40m QIL Cryo_Lab CTN SUS_Lab TCS_Lab OMC_Lab CRIME_Lab FEA ENG_Labs OptContFac Mariner WBEEShop
  Cryo Lab eLog, Page 57 of 61  Not logged in ELOG logo
ID Date Author Type Category Subject
  250   Tue Jul 19 00:36:29 2011 FrankCryostatSensorscryo connector test

started testing how to fabricate the cryo cables/connectors and sensors. The wires are super thin (AWG 36) and not very flexible (phosphor bronze). They behave more like piano wire .

Anyways, disassembled one of the standard 4-pin Lemo 0S sockets to remove all the brass parts. I'm still uncertain about using the chrome plated brass parts.
First of all , brass is outgassing in high vacuum. On the other hand it's chrome plated. Chrome is OK.  But brass has a high zinc content.
The problem is that the high zinc content can cause the brass part  to grow zinc whiskers which can be several mm long and create shorts (link).
So if possible i would like to get rid of that stuff too. The insulator is PEEK.

Below a picture of one of the sockets easily took apart within a minute using the drill press and a file.
Switched to 0S series as without the housing there is (almost) no keying left and so the standard 4-pin connector is hard to mate correctly (4 possibilities, everything is rotational symmetric).
However even blind people can connect the 0S series correctly.

P1030829.JPG

P1030832.JPG

 

  249   Mon Jul 18 02:35:54 2011 FrankLab InfrastructureCryoadditional bottle rack installed

we got the second bottle rack for the Helium bottle installed on Friday. It's located next to the Hydrogen bottle and the flamable liquid cabinet near the entrance.

P1030837.JPG

  248   Sun Jul 17 22:32:14 2011 DmassDailyProgressGeneralAsymmetric Michelson Setup: COVEGA

I set up the Michelson again after swapping the Emcore diode (higher linewidth) out for the Covega Diode (lower linewidth).

Setup:

  • I changed to the Newport 818-BB-30 (2 GHz PD)
  • I used the RF power splitter (DC to 4.2 GHz) at the output of the PD
  • Electronics were the same as in elog:238
  • I'm unsure about impedances in elog:238 and how they effect gain, so I measured the signal @ scope and power @ PD
  • 12.27mW  = 112 mV at scope, as measured by the Thorlabs Powermeter This is dumb and wrong - PD was saturated
  • Newport 818 has a saturation current of 5 mA (0.25V with a 50 Ohm load)
  • Assuming ~1A/W responsivity, this gives us around 4mW input max
  • I set the signal max to 38 mV at the scope using a waveplate and a PBS (so ~4mW at PD, less than saturation current at 1A/W)
  • Vpp = 38.4 mV
  • Vo = 2 mV
  • Contrast = (Vpp)/(Vpp+Vo) = 95%

 

  247   Sun Jul 17 21:16:34 2011 FrankCryostatCryoindividual cryo connectors

The current idea is to have two 25pin connector anchored to the cold plate (one per cavity) and running cables from there to the heaters and sensors attached to cold plate and cavity & shields using small individual connectors where useful.

As mentioned earlier in post #185, we would like to have individual connectors as a nice advantage of the current design is that we can remove the cavity from inside the shield or with the inner shield without having to remove the rest.
Several cryo companies sell standard inserts from Lemo and Fischer connectors made from PEEK as low temp connectors (e.g see here).
As we will have one or multiple sensors on every single shield and on the cavity itself to characterize our thermal design we would like to disconnect individual parts without having to disconnect/remove everything.
So we need some small cryo connectors, preferably 2-pin and 4-pin versions for individual sensors and heaters(?).

I wanted to go for the crimp version as it is much easier to fabricate and we don't have to clean each pin so much. However i didn't pay attention to the minimum size of the wire for the standard crimp version, which is 26 AWG (4-pin) or 24 AWG (2-pin).
However our cables have a size of 32 AWG (quad twist for temp sensors) or 34 AWG for copper wires (for heaters). Even with the special crimp version for smaller wires we probably would have to solder at least the sensor connection. Minimum size is AWG 32 for the smaller pins (see specs for crimp contacts below).

unipole_multipole_Page_009.jpg

Below the part numbers for the connectors (spare parts). We can also use inserts from all standard parts.

Spare_parts_U-M_ang_Page_12.jpg

Spare_parts_U-M_ang_Page_13.jpg

  246   Fri Jul 15 20:07:38 2011 FrankMiscCryohoists for cryostat

As suggested by Warren i searched the web for belljar hoists to raise and lower the cryostat without too much tilt so that the cavity does not slip on the supports.
I found only one company so far:

http://www.lesker.com/newweb/chambers/Motorized_Hoists.cfm?

