40m QIL Cryo_Lab CTN SUS_Lab TCS_Lab OMC_Lab CRIME_Lab FEA ENG_Labs OptContFac Mariner WBEEShop
  SUS Lab eLog, Page 24 of 37  Not logged in ELOG logo
ID Date Author Type Category Subject
  766   Fri Dec 13 14:31:16 2013 ericqDailyProgressCrackleNew Analysis

We have our first results from the new setup, which has effectively eliminated the systematic effect due to differential beam spot motion.

We observe NO 2F or 4F crackling noise in the 550 to 590Hz bandwidth, with an 95% confidence upper limit of ~4.3 e-16 m/rtHz for 2F.

 Visibly, in the time series of the 2F Q, we see no real difference between the on and off portions. Furthermore, the failure to reject the null hypothesis of no difference in 2F Q power and the calculation of the upper limit was done via a one-sided student's t-test. 


So, how do we know the previously troublesome systematic is gone? Well, looking at the time series when the drive is switched off, for instance, we can see that the calibration line behavior is qualitatively unchanged. 


But, to be more thorough, we can look at the ASD of the calibration line mixer output when the drive is on vs. off. Whereas we see a large line appear at the drive frequency in our older data from October when we discovered this problem, the latest data shows no discernible extra Michelson gain fluctuation due to the common mode drive. 




  765   Tue Dec 10 17:58:38 2013 nicolasMiscCryo SUSSilicon Fiber cavity sensitivity to misalignments

If we build silicon fiber cavities without alignment control, how well do we need to have the two fibers aligned in order to have the cavity mode remain on the fibers?

I've made a short script that goes through the numbers, here is an example output:

>> beamshift
L = 10cm, R1 = 35cm, R2 = 35cm.
g1g2 = 0.5102
Beam size w1 = 223.0642 microns, w2 = 223.0642 microns.
For 10mrad misalignment, the beam shift is: 
0.14583cm or 6.5377 beam widths

So if our mirror pad is ~0.5cm radius, a shift of 0.15cm still keeps the beam spot many spot sizes from the edge.

This is just a first guess. We now need to come up with reasonable limits on the mirror radius and what our expected dead reckoned misalignment will be.

Code is in 40mSVN/SiCryoSus/Calculations/pointingSensitivity/

  764   Tue Dec 10 11:54:55 2013 GabrieleDailyProgressCrackleAnalysis of hysteresis measurement

Last night measurement went well. Unfortunately for reasons I don't understand the excitation level was not saved in the data, so I had to reconstruct it.

The blade A gave good data, while it seems we are saturating the driver of the B blade. So I analyzed only the A data.


The upper left plot shows the time evolution of the shadow sensor A output. I divided the period in three parts, corresponding to increasing, decreasing and increasing again force. The colors are maintained in all plots. The upper right plot shows the force/displacement plot. The superposition of the three part is not perfect. As in my previous analysis, I fitted a straight line and subtracted it. The result is shown in the lower left plot. There is a first indication of a non-linear behavior (blade or shadow sensor? To be checked). The final point is clearly different from the initial one. However, if one looks at the time domain, there is a clear drift. Since the measurement took about 14 hours, it is not surprising at all. Therefore, I treated again the raw data subtracting a simultaneous fit of the linear dependency force-displacement and of a linear drift in time. The results are shown in the lower right plot. There is still some not good superposition of the three sections. This is exactly what we should see in case of hysteresis. However, this can also be a residual of the trend not properly cancelled.

To distinguish, we should repeat the measurement with more cycles.

  763   Mon Dec 9 18:21:17 2013 GabrieleDailyProgressCrackleHysteresis

 At 6.15pm I started a hysteresis measurement with small steps. The force is changed in 21 steps following a sine with maximum amplitude of 8000 counts. Each step has a duration of 2400 seconds. 

  762   Sat Dec 7 20:48:29 2013 ericqComputingCrackleCymac Rebooted

Absurd amounts of channel hopping was going on. (Displacement excitation DQ channel wrote error signal data to disk, DTT was getting filter module input as its output...)

Restaring X1KRK didn't help. Restarting daqd didn't help. Rebooted cymac, seems to have worked.

  761   Sat Dec 7 11:23:33 2013 ericqDailyProgressCrackleCrackle Measurement Ongoing

 The Michelson lock is now stable enough to try to make another crackle measurement, chopping the 125mHz drive on and off every hour. We started it up yesterday, so far 17 hours of continuous data has been recorded! However, the alignment is slowly drifting out, the calibration line height has dropped to about 60% of its starting value. 

Update: I went in, touched up the alignment, and started another measurement with higher drive amplitude. 

  760   Thu Dec 5 18:26:48 2013 GabrieleDailyProgressCrackleHysteresis again...

 [Xiaouye, Gabriele]

We started another measurement trial for hysteresis. This time we are trying different frequencies between 1 and 1/512 Hz and different amplitudes between 4000 and 10000 counts. 

Damping is off.

