance was rung up enough that the drive was overpowered, so we didn't see these beats. However, since I can't seem to find the mode exactly... Conclusion - I should run finer sweeps to really pinpoint that resonance. I'll run the sweeps through to morning, work on the thesis until they finish, and then do more ringdowns in the afternoon.

 

  520   Sat May 5 01:40:57 2012 ZachDailyProgressCoating Q37kHz ringdown?

I took the liberty of analyzing Giordon's most recent measurement. It looks pretty good, with an apparent Q of ~1.67 x 107. I think we expect it to be at least this high for a pure FS disc with no coating.

The analysis was done in the following way:

  • Take Giordon's time series data (100 s at 100 Hz = 10000 pts)
  • Rectify using abs()
  • Take a 100-tap moving average
  • Take ln (using log())
  • Fit output to line

The linear coefficient fit.p1 is -1/tau.

37kHz_ringdown.png

  521   Sat May 5 15:13:15 2012 ZachDailyProgressCoating QBevel added, mode at ~37.16 kHz now in COMSOL

 

By intersecting the cylinder with a torus, I was able to get something very close to a flat bevelled edge. I adjusted the size of the bevel until the 37 kHz mode matched roughly what we have empirically. I have updated the 3D COMSOL file

Interestingly, it still doesn't show the 2.6-kHz mode.

Screen_Shot_2012-05-05_at_3.06.41_PM.png Screen_Shot_2012-05-05_at_3.07.18_PM.png

Edit by Giordon:

It seems to predict the 64.448kHz mode that we saw though (centered the frequency around 64khz in COMSOL)

Screen_Shot_2012-05-05_at_3.28.31_PM.pngScreen_Shot_2012-05-05_at_3.28.40_PM.png

  522   Sat May 5 16:17:51 2012 Giordon StarkDailyProgressCoating QSweeps Done

The first 4-5 sweeps below 37kHz showed nothing, and I was suspicious. So I ran another sweep around 37kHz where I know I did a ringdown measurement and had previously found a peak -- this time, the peak/mode isn't there. I'm not sure what's going on. I looked at the set-up, the ESD appears fine (no scorch marks), the substrate looks ok, it's not sticking to the ESD; all the cables are hooked up fine. 

Edit: i made some adjustments and rebalanced the system again, ran a quick sweep from 35kHz - 38kHz and I see a peak. It also appears there might be multiple modes next to each other, so I'm running a finer sweep right now.

 

37.155kHz-37.160kHz.png 

Weird sweeps:

2.728kHz-2.738kHz.png6.430kHz-6.450kHz.png

14.046kHz-14.066kHz.png15.454kHz-15.474kHz.png

37.155kHz-37.157kHz.png37.406kHz-37.426kHz.png

41.182kHz-41.202kHz.png

  524   Sun May 6 02:25:27 2012 Giordon StarkDailyProgressCoating QRingdown measurements and strategizing

[Rana, Rana, and Giordon]

 Recall the list of modes we found by tapping [Zach, Alastair] (in kHz):

  • 2.728
  • 14.056 / 15.464
  • 37.16 / 37.416
  • 41.192 / 41.576
  • 64.448

Rana and I did a power spectrum, input noise, and looked for these peaks again. The only one that was found was the 37kHz mode (37.15740625kHz; 31.25mHz line width). This is being driven at 37.1574kHz sine (163.5mVpk). To get a ringdown measurement, we're waiting for the steady state. Since we know that Q~10^7, tau~100s; the steady state takes a couple of cycles so when we excite it - we're waiting 500s or more until we turn off the drive. That way, what we're seeing is a true ringdown and not just some (I can't describe it any other way) "push-pull" beat between the resonance and the drive [since we aren't driving it RIGHT at resonance].

Ringdown measurement details

  • LIA: 3ms, 100mV, 37.16kHz
  • Drive: 37.1574kHz, 163.5mVpk
  • Peak: 37.15740625kHz

I made 5 measurements of each ringdown (50s each approximately). Each measurement has two snapshots, one right at the beginning when the drive is turned off; one as soon as it finishes saving the first snapshot. We'll use each one to calculate the Q for that particular snapshot. A note: Rana mentioned that high-amplitude decay fast and might undershoot Q, while low-amplitude decays slow and might overshoot Q (shoot == estimate) [relatively].