  245   Fri Jul 15 19:09:27 2011 DmassDailyProgressGeneralAsymmetric Michelson Setup: EMCORE

Quote:

What: An asymmetric Michelson to measure the frequency noise of the laser diodes

Why: I tried to measure the frequency noise of the diodes by locking them to a PMC, but was unable to suppress the error signal into a linear range

Equations: A little algebra on the output of a Michelson:

  1. Vout ~ sin^2(theta / 2)=sin^2(k * deltaL)
    • theta = phase difference between the fields at the recombining beamsplitter of the Michelson
    • k = 2 * pi * f / c
    • deltaL := the difference in arm length of the Michelson
  2. For my tabletop IFO: Vout = Vo + Vpp / 2 * (1+cos(theta)) = Vo + Vpp * sin^2(theta / 2)
    • Vpp is the peak to peak signal of the interferometer
    • Vo is the minimum signal (due to dark voltage on PD and contrast defect)
  3. dVout / dtheta = Vpp / 2 * sin(theta)
  4. theta(Vout) = arccos(2 * [Vout-Vo] / Vpp - 1)
  5. And of course, dVout / df = dVout / dtheta * dtheta / df
  6. dtheta / df = 4 * pi / c * deltaL
  7. dVout / df = Vpp * 2 * pi * deltaL * sin(theta) / c
  8. Which is what we expect, a common mode rejection of frequency noise based on how well we balance the arms

 

 I set up the asymmetric Michelson with the EMCORE diode after doing the symmetric.

What Was Done:

  • I measured the waist position / location coming out of the beamsplitter using the razorblade/PD set w.r.t. a location I marked on the table
Location (in) Size (um)
2 228
3 246
4 270
5 316
  • I fit a guassian beam to the above with MATLAB and got the following beam parameters:
    • wo=220 um
    • wlocation = 1.1 inches
  • I checked what I could do with available lenses to make a waist 1/2-1m past the point I marked (~20-39 inches)
  • If I put a 400mm lens at +18 inches, I get a waist at ~29 inches
  • I put a flat mirror at +29 inches in an attempt to spatially mode match my asymmetric Michelson
  • I realigned my now asymmetric Michelson
    • Contrast = (3.06-0.332)/(3.06-Vdark+0.332-Vdark) = 81%
    • Vdark is the dark offset of the PDA10CS = 15mV
  • I took some spectra with the 2GHz 50 Ohm output PD, but realized it was saturating my PD (5mA was saturation)
    • At the nonsaturated level, assuming 100% contrast, the input noise of the 4395 (10nV/rtHz) would be at 40Hz/rtHz
  • I switched to the PDA10CS to get better gain on the signal below 20 MHz

Data:

  • For the first spectra I used a DC-4.2GHz splitter to split the signal as shown in elog:238. The rest of the electronics were identical to the picture
  • Contrast:
    • Vpp = 712mV
    • Vo = 96 mV
    • Vdark = 8 mV
    • Contrast = (712 - 96)/(712 + 96 - 16) = 78%
  • I also blocked each arm and took spectra to see if the RIN + jitter coupling showed up as a noise source anywhere
  • Each spectra was taken the same way as described in elog:238, by looking at the scope output while taking the spectra until it was ~ mid-fringe for the duration of the measurement

Michelson Phase Noise Spectra:

Filename Vmin [mV] Vmax [mV] Vavg [mV] Theta(V) [rad] dV/dtheta [V/rad]
AM3H1.TXT 152 392 272 2.10 0.307
AM3M.TXT 288 568 428 1.64 0.355
           
  • The phase noise was so high that I had to increase the internal attenuation of the input channel to 20 dB. This effectively raises the input noise of the device to 100 nV/rtHz

Blocked Arm Spectra:

Filename Vmean [mV] Theta(V) [rad] dV/dtheta
AM3B1H
216    
AM3B1 216    
  •  The noise from the 1-arm blocked setups were both below the input noise of the 4395 (100nV) in the range of the measurement, I left the attenuation set as in the measurement.
  • This corresponds to a RIN of 5*10^-7
  • The Emcore high frequency RIN is 10^-7 (elog:188), so we don't expect to actually see this is in our Michelson measurement
  • The RIN+jitter spectra (with 20 dB input attenuation) are almost indistinguishable from the 4395 input noise, so I won't bother posting them.

 The phase/freq noise spectra are shown in elog:225.

 

If we compare the shapes of the high frequency spectra taken in elog:225 (asymmetric Mich) with those taken in elog:238 (symmetric Mich), we see similar features, with the following attenuation:

  • Symmetric 10 MHz Mich phase noise = 10 nrad/rtHz
  • Assym Mich 10 MHz Mich phase noise = 100 Hz/rtHz x dtheta/df = 4*pi*deltaL / c = (100Hz * 41 ns)/rtHz = 4.1 urad/rtHz
  • This factor of 250 suppression would mean that my Michelson asymmetry was 1/250m = 4mm in the symmetric case. This seems like a reasonable number.