  759   Thu Dec 5 09:57:21 2013 GabrieleDailyProgressCrackleAnalysis of last night hysteresis measurement

 The measurement went quite well during the night. Here is a trend of the output signal of the A shadow sensor:


If we restrict to one amplitude and plot the displacement vs the driving force, we expect to see some deviation from a straight line due to hysteresis. Here we should see any deviation due to the hysteresis. In particular we expect the system to follow two different orbits when the force is increasing or decreasing, like in this example from Wikipedia:


Here is an example of the typical results we get:


I fit this data with a straight line and looked at the residual. In following plot I divided the points corresponding to increasing force from those corresponding to decreasing force. There is no clear pattern:


The results are good for the blade A for amplitudes up to about 11000 counts of driving, then some saturation appears. In the case of the B blade the situation is much worse. For driving amplitudes of the order of 9000 counts we start seeing some saturation effect. For larger amplitudes there is a strange behavior when the force is positive, as if the blade is touching something:


However, looking at the good data from the A blade, no sign of hysteresis is detectable.

  758   Wed Dec 4 18:38:16 2013 GabrieleDailyProgressCrackleHysteresis measurement

 [Gabriele, Xiaoyue]

We started a hysteresis measurement, injecting a sine at 1/360 Hz. Every hour we increase linearly the amplitude. We started from 800 counts (roughly 1 micron) and we are increasing the amplitude up to 8000 counts in 10 steps.

We left the damping loops on during the measurement. 

  757   Wed Dec 4 18:36:02 2013 GabrieleDailyProgressCrackleBlade resonance measurements

 [Gabriele, Xiaoyue]

We took some time to measure the mechanical response of the two blades around the resonance frequency. We injected some common noise between 0.5 and 10 Hz and looked at the shadow sensor outputs.

We clearly see the first mode of both blades at about 2 Hz, and the second coupled one at 5 Hz. 


A quick fit gives the following poles and zeros (in Hz):


0.0083102 +     3.8822i
-0.045248 +     6.1484i
-0.002453 +     1.8849i
-0.013101 +     5.0392i
-0.054801 +     5.9915i


-0.01952 +     4.0533i
-0.078004 +     5.8236i
-0.017985 +     1.8791i
-0.018052 +     5.0409i
-0.14897 +     5.9551i

The resonance frequencies are pretty close.

  756   Tue Dec 3 17:01:13 2013 ericqDailyProgressCrackleFirst trial at measurement

 We've now compensated for the plant TF irregularities more accurately, and achieved a UGF frequency of about 150Hz, but were only able to turn on one integrator boost on, as opposed to the two I used to use. Nevertheless, things looked potentially stable enough to go for an overnight, slow chopped measurement, so we figured we should give it a shot. 

I edited the autolocker script (krklocker.py) with the new gain values, and have a script (slowchop.py) running that switches the .125Hz drive generated by an awggui instance on and off every hour, as long as the autolocker reports that it has the Michelson locked. 

Tomorrow, if anyone wants to use the apparatus, simply disabling the autolocker on the krkmaster medm screen and breaking the lock will prevent either script from touching anything thereafter. (If this turns out to be false, killing the python processes manually will do it.)

  755   Mon Dec 2 11:50:42 2013 GabrieleSummaryCrackleResidual fringe modulation due to common mode driving

 As pointed out in the previous entry, the new setup reduced a lot the modulation of the fringe due to misalignment when driving the common mode. To have a more precise estimation of the residual modulation, I used data from the two photodiodes with and without common mode driving, with unlocked Michelson. I computed the envelope of the Michelson fringes taking the maximum and minimum values during each second. The result is here:


The residual modulation is of the order of 3%, as visible in the first stretch of excitation. Interestingly, in the second one, the fringe modulation has a larger amplitude (5%) and opposite phase...

  754   Tue Nov 26 22:57:57 2013 ericqDailyProgressCrackleProgress towards full operation


At the start of the day, we had a very loose "lock" of the michelson. We made a measurement of the plant transfer functions between 100-400 Hz, to get a feel for the deviation from the 1/f^2 shape caused by the first higher order mode of the blades that consistently shows up around  250Hz. (The details of this deviation vary greatly depending on the mounting specifics, and thus change when when we change the setup). 

Eventually, we developed a compensation filter good enough to get the UGF up reasonably higher, and turn on one LF boost at 30 Hz. We're not quite all the way at the level of stability where we could have it locked overnight, but not too far.  

In addition, we were able to align the michelson with a contrast of 95% at one point, which is better than the previous setup ever saw.

More importantly, when driving the blades in common mode with an 800 count .125Hz sine wave, the michelson gain only changed by about 5%, which much smaller than the ~50% change seen in the previous setup at the same drive amplitude. This suggests that the systematic effect of michelson gain modulated noise should be about an order of magnitude smaller than in the previous setup, and hopefully underneath what can be integrated out in a finite amount of time. 

  753   Tue Nov 26 01:03:45 2013 xiaoyueDailyProgressCrackleSystem Installed

Attending: Eric, Gabriele, Xiaoyue

We overlapped the two arms by first eliminate the fringes as much as possible. However the angular motion of the suspended mirrors kept bringing the fringes in. Then we did finer aligning by maximizing the signal amplitude, making sure a good overlapping is giving magnificent interference (either constructive or destructive). A beating envelope was observed.

Then we installed the system into the chamber. Then we went through the procedures to generate the transfer function for the damping loop. We forgot to turn off the damping which at the beginning raised very strange phase behavior. As people walked around affected the signal a lot, we decided to pump the chamber overnight. 

  752   Thu Nov 21 22:58:51 2013 xiaoyueDailyProgressCrackleOptical Setup

Gabriele and I set up a rough Michelson interferometer alignment.

Matlab model -- Gabriele


Side view (left) Top view (right)


where the maroon beams are reflected from end mirrors. 