Other changes to the set-up will be needed on Wednesday (after the thesis is done):

  • the PLCX (focusing lens) has the wrong AR coating for our laser [it's reflecting too much]

The plan for more data by Tuesday:

  • Measure 5 to 6 ringdowns of the uncoated sample that is currently in there [tonight]
  • Put the coated sample back in, and get more ringdown measurements of that [Monday? would need Alastair's help here I assume and need to work around his schedule?]

Plan for the thesis:

  • Talk about finding the relationship between the Q of the uncoated sample, the Q of the coated sample using the relationship
    • 1/Q_coated,f = 1/Q_uncoated,f + \alpha_f 1/Q_B + \beta_f 1/Q_S
  • Go into detail about error propagation, explain what "next steps" are, how to improve the set-up a lot more
  • KEEP MAKING PROGRESS
  525   Sun May 6 03:08:33 2012 ranaDailyProgressCoating QRingdown measurements and strategizing

 

 We translated the beam on the optic by ~1 cm today. It is now close to the bottom edge. I believe this made the sensing of the 37 kHz beam easier.

For version 2 of this setup, we want to use multiple drive connections on the outside and connect to multiple fingers on the ESD. Then we can use mechanical switches to reconfigure the drive force pattern and more easily ring up different modes.

Would be nice to figure out a way to do the same thing for the sensing; i.e. be able to choose among multiple sensing beams easily without re-aligning the whole thing.

  526   Sun May 6 16:42:57 2012 Giordon StarkDailyProgressCoating QFits of ringdown measurements for 37kHz - Uncoated sample

 This eLog will detail the fits involved. This a two-step process. The first step is getting a preliminary tau estimate using the rectifying method (m-file attached). The second step is to fit a decay sine curve knowing the tau and the frequency. There are 10 fits, 2 measurements (a,b) of a ringdown, 5 separate ringdowns. In other words, because the quality factor of the uncoated sample is significantly higher - it takes a lot longer for it to ringdown. Because of this, we can't capture the full ringdown in one snapshot of the oscilloscope. Instead, we capture 50seconds of it, save it, and then capture another 50 seconds of it. (a) = first 50 seconds of ringdown, (b) = second 50 seconds of the ringdown about 1-2minutes later.

 Notation: tek#(a|b). The # corresponds to the ringdown number, which ringdown measurement we're looking at. The a/b at the end corresponds to which portion of the ringdown we are looking at: a = first 50 seconds, b = later 50 seconds.

 

 

Rectification

rectifiedData.png

The rectifying method is plotted for all ringdown measurements above. Note how the slopes of the lines seem to be about the same - which tells us that these ringdowns are going to give us the same tau (or Q) -- something we expect. On a similar note, we also see that the second set of measurements for the ringdowns (#b) are noisier and have a smaller slope (more horizontal = larger tau = larger Q). The first set of measurements (#b) have a larger slope and therefore calculate a lower Q.

Preliminary taus:

Ringdown Number 1st snapshot 2nd snapshot
1 78.82 98.33
2 76.34 101.45
3 77.10 124.73
4 81.17 257.20
5 82.92 238.34

Preliminary Result: tau = (121.64 +- 68.28) 

Q = pi*37.15740625kHz * tau = 1.42 * 10^7 (0.62, 2.22)*10^7

Plots of rectification:

tek0fit.pngtek1fit.png

tek2fit.pngtek3fit.png

tek4fit.png

 

 Decay Sine Fits

Table summarizing the fits:

Ringdown Number 1st snapshot 2nd snapshot
1 77.89 (77.64, 78.14) 79.08 (76.66, 81.49)
2 75.75 (75.53, 75.98) 77.44 (75.43, 79.44)
3 76.16 (75.92, 76.41) 75.03 (72.48, 77.59)
4 77.73 (77.28, 78.19) 75.67 70.96, 80.37)
5 77.6 (77.14, 78.07) 84.81 (79.16, 90.46)

Overall: tau = 77.72 (77.12, 78.32)

Q = pi*37.15740625kHz * tau = 9.07*10^6 (9.00,9.14)*10^6

Screw it - too many plots. Here's a GIANT pdf (click it to see all fits in order: a regular fit, a close up of the fit.)

AllFits.pdf

  527   Sun May 6 21:10:02 2012 ZachDailyProgressCoating QFits of ringdown measurements for 37kHz - Uncoated sample

This is very nice. Can I request that you put the resultant Qs in the table alongside the taus? Also, why are the different trials so different? Please put something meaningful in the legends to distinguish runs---they might as well be Teletubbies' names right now.