 

  244   Fri Jul 15 17:24:53 2011 DmassDailyProgressGeneralThorlabs PDA10CS Transfer Function

My PDA10CS transfer function seemed fairly smooth in its amplitude reponse, so I made a short MATLAB script to interpolate it I could use it for calibrating different spectra I took using it as a detector.

I assumed that the response below 100kHz quickly asymptoted to its DC response, and thus set its value below 100kHz this to ~0dB. Since all my calibrations for the Michelson response were done at "DC" (my pedestal poke frequency was somewhere around 1 Hz), this should be sufficient to calibrate out my PD response.

Relevant MATLAB code:

function [amp]=interp10CS(freq)
% Input must be between 100kHz and 30 MHz - amp is given in dB
cd /Users/dmass/svn40m/trunk40m/docs/CryoLab/FullDiodeChar/MichMeas
TF10CS=load('TF10CS.TXT');

% Parse the PDA10CS transfer function into amplitude and frequency - +10dB
% should give DC value of 0dB - desired

lengt=length(TF10CS);
AMP10CS=TF10CS(1:lengt/2,2);
fre10CS=TF10CS(1:lengt/2,1);


if max(freq) > 3e7
    amp='Freq too big'
else
    amp=zeros(length(freq),1);
    for i=1:length(freq);
        if freq(i)<1.1e5
            amp(i)=0;
        else
        indexup=min(find(fre10CS>freq(i)));
        indexdown=indexup-1;
        fup=fre10CS(indexup);
        fdown=fre10CS(indexdown);
        % This is the log (in freq) interpolation below:
        amp(i)=9.5+AMP10CS(indexdown)+(log10(freq(i))-log10(fdown))/(log10(fup)-log10(fdown))*(AMP10CS(indexup)-AMP10CS(indexdown));
        end
    end
end

  243   Thu Jul 14 18:57:23 2011 JennyUpdatePlotsMore MATLAB curve fitting

I tried fitting the same curve (radiation and conduction in a simplified cavity setup) in MATLAB to a product of exponentials. 

Two equations I tried:

fitequation1.jpg

 fitequation2.jpg

where u(x) is the Heaviside step function. The second one simply allows for a time delay of t_0 in the second exponential.

I used fminsearch again with these new functions and got out very similar fits.

Nodelay:fitparams1.jpgexp_product.jpgDelay:fitparams2.jpgexp_product_delay.jpg

 

 Essentially no difference. Also, there's very little difference between these fits and the single pole fit I did before. I tried to do weighted fitting like DMass suggested (multiplying my error vector by 1/time) but I couldn't get it to produce a good fit.

 

  242   Thu Jul 14 17:38:47 2011 FrankMiscmaterial propertiesCommon Indium-contained alloys and their properties

Common_Indium-Contained_All2.jpg

  241   Thu Jul 14 15:12:43 2011 JennyDailyProgressSimulationSimple model of room temp cavity

 Update: More exciting video (by which i mean the same video with some Bach music added and a slightly faster frame rate)

  240   Thu Jul 14 02:52:56 2011 FrankUpdateDrawingsCAD drawings updated

CAD drawings are now up to date on the svn including all mates and fasteners (outer shield to cavity, no cold plate support yet)

the following details are still missing:

  • long holes for frame length adjustment and mounting
  • holes for accessing set screws for fixing cavity from outside
  • cold plate support structure
  • location/holes for temp sensors/heater cables
  • location/holes for temp sensors

other not important features:

  • material properties for fasteners
  • set screws for frame
  • set screws for fixing cavity

 

  239   Thu Jul 14 00:48:43 2011 FrankHowToCavityshield support

during the meeting this afternoon we had a discussion about how to support the shields the right way (or most clever one). Here a summary of my thoughts:

The following thoughts take only bending of the structure into account, not the mechanical oscillators build by the individual parts.

As we have to thermally insulate the shields from each other and the cold plate we have to support them at small distinct points.
As i don't have the time to do a full mechanical simulations my intuition says that we should support everything close at it's airy points, the same way as we support the cavity to minimize absolute length change.
A better way would be to support them along the full length but the thermal conduction would be too good.
If we support them close to their airy points we minimize the sensitivity so vertical acceleration taking the whole length into account (at the center of the structure)
In fact for the onion type structure the inner part is a little bit smaller so to minimize a length change for the inner part the position is slightly different and we don't care about the center of the shields, we care more about the bottom where we support everything. I don't know how an actual length change of the shields couple into length of the cavity, but reducing it to a minimum can't be bad.
The cavity itself is resting loose on the support points, so the distance between those points will change changing the coupling directly into the cavity from vertical acceleration of the whole structure. This effect might be small.

Anyway, the following rule of thumb is valid for supporting bars (link). The minimum length change is achieved in the case of a symmetrical spacing of the Airy points from the center, approximately 0.577 of the length of the structure.
This agrees with the papers on how to support the cavity the right way and my own simulation for our RT and cryo cavities.