We tilted both of the 45 degree mirrors to deviate the reflected beam with an angle from incoming beams to bypass mirror 1(M1), to be detected by photodiode (PD).

Also it should be noted that the two end mirrors (ENDM1, ENDM2) have height difference of 1cm approximately, so we differentiate the optical path lengths of the two arms accordingly.

The alignment procedure is listed briefly as below:

- prepare the first mirror with a horizontal beam using irises.

- prepare a vertical reference using two irises aligned with a home-made plumb bob. 

- Align the 45 degrees inclined mirrors for vertical beams.

- Make sure the end mirrors are horizontal by overlapping incoming and reflected beams

- Roughly align the entire setup with beams superimposed

- Tweak the 45 degrees mirrors to separate the beam with a small angle (but large enough to bypass M1)

- Recenter or decenter mirrors to extract the symmetric port 

- Install photodiodes and beam dumps

(We may want to order more visible range mirrors / D-cut mirror / D-cut mirror mounts.)

  751   Thu Nov 21 09:26:50 2013 GabrieleDailyProgressCrackleA first analysis of the hysteresis measurement

 I downloaded the hysteresis data and had a first look at it. There are two main things to be addressed.

First, there is the global trend of both shadow sensor signals. Data is still uncalibrated.


I was able to fit it reasonably well with the logarithmic law x(t) = a*log(b*t+1)+c which we used also for the previous creep measurement. However, since one blade is going up and the other one is going down, this can’t be creep. It is likely a relaxation of the rubber below the plate. This also means that the "creep" we believed to see in entry 748 is likely the same phenomenon. However, I could remove the trend from both data sets and look at what remains
There is not much left indeed. I would say we are not able to see some clear hysteresis signature. Maybe it is simply covered by the fit residuals. 
The second strange things are the sudden jumps followed by rather slow relaxation. This happened for both the highest amplitudes of the green trace. It also happened once for the blue trace: in this case the blade was at a constant position and then suddenly moved. In all cases the motion is very fast and there is some ringing after it. Therefore I would conclude that it is a real mechanical displacement (a damping loop should not be able to create ringing…) 
In the B shadow sensor there is a clear 1.8 Hz oscillation damping out. Instead the A signal show a slightly less clear 2 Hz oscillation. We can imagine that they are both the blade resonances.
We should investigate what is the origin go these jumps…
I would say that this data is not very good for hysteresis measurement…
And moreover, this slow relaxation of the plate might create us troubles: if we align the two mirrors attached to blades in order to be horizontal, they will no more be once we move everything inside the vacuum.
  750   Tue Nov 19 22:29:39 2013 xiaoyueDailyProgressCrackleHysteresis Test Set Up

Attending: EricG, Gabriele, Xiaoyue

We installed two steel blades with different stiffness. By changing the hanging weight we are leveled the two arms, also the two end mirrors roughly. Leveling the whole stage by a more careful balancing of bearing weight, or does it matter much to the noise issue was left as an open question.

Eric walked me through the control system including the matlab program, emdm interfacing, foton, data viewer, striptool, diagnostic test tool, cymac terminal for signal generation. For every new system, a transfer function should be investigated by sweeping the driving frequency; or we can drive with a range of frequencies and extract the information from the output using Fourier analysis, but the coherence function exhibited a poor correlation (<<1, in a random fashion) between input and output signal when we tried this method.

We went through Eric’s python code for hysteresis analysis. We are driving the system on(+) – off – on(-) – off series with the sign indicating the direction of the driving force. The time interval between consequent actions is half an hour. We set it to run two cycles for each of the three levels of driving voltages.

  749   Mon Nov 18 19:18:36 2013 nicolasDailyProgressSi CantileverCantilever is back in the dewar

I've glued the heater and temperature sensor to the SS block and wired them to the outside through the feedthrough.

I set up the laser/split diode readout. The cantilever is flapping and being recorded. GPS of start of excitation is 1068866140.

The vacuum gauge reads 4e-7 torr. I am not convinced this is possible, but I did bake the chamber over the weekend.

  748   Mon Nov 11 18:04:16 2013 ericqDailyProgressCrackleMaraging Blade Creep

I put in a maraging steel blade back in the tank last week, to do some hysteresis measurements while we wait for a new post to come back from the shop for the new setup geometry. 

Just as a quick observation, we can see some creep (not crackle) happening. The plot here shows about a 1 micron downwards settling over a few days, time constant looks roughly like a day and a half (-1100 to ~-1030 over 2.5 to 4d)


  747   Fri Nov 8 18:45:30 2013 ericqDailyProgressCrackle"Cyclostationary" Shot Noise

Prompted by conversations with Gabriele, we discovered a previously unconsidered systematic effect in the crackle data. 

In general, we are trying to use periodicity of the crackling noise to tell it apart from stationary noise, such as shot noise. Since the incident power on the photodiodes when locked at half fringe is constant, I assumed the influence of shot noise on the differential displacement signal would be stationary as well. 

But, consider the following:

We induce a common mode drive in the blades. This causes a finite misalignment of the Michelson end mirrors (order of tens of urad), and due to the particular geometry of the setup, the beam spots on the photodiodes move in opposite directions. Thus, the Michelson gain changes with time. 

As far as the displacement signal is concerned, this is fine, as we can track the Michelson gain in real time with the calibration line at 1.5 kHz. 

However, the shot noise in terms of displacement noise looks like  M(t) * s(t), where M is michelson gain in [m/W] and s(t) is the stationary shot noise time series in [W]. 