  528   Mon May 7 03:29:22 2012 Giordon StarkDailyProgressCoating QFits of ringdown measurements for 37kHz - Uncoated sample

Quote:

This is very nice. Can I request that you put the resultant Qs in the table alongside the taus? Also, why are the different trials so different? Please put something meaningful in the legends to distinguish runs---they might as well be Teletubbies' names right now.

It's pretty hard to see that bright pink! I've updated the post with calculated overall Q for both the rectifying method you did as well as the sine decay fits. One of the reasons the cases are different (I assume you're looking at that first case tek0* versus all other cases) is the amount of time taken to let the mode ring up before turning the drive off, ~60 minutes versus ~10 minutes for the other cases. Because we are so close to a mode with a high Q, it is taking long enough to ring up that we can see the resultant beats from not driving it right at the resonant frequency. Also, as you can see from the actual fits later on - there's not much extra precision gained with driving it for 60 minutes versus 10 minutes; in the interest of time - 10 minutes is enough.

For future runs - I will put more meaningful names. What are teletubbies?

  529   Mon May 7 10:34:27 2012 ZachDailyProgressCoating QFits of ringdown measurements for 37kHz - Uncoated sample

Quote:

Quote:

This is very nice. Can I request that you put the resultant Qs in the table alongside the taus? Also, why are the different trials so different? Please put something meaningful in the legends to distinguish runs---they might as well be Teletubbies' names right now.

It's pretty hard to see that bright pink! I've updated the post with calculated overall Q for both the rectifying method you did as well as the sine decay fits. One of the reasons the cases are different (I assume you're looking at that first case tek0* versus all other cases) is the amount of time taken to let the mode ring up before turning the drive off, ~60 minutes versus ~10 minutes for the other cases. Because we are so close to a mode with a high Q, it is taking long enough to ring up that we can see the resultant beats from not driving it right at the resonant frequency. Also, as you can see from the actual fits later on - there's not much extra precision gained with driving it for 60 minutes versus 10 minutes; in the interest of time - 10 minutes is enough.

For future runs - I will put more meaningful names. What are teletubbies?

 I didn't make it pink.

  530   Wed May 23 10:25:00 2012 ZachDailyProgressCoating Q633nm optics in

I received the red optics detailed in SUS:474 while I was gone last week. In case you're too lazy to click, they are

  • 2x PBS
  • 2x HWP
  • 2x QWP

 

  531   Fri May 25 14:44:47 2012 ZachMiscCoating QTurbopump borrowed for gyro

I borrowed the Hi-Cube turbo to pump down the gyro chamber. I turned off the large turbo and sealed the tank, and it seemed to still be below 1 uTorr after ~15 mins, so the leak rate is reasonable.

I will replace it before the end of the day.

  532   Fri May 25 20:15:39 2012 ZachMiscCoating Qpump replaced

I put the pump back on the CQ tank. The pressure had only risen to 4 uTorr over a day of being sealed off, so I spooled up the Hi-Cube, opened the valve and then reengaged the large turbo.

  535   Fri Jun 15 18:16:17 2012 ZachMiscCoating Qturbopump re-borrowed for gyro

I have once again borrowed the floor-standing HiCube turbopump to use on the gyro chamber.

Since no one is actively using the CQ chamber right now, I will probably just keep it in the ATF for the moment. The bell jar seems to hold ~uTorr (I meant mTorr -- ZK) vacuum after a day of being sealed, so it shouldn't be a problem to leave it as is.

  537   Sat Jun 23 22:15:26 2012 ZachMiscCoating QTank at ~160 mTorr

Since I am still using the HiCube pump for the gyro, I figured it was a good idea to check the pressure level of the bell jar every now and then.

It is at ~160 mTorr right now, which I assume is OK to leave it at (?).

I removed the pump on 6/15, so it has taken ~8 days to get to this level.

  538   Fri Jun 29 15:57:52 2012 Mike J.MiscCoating QTank at ~280 mTorr

Quote:

Since I am still using the HiCube pump for the gyro, I figured it was a good idea to check the pressure level of the bell jar every now and then.

It is at ~160 mTorr right now, which I assume is OK to leave it at (?).

I removed the pump on 6/15, so it has taken ~8 days to get to this level.