This is approx how the current support of the shields and cavity is designed. The structure does not allow supporting everything at it's airy points as the supports have a finite thickness and we need some space to adjust the spacing for the cavity support.
So some of them are slightly off.

 

Here a nice thesis on mechnical designs: https://e-reports-ext.llnl.gov/pdf/235415.pdf

 

  238   Wed Jul 13 17:22:52 2011 DmassDailyProgressGeneralSymmetric Michelson Measurement

 

The first thing to do seemed to be to get a measurement of the symmetric Michelson length noise.

Setup:

MichelsonS.png

Procedure:

  1. I used a BNC T to split the signal
  2. I balanced the length of the arms by eye, to maybe ~cm matching
  3. I used a PDA10CS as "PD"
    • Checked the lens makes a reasonable spot at the PD with razor scan (as described in elog:162)
    • wo = 142 um @ PD
    • PD diam = 1mm
  4. I got the beams to overlap via eye
  5. I poked one of the end mirror posts and tuned the alignment to increase the contrast
  6. Vpp = 904 mV    //    Vo = 28 mV    //    Vdark = 8 mV  (Vdark is offset with no light on PD, Vo is min voltage with light on)
    • Vmax = Vpp+(Vo-Vdark) = 904 + 28 - 8 = 924 mV
    • Vmin = Vo-Vdark = 28 - 8 = 20 mV
    • Contrast = (Vmax-Vmin)/(Vmax+Vmin) = (904)/(944) = 95.7%

Data: 

Filename Vmin [mV] Vmax [mV] Vavg [mV] theta [rad] dV/dtheta [V/rad]
LNZ3H.TXT 552 624 588 1.33 0.439
LNZ3M.TXT 672 784 728 0.99 0.378
LNZ3L.TXT 520 696 608 1.28 0.433
  •  All calibration as described in quoted elog
  • dtheta/dL = 4*pi*f/c = 4*pi/lambda
  • dL/dtheta = 1.55 um/(4*pi) = 123 nm/rad
  • The data was taken by watching the Michelson output on a scope while taking spectra, and retaking spectra until the Michelson was ~mid fringe for the duration of the spectrum.

Plots:

Here is the voltage noise spectra taken as described (click for a high res PDF)
SymMichVolts.pdf

and calibrated (calibrated for Michelson response and PDA10CS response) into phase noise of the Michelson:

SymMichRads.pdf

The mid trace looks ~a factor of 2 lower than the others, which merits a quick sanity check to make sure this isn't wildly inconsistent with anything else I know:

I looked through my notes and thought about it.  I don't think this was some missing or additional 50 Ohm terminator. I think this can be explained as follows:

The middle trace was taken farther off mid fringe than the others (see table for range of voltages), which would cause it to have lower sensitivity, and show up

as a lower voltage (which only explains the uncalibrated difference). For the calibration, since I didn't actually look at the average value and just recorded the

min/max, if the signal was higher during this time (closer to dark), the average would underestimate the dtheta/dV calibration.

To be quantitative, the dV/dtheta calibration at the high point of 784 mV is a factor of sin(arccos(2*(784-28)/924-1)) = 0.77 lower than at mid fringe.

This wouldn't explain the factor of two.

In summary: I trust the calibrated Michelson phase noise levels to a factor of 2

Attachment 3: SymMichRads.pdf
SymMichRads.pdf
  237   Wed Jul 13 15:30:20 2011 Warren JohnsonCryostatDrawingstwo cavity support structure

Here is a .pdf drawing of a proposed support structure for the thermal shields and the cavities.   It is an aluminum frame which bolts to the bottom of the top plate.  The VectorWorks drawing file is appended.

cavity_mount_7_10.pdf

Attachment 2: cavity_mount_7_10.vwx
  236   Wed Jul 13 15:01:51 2011 FrankUpdatePurchasesall orders checked

checked all open orders. Everything is on time or only delayed by a few days except for the dual power supply which is still missing (we got the high voltage one and the rack mount kit).

The dual 30V-3A supply is currently backordered with a delivery date of Aug 24th. They quoted it to be in stock but sold all of them before our order was placed.
I don't know if it makes sense to look for an alternative place which has it in stock as we don't need this super urgent and we can always borrow something from the PSL for a few weeks.

There are also some special nuts missing to mount them to the rack kit. I talked to the company and they figured out that they are missing in all brand new kits they ship . So will have to wait for them as well.

  235   Wed Jul 13 11:33:38 2011 JennyDailyProgressPlotsCurve fitting in MATLAB

I did a single-pole fit of my temperature-evolution data from COMSOL. I generated a plot much like the IGOR plot in my last entry, but using MATLAB this time.