My analysis code looks for crackle noise in the form of cos(2f_d t) c(t), so if M(t) has a component varying like cos(2f_d t), it will manifest as apparent crackle power.

 In short, it does, and this effect is responsible for the results I've measured thus far

Our short-term solution will be to rebuild the experiment in a geometry in which the spots move in the same direction, thus hopefully suppressing the change in Michelson gain enough where we can set an upper limit on mechanical crackle below the direct noise floor of the experiment. 

To go into more detail, I verified that this was the case in the following manner:

I replaced the time series PD data with random white noise corresponding to the shot noise of the laser power present in the michelson. I then ran it through the identical analysis routines as the real data, while "correcting" the fake data with the same calibration line data as the real data. I.e. simulating absolutely no crackle noise in the experiment, but the existence of the misalignment effect. The "crackle power" found in each simulated case corresponds well to the real data, even reproducing the change in magnitude for different drive amplitudes (since the different drive amplitudes cause different amounts of Michelson gain variation), and for both Maraging Steel and Stainless Steel blades. Here are the plots showing the results of this simulation:


  746   Tue Nov 5 21:37:11 2013 ranaComputingGeneralAndroid app for getting GPIB data?


  745   Tue Nov 5 14:40:30 2013 ericqDailyProgressCrackleMultiple amplitude Stainless Steel Measurement

 Gabriele suggested running multiple amplitudes in one data run, to move things along quicker. I did such a run on the stainless blades last night, at three amplitudes. Here's what came out:


We see the amplitudes scale with the drive amplitudes. Instead of the running medians that I used in the last post, I just plotted a thick line where the mean of each segment is, and thin lines showing the standard error of the mean (i.e. the std of the data / sqrt(no. of data points)). I will repeat this kind of run tonight to get better statistics on the amount of noise at each amplitude, since I'd like more than one recorded segment for the 600 count drive for instance, and then repeat for different frequencies. 

  744   Sat Nov 2 16:24:59 2013 ericqDailyProgressCrackleSome measurements on Stainless Blades

The stainless steel blades are fussy and noisy. Eventually recorded enough stable data to run my analysis codes on it.

I had to use a different bandwidth than in the last analysis, since there was too much excess noise in that band in the stainless blade's spectrum. Here, I've used the 740-790 Hz range. 

A crackle-like signal is observed! What is being plotted? We're searching for a signal like cos(2wd t)c(t). The y-axis is the average PSD of c(t) in the analyzed bandwidth. The x points are individual results of windowing a portion of the signal, and doing the 2f demodulation. The lines are a running median filter on these points, as the average value is what we're after. 


This is a comparable, but smaller, amount of noise to what was seen in the maraging steel blades. This is potentially consistent with the fact that the stainless blades are loaded with ~700g, whereas the maraging steel blades were loaded with >2kg. However, we did expect these blades to be reasonably "cracklier".  There's more to be done with this to make a proper comparison, which I am working on.

  743   Thu Oct 24 16:44:26 2013 ericqDailyProgressCrackleBarkhausen Noise Estimate

 Made some progress with the barkhausen noise systematic estimate...

In T080355, Rai finds an empirical relationship between applied external magnetic field, and the Barkhausen noise in the magnetic moment of the Neodymium magnets he tested. In order to estimate the systematic signal due to Barkhausen noise from the Neodymium driving magnets in my setup, I assume that the magnetization per unit volume of my magnets exhibits the same external field dependence as Rai's magnets.

Then, using the parameters of my coils and magnets, I find how much force noise I might expect from the barkhausen noise induced by the drive coil on the drive magnet. Then, it is simple to convert from a force noise spectrum into the integrated RMS displacement power in the bandwidth I analyzed in earlier results (400-500Hz). 

The result is an expected 7.3 x 10^-17 m power due to Barkhausen noise, whereas I observed 1.3 x 10^-14 m power difference when the drive was switch on, which is a little over 2 orders of magnitude higher. 

Thus, I conclude that Barkhausen noise was not a significant effect in my setup. My calculations are attached.

Attachment 1: 2013-10-24_16.29.00.jpg
Attachment 2: 2013-10-24_16.29.15.jpg
  742   Wed Oct 23 10:02:26 2013 EvanDailyProgressSi CantileverCryo cantilever dewar: heat tape turned off

I turned off the heat tape on the cryo cantilever dewar at GPS time 1066582781. The pressure was 1.673×10−4 torr and the temperature was 76.8 °C.

  741   Mon Oct 21 18:26:46 2013 ericqDailyProgressCrackleOct 21 Crackle Meeting Minutes

Crackle Meeting Oct 21

Attending: EricG, EricQ, Norna, Koji, Gabriele, Rana

  • Stainless steel blades are installed. Measurements will be done this week, to compare to maraging steel. Carbonsteel is next. 

  • We need to examine the choice of filters, windows, etc. in the signal processing chain, to be sure we're achieving maximum sensitivity. Literature about lock-ins should be helpful. We can empirically test out different choices on undriven data that exists. 

  • Discussed crackle propagation. What paths exist for crackle originating at different locations to move the test mass? Can twisting blade modes excite the fiber violin modes? Scaling of crackle from small to big blade also warrants further investigation. 

  • Discussed hysteresis (and contrasted to anelastic effects). Will measure hysteresis on carbon, stainless and maraging steels to try and connect with crackle measurements. 