 283 mTorr at Fri Jun 29 15:57:02 2012

  540   Mon Jul 2 17:03:47 2012 Mike J.MiscCoating QTank at ~350 mTorr

 

 348 mTorr on Mon Jul 02 17:02:10 2012

  541   Fri Jul 6 14:00:52 2012 Matt A.DailyProgressCoating QProgress so far

Monika and I have been working to get to the Q measurement setup working again.

 

To date, we've:

  1. collected the SR785 signal analyser from Rana's Lab,
  2. stolen the HiCube back from the Gyro,
  3. gotten an oscilloscope from next to the cryostat (it didn't look like anybody was using it),
  4. Adjusted the polarization to maximize signal,
  5. Pumped down to ~10^-7 torr, and
  6. found the 37 kHz mode. -just from looking at the signal analyzer, the Q seems to be around 10^7, but that's a really rough estimation.

So that basically shows that the setup works, and we can take measurements.

Next, the plan is to actually record some measurements. I've got a laptop coming from Larry next week, and I've got a copy of Steve Penn's LabView code for measuring Q's. I'm going to see if I can get Monika to install all that and get it running so we can actually make measurements and compare them to Giordon's. We can also do some mode hunting, as it looks like he only ever found one mode on this sample. I don't want to open up the tank and reposition anything until we've got comparable ringdowns though. I think the exciter plate geometry/location is probably what's making it hard to find modes.

 

Once that's set up and working, we'll start looking into the comsol stuff. I imagine Monika will be gone by then, but I'm interested in figuring it out.

  542   Fri Jul 6 15:51:36 2012 AlastairDailyProgressCoating QProgress so far

 

 I have a laptop with a LabView license that you can use, that was bought for that setup.  There is also a National Instruments card that goes with it.

  544   Wed Jul 11 15:33:56 2012 Matt A.DailyProgressCoating QSystem working

 We got the laptop from Alastair with LabView and after some reworking, got Steve Penn's data-taking software to run. 

 

We also located a bunch of modes by hitting the table, and Monika is searching through them right now.

 

We rang up one of the butterfly modes just to test the system, but the Q looks to be around 90,000--Way too small. We'll try a couple more modes, but my thinking is that this is the butterfly x mode, so there's probably some loss at the suspension point that's really limiting us.

  546   Mon Jul 16 20:17:04 2012 Matt A.DailyProgressCoating QQ Measurements

 Monika is continuing to mode-hunt and take ringdowns.

 

In the meantime, I've written a quick and dirty Matlab code to get some Q's out.

Here they are:

We found two modes around 37 KHz, which is where Zach and Giordan found a mode on this sample. They're about 2 Hz apart, and I've labeled them 'Upper' and 'Lower' for what I hope are obvious reasons.

These modes both give Q~1million, although the lower mode seems to have a lower Q, possibly due to loss to the suspension.

 

We also measured one of the butterfly modes around 2700 Hz. This was a quick test, and it's not clear which of the butterflies this is. 

It gives a much lower Q, about 2e5.  If you look at the decay of the envelope though, you see that it's not just an exponential decay. I'm not going to guess at what might be causing this, but there are loads of possible reasons. We'll look into it.

 

 

  604   Wed Nov 7 15:27:04 2012 Matt A.SummaryCoating QCoating Q analysis

 I've been working on the COMSOL model for analyzing Coating Q measurements. 

 

First, I built the COMSOL model (attached, TwoPartDisc_Base.mph) out of two connected cylinders, one for the coating and one for the substrate, using typical thicknesses of a single layer on a 3" substrate. The substrate was made of silica, and the coating was made of tantala (nominally, I took the material params for silica and changed the Young's modulus to 140 GPa and the Poisson's ratio to 0.23). In a real sample, the silica substrate is beveled and there is a thumbnail section of the coating missing (for welding to the suspension fiber). It is generally assumed that these differences are negligible.

 

First, I calculated the energy ratio per coating thickness for a thin film on a thick substrate, i.e. dU_film/U_total. This is the typical value reported in the literature for extracting the coating loss from the measured loss of a coated substrate. These values can be compared to the literature values to make sure my model isn't completely wrong. Unfortunately, not many papers actually list what values they use, and even fewer list the Young's modulus of the coating, although I'll assume it's some calculation for a multilayer coating made of silica and tantala, so it'll be somewhere between 70 and 140 GPa. 