 

Here is my code:

cav_vac_probedata.m

%Loads comsol point probe temperature data and creates two vectors from
%the data: timz and temp

clear all
probedata=load('6-30_cavity_vactube_probedata');

lent=length(probedata);
ctr1=0;
ctr2=0;
timz=zeros(1,301);
temp=zeros(1,301);
for i=3:lent
    if (rem(i,2)==0)
        ctr1=ctr1+1;
        temp(ctr1)=probedata(i);
    else
        ctr2=ctr2+1;
        timz(ctr2)=probedata(i);
    end
end

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

onepolefit.m

function sse = onepolefit(params,Input,Actual_Output)

%This f'n takes as its inputs
%three parameters (boundary temp, temp step, and time const.),
%a vector of x-values,
%and an associated vector of y-values.

%It then generates a vector of fitted y-values by applying
%an exponential approach function (single pole), with the given parameters,
%to the x-values.

%It then returns the sum of the squares of the entries of the difference
%between the fitted y-vector and the actual y-vector

T_0=params(1);
A=params(2);
tau=params(3);
Fitted_Curve=T_0-A.*exp(-Input/tau);

Error_Vector=Fitted_Curve - Actual_Output;
% sse stands for sum of squares error
sse=sum(Error_Vector.^2);

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

onepolefitting

function [outputs] = onepolefitting(startparams, xvals, yvals)

%This f'n takes as its inputs
%three starting guess parameters (boundary temp, temp step, and time const.),
%a vector of x-values,
%and an associated vector of y-values.

%It then calls the onepolefit f'n and minimizes its output (sum of squares error)
%using the fminsearch f'n. It outputs info on iterations and the
%fit paramters settled on.

%It then plots the input data, the fitted curve, and the residuals

options=optimset('Display','iter');
fitparams=fminsearch(@onepolefit,startparams,options,xvals,yvals);
fityvals=fitparams(1)-fitparams(2)*exp(-xvals/fitparams(3));
residz=yvals-fityvals;

figure(1)
hold off
clf
subplot(2,1,1)
plot(xvals,yvals,'x')
hold on
grid
plot(xvals,fityvals,'r')
xlabel('Time (s)')
title('Cavity Response to Vactube Temp Step and Single Pole Fit')
ylabel('Temperature (K)')
legend('Cavity Center Probe','Single Pole Fit',...
    'Location','SouthEast')
hold off

subplot(2,1,2)
plot(xvals,residz,'*')
hold on
xlabel('Time (s)')
title('Single Pole Fit Residuals')
ylabel('Temperature (K)')
legend('Residuals',...
    'Location','SouthEast')
grid

outputs=fitparams;

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

In the command line I ran

cav_vac_probedata

guessparams=[295.1 0.1 7200]

%First entry is initial temp, second entry is temp step, and third is my guess for the time constant from eyeballing the data

onepolefitting(guessparams, timz, temp)

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

The fit parameters it settled on were:

Boundary temp: 295.1 K

Temp Step: 0.1 K

Time Const: 7132.4 s

Here is my figure:

 6_30_onepole_matlab.jpg

Next Step: Find a better fitting function (two, maybe three poles) and try to get the residuals to go to zero.

DYM: If you think you have one low pole which is dominant, you can apply some sort of weighting to treat the earlier time points more importantly for your fit than the later time points, maybe multiply the terms in your error vector by 1/time

  234   Tue Jul 12 16:33:52 2011 DmassDailyProgressGeneralAsymmetric Michelson Setup

What: An asymmetric Michelson to measure the frequency noise of the laser diodes

Why: I tried to measure the frequency noise of the diodes by locking them to a PMC, but was unable to suppress the error signal into a linear range

Equations: A little algebra on the output of a Michelson:

  1. Vout ~ sin^2(theta / 2)=sin^2(k * deltaL)
    • theta = phase difference between the fields at the recombining beamsplitter of the Michelson
    • k = 2 * pi * f / c
    • deltaL := the difference in arm length of the Michelson
  2. For my tabletop IFO: Vout = Vo + Vpp / 2 * (1+cos(theta)) = Vo + Vpp * sin^2(theta / 2)
    • Vpp is the peak to peak signal of the interferometer
    • Vo is the minimum signal (due to dark voltage on PD and contrast defect)
  3. dVout / dtheta = Vpp / 2 * sin(theta)
  4. theta(Vout) = arccos(2 * [Vout-Vo] / Vpp - 1)
  5. And of course, dVout / df = dVout / dtheta * dtheta / df
  6. dtheta / df = 4 * pi / c * deltaL
  7. dVout / df = Vpp * 2 * pi * deltaL * sin(theta) / c
  8. Which is what we expect, a common mode rejection of frequency noise based on how well we balance the arms

 

  233   Tue Jul 12 01:05:29 2011 FrankMiscCavitycavity support - some more pictures

here some more pictures of the support frame with and without the cavity. The cavity is sitting on small interchangeable rods where we can easily change the conductive heating by interfacing area or conductivity of the material chosen.

Those pieces can be either glued using the varnish or screwed to the frame as illustrated.The distance between the support points can be adjusted with the frame and locked.