  • We want to measure crackle at more drive frequencies, for potentially publishable results. This will be done after the carbon and stainless steel measurements. Should we measure all amplitudes and frequencies in the "bad" steels, or only a few to contrast?

  • Talked about systematic effects. My efforts on the Barkhausen noise calculation will be posted soon. DAC crackle noise can be remeasured with an AD620 instead of AD829, for better common mode rejection. ADC crackle should be measured as well. 

  • Discussed Crackle v2 experiment. How should blades be mounted / aligned in either Fabry-Perot or Michelson topologies? Better seismic common mode rejection is possible with horizontal mounting, but we lose the pre-stress of the masses used when loaded vertically. Eric and Gabriele will make some cartoons and quantitative estimates of noise in different measurement schemes, to be discussed at the next meeting. 

  • Also to be delivered next meeting: outline for Crackle paper. Need to consider what the target audience is: materials crackle people who would want to hear about a measurement of this effect or instrumentation people who want to hear about a new noise source

  740   Tue Oct 15 18:05:32 2013 nicolasLab InfrastructureSi Cantileverprobably a leak in my pump line

This is the pressure trend when I close off the valve between the pump line and the cryostat. I am not totally sure how to interpret this, but the fact that the pressure didn't go much lower (like the 10^-6 region) when the pump is only pumping the line makes me think that there may be a leak in my line. Something to investigate.


Attachment 1: pumplineleak.png
  739   Tue Oct 15 17:56:39 2013 ericqSummaryCrackleCrackle Updates

 An update about Crackleland is long overdue...

More measurements along the lines of ELOG 731 were made, showing a DC shift in the right fourier component of the displacement power that is consistent with Crackle noise. These results were presented at the LVC meeting in Hanover (slides here). I've also updated the "crackledoc" with these latest measurements, partially motivated by the need to inform Gabriele about the state of things (T1300465 (Still a few things missing that I want to update soon, though...))

At last week's crackle meeting, we decided that these measurements meant full steam ahead for the design process of crackleV2, to try and get down to lower frequencies that are more relevant for the suspensions, since seismic noise currently limits us up to the hundreds of Hz. We brainstormed some designs then, and I spoke with Gabriele over Skype yesterday. 

Simultaneously, I've had some replicas of my blades made from Carbon Steel and Stainless Steel. We want to throw these into the current setup to hopefully see different amounts of crackle in each material. (If the results don't change much, then maybe we're just seeing systematics...)


  738   Tue Oct 15 17:56:39 2013 nicolasLab InfrastructureSi Cantileverbake successful, going to bake more

Cryo cantilever dewar.

Here is the pressure trend (30 hours) after turning off my heat tape.


It quickly drops to ~ 1 x 10^-5 torr. Before the bake, it was never getting below 4 or 5 x10^-5 torr.

Because of this success, and because no work will be done on this chamber for the next week and a half. I will keep baking to further remove contamination.

New bake started at 1065921266.

  737   Mon Oct 14 18:05:35 2013 nicolasLab InfrastructureVacuumbake ended at 1065834142


pressure trend during bake

  736   Fri Oct 11 16:57:23 2013 nicolasLab InfrastructureVacuumHV feedthrough installed, vacuum baking


I put the new HV feedthrough on the cantilever dewar. I borrowed the heat tape from the CTN lab and I am going to bake the chamber over the weekend at 80 deg C, it's currently pumping and heating up.

 cantilever dewar is baking at 80C and pressure is being recorded to C5:VAC-P2_PRESSURE. data logging started at gps time 1065570956. (4:55pm local time Friday).

  735   Fri Oct 11 13:33:22 2013 nicolasLab InfrastructureVacuumHV feedthrough installed, vacuum baking

I put the new HV feedthrough on the cantilever dewar. I borrowed the heat tape from the CTN lab and I am going to bake the chamber over the weekend at 80 deg C, it's currently pumping and heating up.

  734   Fri Oct 11 13:31:56 2013 nicolasMiscVacuumcantilever dewar leak rate data


To measure the leak rate, i valved off the pump and let the pressure rise and wrote down some numbers:


t (m:ss) | P (torr)
0          4.5e-4
30         2.6e-3
60         3.6e-3
2:30       6.5e-3
5:00       1.1e-2     

 2 days later I got this:

t (m:ss) | P (torr)
0          5.5e-4
30         1.4e-3
60         1.9e-3
2:30       3.2e-3
5:00       5.3e-3    
  733   Wed Oct 9 18:40:35 2013 nicolasMiscVacuumcantilever dewar leak rate data

To measure the leak rate, i valved off the pump and let the pressure rise and wrote down some numbers:


t (m:ss) | P (torr)
0          4.5e-4
30         2.6e-3
60         3.6e-3
2:30       6.5e-3
5:00       1.1e-2     
  732   Thu Sep 19 18:58:34 2013 ericqMiscCrackleSystematics

In thinking about systematics, I steered one of the michelson beams off of the PDs, drove the blades, and put the data through my analysis pipeline.

No DC shift in the 2F Q power was observed, and the fluctuations were ~100 times smaller than the offset shown in the last ELOG's plot. This eliminates effects from beam motion on the PD surface (and maybe things like electrical pickup too?) as contenders for significant systematics. 