Energy Ratio Comparison
Mode R (Harry 07) R (Matt)
Bf 1584 2269
Dh 1659 2491
Bf2 1581 2276
Dh2 1624 2408

These are the first four modes of the substrate, at frequencies Bf: 2700, Dh: 4000, Bf2: 6000, Dh2: 9000 Hz, and usually the only four that are measured in the literature. The difference in energy ratios come from the difference in Young's modulus, and possibly Poisson's ratio. I've done a sweep of the Young's modulus of the coating from 70-280 GPa and you can see that these numbers agree around 100 GPa, which seems reasonable for a multilayer coating. The plot below shows the four model results below for the first 32 modes (many are double modes, and have the same frequency and energy ratio). The red points are equivalent to a pure silica coating (ignoring the effects of Poisson's ratio), the Green points would be for pure tantala, the blue would be for pure hafnia, and I don't know why I did the black, we're not likely to use a material that stiff.

Yf_EnergyRatio.png 

Varying the Poisson's ratio also has an effect on the energy ratio, but reasonable values of Poisson's ratio are pretty much all in the range of 0.2-0.3 (although silica is at 0.17). The result of varying within this range is shown below.

Pf_EnergyRatio.png

In most cases, this variation isn't a big deal.

Finally, I calculated the film bulk and shear energy ratios, U_bulk-film/U_film (R_bulk) and U_shear-film/U_film (R_shear), using the equations in Hong 12. (n.b. I had to use the equations in Hong to calculate the total energies as well, as the sum of shear and bulk seemed to add exactly a factor of two to the total energy calculated by COMSOL. I still don't know why this is true, but it doesn't matter to the ratios because this factor is removed in the ratio.)

For completeness, the equations used in comsol are:

U_bulk = 3D integral of (0.5 * solid.K * (solid.eel11+solid.eel22+solid.eel33)^2)

     where solid.K is the bulk modulus and solid.eel## are the diagonal elements of the strain tensor at each point.

U_shear = 3D integral of (2/3*solid.G*(solid.eX^2+3*solid.eXY^2+3*solid.eXZ^2+solid.eY^2+3*solid.eYZ^2+solid.eZ^2-solid.eY*solid.eZ-solid.eX*solid.eY-solid.eX*solid.eZ))

     where solid.G is the shear modulus and solid.e## are various components of the strain tensor at each point.

Below are two plots from the two parametric sweeps done above. The first shows the bulk energy ratio as Young's modulus is changed, and the second shows the bulk energy ratio as the Poisson's ratio is changed. Of course, the plots for shear ratios would just be 1-R_bulk.

Yf_Bulk.png

Pf_Bulk.png

As you can see, varying the Young's modulus has no effect on the bulk and shear energy ratios, which makes sense as the total, bulk, and shear energies are all linear with Y:

solid.K = Y/(3*(1-2*sig))       [sig = Poisson's ratio, Y = Young's modulus]

solid.G = Y/(2*(1+sig))

Changing the Poisson's ratio does change the energy ratios because the bulk and shear energies are not linear with sig, it it obviously makes a big difference, even with a small variation in sig (0.2-0.3).

 

It seems that it'll be important to get the Young's modulus and Poisson's ratio right when we measure different coatings directly.

 

Finally, there are two modes that are considered 'shear modes'. These are seen at ~37 kHz and ~42 kHz in the plots above. They are especially obvious in the plots of film to total energy ratios, where they are much lower than the other modes. This makes sense as most of the strain is in the direction perpendicular to the coating, where the cross-sectional area is very small. So while these modes might be interesting in terms of their bulk and shear film ratios (at least the 37 kHz mode, the other seems to approach the various butterfly modes), it would be very difficult to extract a film loss with them, as the effect of the film on the total loss would be very small. 

Instead, I think it's reasonable to focus on the numerous butterfly and drumhead modes, as there seems to be a clear divide in terms of their film shear and bulk ratios, and they all have larger total ratios. In fact, as a number of papers have already measured the film ratios at at least one butterfly and one drumhead mode, we can go back and look at their results (assuming our own Poisson's ratio (sig = 0.23).

The uncertainties on these will be large, since in most cases, only one drumhead mode is measured only once, and the butterfly modes will have multiple measurements. Also, the drumhead mode will most likely have higher suspension loss, and so its loss can be over-estimated. Finally, while the lowest measured loss is usually the one reported in the literature, this is usually just an upper limit, and the true loss might be much lower. Needless to say, the uncertainties on these are not really Gaussian. That being said, the errors I quote below are from the lscov function in Matlab, which does a least-squares fit to the data and assumes gaussian uncertainties. Take it as a zero-order measure if anything. 