The cavity can be placed inside the frame and fixed from the top using a setscrew on each ring. The frame (including the cavity) is then inserted into the inner shield.

The setscrew can be reached trough holes in the shield from outside to ones everything is in place we can unlock the cavity if we want. The current plan is to make the frame out of SS for less conductivity from shield to cavity.

Differential thermal expansion might be a problem which has to be checked. Total weight of one cavity including everything: 2.34kg.

Any comments are welcome. The plan is to submit everything for fabrication mid this week.

.cavity_frame.JPG.

cavity&frame.JPG

all.JPG

Attachment 2: cavity&frame.JPG
cavity&frame.JPG
  232   Sun Jul 10 23:44:34 2011 FrankCryostatCavitythermal shields+cavity support - drawings for Warren

these drawings are for Warren for further calculations and discussion only. So no further detailed explanation at this time.

warren1.png

warren2.png

warren3.png

warren4.png

warren5.png

warren6.png

3.5in_shield_support.jpg

 

 

  231   Sun Jul 10 21:03:59 2011 dmassDailyProgressLab WorkTo Do
  • 3040 back to MIT Shipped
  • Cryostat status update
  • IR viewer
    • Summarize what we know and whats out there so we can grab one sensibly
  • Gold PD tuning / real meas
    • Look at where we are going to use them in our setup to choose an intelligent sideband frequency
    • Tune to this freq
    • Measure noise / transfer functions
  • Ship windows to AP for coating
  • Update ganttchart on svn
  • Update purchase spreadsheet
  • Indium - get it
  • LN2 - call
  230   Sun Jul 10 13:53:04 2011 FrankNotesCryoThermal conductivity of cryogenic UHV materials

Some links to documents containing data for thermal conductivity of several UHV compatible materials which are not on the wiki so far (PEEK, Macor, Vespel etc.)

http://reference.lowtemp.org/Woodcraft_LTD13_materials.pdf

  229   Fri Jul 8 13:16:08 2011 DmassDailyProgressGeneralThorlabs PDA10CS Transfer Function

In the course of calibrating the output of my Asymmetric Michelson, I measured the transfer function of the Thorlabs PDA10CS using the Jenne Laser setup at the 40m (as described in elog:219) - I didn't have the splitter on the output of the 1611 when taking this transfer function for calibration purposes.

From one of Alberto's measurements, the 1611 has the following properties:

  • DC/RF output impedance: 50 Ohms
  • DC Transimpedance: 1e4 V/A (1MOhm referred)
  • RF Transimpedance: 700 V/A (50 Ohm)
    • I am interpreting the pink text as the load with which the transimpedance was measured - I think this means that when the RF is plugged into a 50 Ohm load, I still get 700 V/A

The Thorlabs PDA10CS lists its transimpedance:

  • 750 V/A under a 50 Ohm load at 0dB gain setting (which was used for this measurement)

The calibration I used:

  1. We need the ratio of photocurrent I_1611 / I_10CS
    1. This may not be 1 even if the beamsplitter is perfect, because of different QE's / reflection loss
  2. (Vdc_1611 / Zdc_1611) / (Vdc_10CS / Zdc_10CS) = I_1611 / I_10CS
    • Vdc_161 =  -2.29V (inverted)    ( Measured with MultiMeter)
    • Vdc_10CS = 0.142V                ( Measured with MultiMeter)
  3. I_1611 / I_10CS = (2.29 / 1e4) / (0.142 / 750) = 1.21
  4. What I measured was Vrf_10CS / Vrf_1611 in dB, so to get Zrf_10CS:
    • (Vrf_1611 / Zrf_1611) / (Vrf_10CS / Zrf_10CS) = I_1611 / I_10CS
    • (Vrf_1611 / Vrf_10CS) * (Zrf_10CS / Zrf_1611) = (I_1611 / I_10CS)
    • Zrf_10CS = (I_1611 / I_10CS) * (Vrf_10CS / Vrf_1611) * (Zrf_1611)
  5. Zrf_10CS = (1.21) * 10^(Transfer Function / 20) * (700 V/A)

 Here is the spectrum that I measured:PDA10CSTFcal.pdf

The DC levels most definitely do NOT line up. I am not sure if it is my calibration, or if the 700 V/A transimpedance gain for the 1611 is just the gain of the inverting amplifier, and I lose a factor of two from having a 50 Ohm load on a 50 Ohm output impedance...so maybe as measured by the 4395 I get 350 V(at 4395) / Amp(out of PD). Maybe there is something even more boneheaded that I missed though.

None of this matters for my immediate purposes, since I calibrated my Asymmetric Michelson as Volts on the PDA10CS, and IF I assume that the PD transimpedance is featureless below 10kHz, I know everything I need to know even if I don't have the response in Ohms

  228   Thu Jul 7 23:07:48 2011 FrankCryostatCryocavity support/ radiation shield

as we don't have reliable data on how we have to design the radiative shields exactly to get enough attenuation from the temperature fluctuations of the heated part my current design will have two simple radiative shields and a simple cavity support for the first iteration. My current plan is to have two shields (as we have the space for more than one) and we can try several different configurations of how and where we heat the shields/cavity which i will describe later.