At this point, the possibly relevant systematic effects that I can think of are:

  • Nonlinearity in the DAC
    • Since the DAC is discrete in nature, the drive force exerted on the blades will not be free of harmonics. This could enhance the DAC noise in a periodic fashion. To get to the level of the observed 2F Q power, the DAC voltage noise during the drive would have to be ~10x higher than in its quiescent state, which is about 1uV/rtHz, and be fluctuating 90 degrees out of phase with the drive signal, to show up in the 2F Q. I can try to estimate this by recording DAC output of the usual drive signal, and filter the data through my blade models. 
  • Nonlinearity in the ADC
    • Though the ADC noise doesn't physically move the masses, in terms of sensing noise, the ADC noise floor is closer to the signal than the DAC noise. If there is a upconversion effect on the ADC input, it could show up as crackle. I don't know much about ADC input problems, other than the need for whitening and AA, which I have rudimentary forms of...
  • Barkhausen noise in the drive magnets
    • Just as the blade is experiencing a slowly varying force, the actuation magnets are experiencing a slowly varying external field from the drive coils. Thus, the magnetization of the magnet can be fluctuating in a crackling fashion, thereby changing the force imparted to the blade. This is hard to calculate, but known to exist at some level...
  • Other physical crackle?
    • Maybe I'm seeing crackling in the base / tip clamps and not in the blade material? Then again, this is not necessarily an error, it just warrants eventual trials of different clamping schemes down the road...

Tonight and tomorrow are my last chances to make any measurements before I fly out of town on Saturday, so I'm racking my brain for anything else that might matter... I'm not sure what else could manifest when the drive is on that would not exist when the drive is off. 





  731   Tue Sep 17 15:26:58 2013 ericqDailyProgressCracklePreliminary Slow Chop results

As mentioned in the previous elog, I wrote a script to turn the Low-F drive off and on every hour (actually was 58 minutes for some reason). 

Here are the results of the 2F Q. The common mode drive was .125 Hz, 800 counts (haven't calibrated into meters yet, but order of microns). Here, I've analyzed the 400-500Hz band of the signal. It looks like something is happening... this really suggests that I really need to think about the systematic effects that may occur when turning on the drive, and estimate their magnitudes. 


  730   Tue Sep 17 03:02:12 2013 ericqDailyProgressCrackleAnalysis Testing

 Much to be ELOG'ed about recent efforts, but for now a quick note about analysis. I wrote a test which took stretches of locked, undriven, data and injected crackle at some level into the signal. I then ran my fourier analysis code on the new datastream, to come up with a number for the crackle power in some bandwidth. I then used the knowledge of the alpha used to create a time series of the crackle signal, and directly integrated the power spectrum to get the crackle power directly. 

Punchline: they match up nicely - my analysis seems to work. For anyone interested, a copy of all my current code lives in /svn/trunk/crackle/MATLAB/ . 

Here I plot the result of the analysis code, along with 95% (2sigma) confidence intervals for its output. I've also included the background (non-crackle) power in the signal in the analyzed bandwidth, to show that we can still recover crackle underneath the background with confidence. This plot was created with simInjection.m in the above link.


I'm currently running the interferometer pretty much every night to collect data at various drive frequencies / amplitudes. Today I wrote a script to switch the drive on and off about every hour, so we can hopefully see a square wave shape in the output of the analysis code. 



  729   Thu Sep 12 00:29:23 2013 ranaComputingDAQX1KRK model restarted


I restarted the DAQ and the KRK model today at about 11AM local time to increase the acquire rate of the accelerometer channels.

The autolocker script died when it couldn't access the epics channels, I restarted it.


  728   Tue Sep 10 11:15:31 2013 nicolasComputingDAQX1KRK model restarted

I restarted the DAQ and the KRK model today at about 11AM local time to increase the acquire rate of the accelerometer channels.

The autolocker script died when it couldn't access the epics channels, I restarted it.

  727   Sun Sep 8 21:51:14 2013 haixingMiscSUSissue of matlab function "margin()" with an unstable plant

I used the matlab function margin() to plot the phase and gain margins for the open-loop transfer function for maglev. It seems to give an incorrect answer. Here is what I got:


As the gain margin is negative, this indicates that the system (plant + controller) is unstable. However, this is not the case.

I used the matlab function nyquist() to make a Nyquist plot, and this is what I got:

The contour circles -1 counter-clock wise once, and this satisfies the Nyquist stability criterion, as the plant (in my case the plate can be modeled as a mechanical object attached to a negative spring) has one pole on the right-half complex plane. Basically, my plant together with the controller in indeed stable, which is also the reality.

Therefore, this seems to indicate that nyquist(), instead of margin() is the right way to examine the stability in the case with an unstable plant in matlab.

  726   Sat Sep 7 19:02:44 2013 ericqNoise HuntingCrackleNew Noise Budget

 New noise levels seem to be holding up!

As mentioned in the last elog, I retook all of the noise budget data, here is everything!



The autolocker is running, I already have a few hours of locked, in-vac, data with simultaneous table accelerometer, which will be the next thing I look at, since the seismic trace still isn't completely correct.  (The trace on the plot is from real table noise data, and modeled stack+blade vertical TFs, no horizontal motion accounted for...)

  725   Fri Sep 6 17:07:39 2013 haixingSummarySUSissues to be investigated

Since the plate is levitating, we are now in the position for real work. Here are the two major issues to investigate in the plan.

1. TF of the plate and cross coupling among different degrees of freedom (important in order to optimize the control servo)

We will measure various transfer functions to characterize the plate TF and cross couplings. We will build six degrees of freedom simulink model based on Georgios's work of three DOFs, and try to make a match between the model and the system.