Also, reading form the Hong paper, the ratio of phi_bulk to phi_shear is the figure of merit when determining how the difference between the two loss angles affects the coating brownian noise in the detector. Higher values are generally worse. In current calculations, it is assumed that this ratio is one.

From Penn et al. 03:

Samples Measured
Sample X-lambda silica layers Y-lambda tantala layers # layers
B 1/4 1/4 2
C 1/4 1/4 30
D 1/8 1/8 60

E

1/8 3/8 30
F 3/8 1/8 30

 

Results
Sample Modes Measured phi_bulk (1e-4) phi_shear (1e-4) phi_bulk/phi_shear
B Bf+, Bfx, Dh 3.4 pm 2.0 3.1 pm 0.4 1.1
C Bf+, Bfx, Dh 3.9pm0.2 2.6 pm 0.1 1.5
D Bf+, Bfx, Dh 3.7 pm 0.4 2.7 pm 0.1 1.4
E Bf+, Bfx, Dh 7.2 pm 0.1 3.7 pm 0.1 1.9
F Bf+, Bfx, Dh 2.7 pm 0.8 1.5 pm 0.2 1.8

Comparing B and C, it seems like the effect is not solely due to the coating/substrate interface. Comparing C and D, we see that the number of layers doesn't seem to make much difference. Comparing C, E, and F is confusing, as if only one of the coatings was causing the increased shear loss, C would fall between E and F. I haven't decided what to make of this, other than that these could be influenced by the uncertainties discussed above. The drumhead mode in E was higher than one might expect, so it could have been an errant measurement, and the modes in F were all about a factor of 2 lower loss, so they might have greater environmental influence.

 

From Crooks et al. 2004 (loss values estimated from plot)

Samples
Sample X-lambda silica layer Y-lambda tantala layer # layers
A 1/4 1/4 30
B 3/8 1/8 30
C 1/8 3/8 30

 

 

Results
Sample Modes Measured phi_bulk (1e-4) phi_shear (1e-4) phi_bulk/phi_shear
A Bfx, Bf+, Dh 3.6 pm 0.2 2.3 pm 0.1 1.6
B Bfx, Bf+, Dh, 2Bfx, 2Bf+ 2.6 pm 0.3 1.4 pm 0.1 1.9
C Bfx, Dh 6.8 3.1 2.2

 Comparing A, B, and C is the same as comparing C, E, and F above, and we see roughly the same trend, but with slightly higher ratios. Perhaps the two materials cancel each other somehow, perhaps it has something to do with layer thickness, or it could be due to offsets from our assumed Poisson's ratio.

 

 

 

 

 

 

 

 

 

 

 

 

  605   Wed Nov 7 15:55:35 2012 Matt A.SummaryCoating QCoating Q analysis

I also analyzed data from Harry 07 on titania-doped tantala, but the elog wouldn't let me write anymore. Here it is:

 

 

Samples
Sample titania concentration (cation %)
0 0
1 6-8.5
2 13-20.8
3 24-22.5
4 54.5-54

All samples were multilayer 1/4-lambda stacks of silica and Ti:tantala, at 30 layers.

Results
Sample Modes Measured phi_bulk (1e-4) phi_shear (1e-4) phi_bulk/phi_shear
0 Bfx, Bf+, Dh 3.3 2.4 1.4
1 Bfx, Bf+, Dh, 2Bfx, 2Bf+ 3.4 pm 7.6 3.2 pm 1.8 1.1
2 Bfx, Bf+, Dh, 2Bfx, 2Bf+, 2Dhx, 2Dh+ 2.6 pm 0.3 1.5 pm 0.1 1.7
3 Bf, Dh, 2Bf, 2Dhx, 2Dh+ 2.8 pm 0.3 1.4 pm 0.1 2.0
4 Bfx, Bf+, Dh, 2Bf, 2Dh 3.1 pm 0.3 1.4 pm 0.1 2.2

It looks like adding titania actually increases the bulk/shear ratio. The only point that does not fit this trend is sample 1, which has huge errorbars, due to some unusually large losses on the Bfx and Dh modes. If we remove just the Bfx mode, the ratio shoots up to 4.8, and if we remove the Dh and the Bfx, we get down to 1.8. Unfortunately, in removing the Dh, we lose out on a mode with very different energy ratios than the Bf modes.  