Limited by the Cryostat design/dimensions, the first shield will have the maximum diameter limited by the beam height above the experimental chamber bottom plate and the distance between the two cavities giving us maximum space at this time. The distance to the bottom of the chamber is 2.5", the distance between the two cavities is 5" (so also 2.5" max per shield). However we have much more space to the cold plate for suspending/isolating the shields from there. So we don't have to take this into account for the shields design right from the beginning.

Assuming a circular cross section of the shield (a simple round tube for a start to keep it simple and cheap) the maximum diameter will be around 4.5" to have some space between the two cavity shields and the shield and the bottom plate. I picked 4.5" diameter for the outer shield having a wall thickness of 1/8" at the moment as we have that material in the workshop basically for free. The tubes will be made out of Aluminum 6061-T6. I first wanted to go for Copper but the weight of the shields would be more than the entire cryostat itself so i think we should go for Aluminum. The difference in conductivity is not that big so thermal gradiants in longitudinal direction should not be an issue and we can also plate Al with gold if we want for emissivity control of the surface. The stiffness of a round tube with 1/8" diameter is pretty high, however it might be not high enough for the final stability we would like to have but for a first iteration i think it's OK. But i didn't do any simulations so far. I think we first have to figure out all the thermal stuff before we think about how to get rid of the limitations by seismic noise. The support will still be adjustable and at the airy points, but we could optimize the stiffness by making the tubes thicker e.g. I think the current way to go is to figure out how and especially how good we can control the temp of the cavity and then we redesign the thing anyways.

The diameter of the cavity will be 2", so assuming a first shield with 4.5" diameter and 1/8" wall thickness there is enough space for a second shield. Current stock on campus is 3.5"x0.125", so again, i think there is enough space for supporting the inner shield on the outer shield and supporting the cavity inside the inner shield. The first problem arises from the fact that there is only limited space between the shields or the cavity for an adjustable support, even with 1/8" wall thickness. It's not a big deal but the space is a bit limited for fancy designs. So the thermal conductivity might be rather high even if we choose the right low conductivity material. But i don't wanna go for something too thin. I don't know how flexible the first design has to be for the iriteration but i think we should go for an easy solution to figure out a couple of things first. There will be a long way unitl we will be limited by seismic noise.

So i talked to our mechanical guys in downs and we don't wanna use any G10 parts in the experimental chamber if we have optics inside. PEEK would be nice but we can't afford it, the raw material would be probably somewhere around $1k or more (because of the large diameter of the tubes). So instead i would go for PTFE for the insulating spacer between the two shields and something similar for supporting the cavity. It's a pretty good insulator and if we make it wide enough it's also stiff enough. The current LIGO refcav's are also sitting on PTFE spacers.

The support for the cavity will probably look like a stainless rings with some rods used as a spacers to determine and fix the distance between the supports. The cavity is sitting on top of some post attached to the two rings which i don't know details yet (TBD). The idea is that these posts can be replaced with different sizes/material to design the right thermal conductivity and rigidity for the cavuty.
Materials have not been chosen yet for this part. The cavity will be supported by those and then fixed from the top using a set screw which can be accessed from outside the shields, So the distance of the support can be adjusted, fixed, the cavity can be loaded into it and then fixed before slided into the first shield. The first shield can then be slided into the second shield which is supported in a clever way from the cold plate. If we use the outer shield as the first heater stage to bring the cavity up close to 125K we don't have to disconnect heater and sensors whenever we change something inside the first shield.

I verified the current design with our mechanical workshop in Downs and we have all the material in stock and can get everything within 3 weeks or so.

Will talk to Warren about some details tomorrow and then post some images.

.

 

  227   Thu Jul 7 15:38:00 2011 FrankCryostatDrawingsVaccum feedthroughs - flange mofifications

attached a pdf of the drawing for the simple blank flange modifications for the feedthrough connectors. As usual all cad files are on the svn.

NW25_blanck_Fischer_Connector.pdf


 

 

  226   Thu Jul 7 02:08:50 2011 FrankCryostatVacuumvacuum schematic

here the first draft of the vacuum schematic including all important features. The valve for the insulating vacuum comes with the cryostat and is welded to it. The rest is standard KF25 equipment.

Cryostat_vacuum_schematic.png

  225   Tue Jul 5 20:11:13 2011 DmassDailyProgressGeneralEmcore Frequency Noise

I will post more details on how I measured this - for now here is the best upper bound on the Emcore's frequency noise as measured using a Michelson with a ~1m asymmetry.