2. Noise budget (to pin down the major noise source)

(a) sensing and actuation noise

We will calibrate the noise from the hall effect sensor. If it is confirmed to be the major noise, we can switch to the optical lever sensing scheme as planned. The coil is quite weak, in terms of voltage to force conversion factor, and it is 5mN/V. The thermal noise of the coil may not be important (to be confirmed with more rigorous analysis).

(c) acoustic noise

Right now the system is exposed in air, and it is anticipated that the acoustic noise is quite significant. To mitigate this noise, we can use a bell jar to cover it which can give a reasonable level of noise isolation.


(b) seismic noise

We will make a correlation measurement between the sensor output and the seismometer (or accelerometer) to see where the seismic noise dominates.

(d) ambient magnetic field noise

We will use two low-noise honeywell hall effect sensors [link] to measure the ambient magnetic field. To get a better sensitivity, we will use differential measurement by shielding one (together with instrumentation amplifier for amplifying the readout).

(e) thermal noise of the magnet

The major noise source comes from the random jitering of the magnetic moments due to thermal excitation. We can find the literature on how to analyze this kind of noise.

(f) long-term drift

We know little about the long stability of the magnets and also how the temperature drift affects the magnets. This needs to be investigated.


  724   Fri Sep 6 15:47:40 2013 haixingSummarySUSachieving stable levitation of the plate

We levitate the plate by controlling three vertical degrees of freedom---one translational, and two tilts. Right now even without applying controls to the horizontal direction, the system is quite stable.

Here we describe the major steps for achieving levitation:

1. Move the plate close to the equilibrium point by using the DC motors and strain gauge.

We started off from the point where the magnetic force is stronger than the gravity force. Basically, the plate is touching tips of DC motors mounted on the top fixed platform. By using the strain gauge attached to the tips, we can tell how how far we are away from the equilibrium point where the strain gauges are not stretched. We slowly actuate the motors to push the plate close to the equilibrium point.

2. Lock one degree of freedom

We started from some generic PID controller. After many random trials, we end up with a controller that was barely doing the work, and the system was marginally stable. We then measured the closed-loop transfer function of the system and use simulink to model this one degree of freedom: model_1DOF_v0.slx and model_1DOF_script_v0.m

After tuning various parameters, we got a reasonable match between the model and the measurement. In particular, the parameters we found go as follows:

mass    = 0.5 kg        % mass
K          = -180 N/m   % negative spring constant
Vm       = 250 V/m    % hall effect sensitivity
Gby      = -27dB       % the bypass gain from the coil to the hall effect sensor
Gcomp = -39.5dB    % the residue bypass gain after compensation

To obtain a better controller, we tried to use "sltunable" class in the matlab, in particular using the functions "systune" and "looptune" by specifing the target phase and gain margin. Somehow, it did not produce the desired result. We found out that the calculated phase margin and gain margin are not correct (we will explain this in more details with another elog entry). We then used "nyquist" function to design the feedback loop and we used the following parameters for the controller, which gives a reasonable phase and gain margins.

DGain = -4*pi*13;                 % the constant gain
Dzs     = [-2*pi*0.5; -2*pi*3]; % zeros for the controller
Dps     = [0;-2*pi*13];            % poles for the controller

After applying this new controller, we got a quite stable levitation. We redid the open-loop transfer function measurement. The agreement between the model and the system is shown by the figure below (the blue line is the model and the green line is the measurement data):


The difference around 4Hz and 20Hz could arise from a simplified model for the compensation (assuming a constant compensation gain, and the reality is more complicated), and we need to refine the model.

2. Repeat the same process and tune the feedback controllers for the other two degrees of freedom.

3. Slowly ramp up the gain for the other two controllers and finally levitate all three DOFs.

We found that if we engage the controllers abruptly, the system will rapidly destabilized. Instead, we had to slowly ramp up the gain of the controllers so as to approach the final stable state softly.

The figure below shows the signal from all seven sensors (one redundant in the horizontal direction).


We can see the resonant frequency is around 2Hz (to be confirmed by future TF measurements), which is quite high. After fully characterizing the system, we will need to tune the DC magnet force with DC control coils to make the equilibrium point closer to the force maximum, where the rigidity is low.


Here is the link to the video of the levitating plate:  link


Attachment 3: data_model_comparsion_1DOF.png
  723   Fri Sep 6 12:46:48 2013 ericqNoise HuntingCrackleTentative Good News

 Last night, after putting back the newly glued picomotor mount, and making extra sure are cable were well behaved, I noticed the free swinging michelson signal looked much slower than before, which was promising. However, every time I aligned, it slowly misaligned; I figured the ultra-soft damping rubber was still settling. Overnight, it slowly reached a steady state (looking at the shadow sensor signals over the night, it took a few hours). 

Today, the michelson easily locks. I didn't get the UGF very high (~50Hz or so), because some of the odd plant TF features have moved around, so I'll have to do some compensation filtering to get as high as I have in the past. However, I though it would be more useful to just go ahead and look at the noise. Here's what I did (which is pretty straightforward, but included for completeness):

  • Locked the Michelson with servo+boosts
  • Measured the Loop TF in DTT, by injecting a swept sine immediately after the servo output, and looking at drive output / servo input TF. 
  • Established the UGF, recreated my servo, plant and loop TFs in MATLAB
  • Recorded 1 minute of locked michelson data, used fringe to fringe counts of servo input to convert to displacement
  • Got power spectrum via pwelch, multiplied by magnitude of 1-G
  • Plotted on top of latest noise budget. 