 

I should also note that the first coatings in the last post were from REO, the second were from LMA, and those above are also from LMA.

Also, if I didn't list any uncertainties in the tables, it's because there are effectively only two points and two unknowns, so the fit is perfect. For example, even though sample 0 above has three modes measured, both Bf modes have the same loss, and the same energy ratios for the two butterflies, so you're reduced to two points: Bf and Dh.

  368   Mon Oct 24 16:18:45 2011 Giordon StarkHowToCOMSOL+MatlabHow to set up COMSOL to communicate with Matlab (Mac OSX Lion)

  Mac OSX Lion came out pretty recently and COMSOL 4.2 [at the time of the writing] does not successfully install on these OSes. There's two issues in general - the first issue is that Lion changed the way some of the paths work (that COMSOL depends on) and COMSOL will throw seemingly unrelated errors in trying to start up.

install COMSOL 4.2 Update 3 (or better): http://www.comsol.com/support/updates/comsol42p3/

After installing, we need to update the MATLAB paths. Matlab normally installs in a directory under /Applications called 'MATLAB_R2010b'. Rename it by changing the underscore to a dash [COMSOL interprets the underscore as a 'space' which means it will never find it]. The new name should be 'MATLAB-R2010b'. Next, navigate to the COMSOL42/bin folder in /Applications. Depending on which mac build you're running (32-bit or 64-bit) - select the appropriate 'maci##' folder.

Inside are 4 initialization files to change:

  • comsolserver.ini
  • comsol.ini
  • comsolbatch.ini
  • comsolcompile.ini

On the very last line of each file - it should read:

 

-Dcs.mlroot=/Applications/MATLAB-R2010b.app

Once you've made the changes - you should be able to launch COMSOL+Matlab without (m)any issues.
  161   Tue Oct 12 10:51:36 2010 AlastairMiscBladesUseful document

 We're going to use this elog to store some of our lab work on blades that will be going on in the SUS lab.

As a first entry here is a useful document on the ALIGO blade design on the DCC:  LIGO T030107  by M.V.Plissi

  193   Thu May 5 21:39:07 2011 tara,ryanThings to BuyBladesblade holding block

We made a drawing for a structure hat will hold the maraging blade. The details aren't complete yet. The holes for the clamping will be  identified,  but the sketch shows the rough idea.

     We want to clamp the blade to a structure. The drawing for the clamp will be provided by Ryan (he found it in the dcc.) The structure is consisted of the base and the pillar. Although a monolithic structure is better, it might be to expensive to carve out a big piece of Al block, so Koji suggested that we do it like this. The base will be mounted on the table, and the pillar will be mounted on the base by 4 screws. The height of the pillar is not decided yet. It depends on how big the Al mass block we need to pull down the blade by its weight, and how the mirror for reflecting the beam up will be mounted, but it should be around 6 - 8 inches.

    The mass block will be used for mounting the end mirror of the interferometer + a translational stage. This way we can steer the beam with 2 mirrors and adjust the arm length. We will determine the weight, so we can estimate the size of the mass block, assuming we will use Al.

 

  196   Mon May 9 22:25:26 2011 tara,ryan, mingyuanThings to BuyBladesblade holding block

 We made a sketch for the weight clamp that will carry the mass block on the end of the blades. This will be done in Solidwork tomorrow.

 

   We plan  to load a block of mass under the tip of the blade by using a pair of knife edge pieces so that the rubbing between the mass block and the blade is minimized.

 The edge of the blade cannot be too large, or it will be noisy when the blade is driven. On the other hands, if the blade angle is too small (sharper blade), the stress on the blade due to the weight will be too large and cause plastic deformation on the blade, which we don't want. We plan to make it flat ~ 1mm wide, with 120degree open angle.

 The yield tensile strength of maraging steel is ~ 1 -2 GPa. With the contact area at the knife edge we can calculate the maximum clamping force.

The width of the edge is ~ 5cm

The thickness of the edge ~ 1mm.

so the maximum force should not exceed ~ 1 GPa x 0.05 m x 0.001 m ~10^4 newton.

We will use spring washers to make sure that we do not tighten the clamps together with too much force and cause plastic deformation on the blade.

 

 

ELOG V3.1.3-