Here's the plot! (The rolloff at 2kHz is due to the fact that I analyzed data at 4k, whereas the older traces were spectra of live 64k data)


So, I may have had a huge win... However, given the magnitude of difference, I want to be extra sure, so I'm going to spend the rest of the day retaking data for the noise budget curves and making sure I'm not missing any multiplicative factors anywhere... I think I can squeeze in an accelerometer during an in-air measurement too.

Attachment 1: goodnews?.pdf
  722   Thu Sep 5 17:19:02 2013 ericqNoise HuntingCrackleNo blade isolation

EDIT: Koji points out that my attempt to see blade isolation is misguided. If the blade is isolating, the razor is steady in the inertial lab frame, thus the shadow sensor, which is rigidly mounted on the stack, will measure stack motion, not blade motion. Thus, the TF from accelerometer displacement to SS signal would be unity. So, I cannot conclude that my blades are not isolating. 

Some uninteresting headaches have been taking up time (broken picomotor control pad, broken picomotor epoxy joint, cymac being fussy)

Though I thought I had fixed my cable routing, loading up the stack with optics and blades shifted the stack in a position in which shorting was occurring again. I replaced the second set of rubber with the "ultra-soft" rubber; though it's resonant frequency is higher, leading to less 1/f^2 isolation, its Q is much much lower, which I hope will reduce the noise in the tens of Hz (which vastly dominates my RMS motion), possibly at the expense of noise further up. 

While I'm waiting for epoxy on a picomotor mount to dry, I decided to simultaneously measure a vertical accelerometer on the table, one on the stack and the two shadow sensors aligned to the razor blades mounted on the blade tip. 

While the stack TF looks ok, my blades do not seem to be doing any vertical vibration isolation. You can see the ~2Hz resonance of the blade, as well as a parallel mode around 5-6Hz which has often been seen before, but no rolloff after that...  (The frequency region plotted is where there is reasonable coherence between the signals, a wiener TF looks similar for the second plot)


A better test of this will be running an accelerometer on the stack along with a locked ifo, I'll attempt this next. 

  721   Sun Sep 1 18:15:10 2013 ericqNoise HuntingCrackleStack Isolation Material TFs

Rana ordered various kinds of rubber to see if I could get a lower resonant frequency of the stacks. I set up two accelerometers, one on the table, one on a stack, and took one minute of data, and did some wiener filtering to figure out the table-> stack motion transfer function. Setup looked like this (posts added for extra mass):



The materials were:

Super-Soft Neoprene Rubber, 3/8" Thick, 6" X 6", 20A Durometer  

Ultra-Soft Polyurethane Strip, 3/8" Thick, 2" X 36", 50 Oo Durometer  

Ultra-Soft Polyurethane Sheet, 1/2" Thick, 12" X 12", 40 Oo Durometer

My current stack rubber (material properties unknown)

Some foam that I found in a cabinet, picture:2013-08-30_18.55.30.jpg

For the first three materials, I had to stack two sliced rectangles to approximate the same height as the current stack rubber. Coherence from 5-60Hz was good on all of the measurements, here are the TFs:


Unfortunately, none had a lower resonant frequency... Though the "Ultra Soft" materials have high damping, so it may be useful to combine rubbers. I've also ordered some "Super-Cushioning Polyurethane Foam" cylinders to test as well. 

  720   Fri Aug 23 19:18:11 2013 ericqNoise HuntingCrackleSeismic Noise

I'm having trouble getting good TF estimates from Wiener filtering, I think because the ranges of high coherence are kind of narrow. Additionally, given that the displacement spectra of the table and of my ifo looked like the stack wasn't doing much, I was suspicious that my inattention to cabling led to a mechanical short of the stack.

So: I ripped everything out and put the accelerometers directly on the stacks!

Here are some displacement spectra, and a vertical "TF"s of the stacks obtained by dividing the noise spectra. 



 In the 1 plate case, I had the chamber lid open, which led to the broad peak at ~120 Hz. 

In any case, I see peaks at ~9Hz And ~24Hz and more isolation that I was getting before. I'm going to put things back together much more carefully. 



  719   Fri Aug 23 16:25:06 2013 Ed Taylor and Nic SmithSummaryRingdownLog plot and log fit of decaying ringdown for the 3rd mode

I created a log plot for the decaying data and fitted linearly to the data.

Yechh!!  JPG, no error bars, no details of fit ??!!!

Attachment 1: LogofRingDownMode3.jpg
  718   Fri Aug 23 15:13:26 2013 Ed Taylor and Nic SmithSummaryRingdownExponential Fits to Ring Down Data

When performing the ring down, we obtained sets of data which contained time varying votlage of the signal recorded from the split photodiode. We utilized the pwelch function in Matlab and obtained a PSD for each data set. The decaying peak of the resonant frequency was plotted seperately which then formed a decaying exponential. I then applied a fit to each set of data.

Attachment 1: Tao.pdf
Tao.pdf Tao.pdf Tao.pdf
  717   Fri Aug 23 14:09:09 2013 Ed Taylor and Nic SmithSummaryRingdownComsol Frequencies and Observed Resonant Frequencies

The observed resonant modes of the oscillating cantilever differed substantially from frequencies computed through comsol (more so at higher frequencies)

Attachment 1: Comsol_Frequencies.pdf
ELOG V3.1.3-