 

A few notes:

  • The length noise curve was made using a symmetric Michelson layout, made as symmetric as I could by hand
    • I suspect this is all frequency noise up past 5 kHz, with suppression corresponding to how well I balanced the arm lengths of the Michelson
  • The "Intensity noise curves" (cyan, yellow) were both buried in the input noise of the 4395 - this corresponds to a RIN of ~(100nV/rtHz) / (200mV) or RIN = 5x10^7/rtHz
  • The funny shape in the noise curves are due to the PD calibration - I measured the transfer function of the Thorlabs PDA10CS and divided it out from everything
Attachment 1: EmcFreqNoise.pdf
EmcFreqNoise.pdf
  224   Sat Jul 2 15:46:10 2011 FrankThings to BuyCryoadditional items missing
  • Faceshields / safety glasses
  • Cryo gloves
  • small dewar (e.g. for temp sensor test)
  • bottle rack (for He bottle)
  • Funnel for LN2
  223   Sat Jul 2 15:16:22 2011 FrankCryostatVacuumKapton tape outgassing

Usually 3M Kapton tape, type 1205 (with acrylic adhesive) is used in space and vacuum applications. The problem is that most cryo companies don't say what they sell and they only focus on the cryogenic properties, not on outgassing for most of their stuff.
E.g. G10 is a nice material for cryogenics but is outgassing a lot at RT and contaminating the vacuum. As we want to keep the outgassing of organic material low in our experimental chamber i've checked all parts for their outgassing properties. While looking for specs of the Kapton tape used in cryogenics i found these documents:

http://www.soreq.gov.il/pdf/pirsumim/pirsimimfull/1534.pdf

http://klabs.org/richcontent/General_Application_Notes/nasa_advisories/na-gsfc-2002-03.pdf

http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20010041071_2001064675.pdf

It turns out that most companies where i could find out what they have also sell 3M 1205 for cryogenics.
Also NASA uses 3M 1205 in some of their cryogenic experiments:

Development of the Materials In Devices As Superconductors (MIDAS) Experiment (NASA/TM-1998-208440)
"Materials to be used in a vacuum must be carefully evaluated. As one example, 3M
Kapton tape 5413 looks identical to Kapton tape 1205, and is packaged similarly, but
5413 contains a silicone-based adhesive that is not compatible with vacuum work. The
1205 tape has an acrylic adhesive that can be baked to reduce outgassing.
"

I couldn't find a source what LIGO is using at RT. Does anyone know? What are you guys using at the 40m?

does anyone use the 3M™ Low Outgassing Tape 6670

  222   Thu Jun 30 19:10:08 2011 JennyDailyProgressSimulationSimple model of room temp cavity

I changed the geometry, material properties, and temperature of my model to more closely mimic the cavity and vacuum tube in the lab downstairs.

I've now saved all of my comsol models, images, and text files to the SVN in case anyone wants to look at or fiddle with them.

File title: 6-29_cavity_vactube
    •    2D axisymmetric model
    •    Materials
         ⁃    295 K fused silica cavity (Corning 7940, solid, NIST SRM 739 Type I)
         ⁃    295 K stainless steel vacuum tube (UNS S30400, solid, polished)
    •    Geometry
         ⁃    cavity diameter = 2"
         ⁃    cavity length = 8"
         ⁃    vacuum tube diameter = 8"
         ⁃    vacuum tube length = 22" (this is not counting flanges)
         ⁃    vacuum tube thickness = 0.12"
    •    Physics
         ⁃    surface-to-surface radiation
              ⁃    ambient temp: 295 K
              ⁃    no heat flux from boundaries
              ⁃    strain reference temperature: T
         ⁃    conduction within cavity and shield
              ⁃    both cavity and shell initially at 295 K
         ⁃    boundary step function
              ⁃    outside of tube set to 295.1 K at time t = 0
    •    Results: time-dependent
         ⁃    probing temp of inside of cavity
         ⁃    Time constants:
                  1) Conduction through vac tube: 5 seconds
                  2) Radiation to cavity: 2 hours

It looks like this at time t = 0: 

6-30_cavity_vactube_geom.png

Videos, demonstrating conduction through vacuum tube and radiation between tube and cavity

  • Video #1 (First 8 seconds show heat propagating through stainless steel vacuum tube.)
  • Video #2 (All 30,000 seconds of simulation, showing heat  propagating to inner cavity from outer tube.)

 

I probed the temperature evolution of the point at the very center of the inner cavity. I fit it to an exponential on Igor Pro and got the following fit:

T(t)=T_0-A*exp[-t/tau] with

  • T_0 = (295.10000 +- 0.00005) K
  • A = (-0.10268 +- 0.00008) K
  • tau= 7180 +- 15 s

 

Here's the plot (bottom curve, blue dots) and fit (red line) with residuals plotted up top (also blue dots).

6-30_cavity_vactube_plotres.png
 

My next task is to find a function that fits better, that takes into account the conduction through the stainless steel and fused silica.

 

Notes on materials:

  • Corning 7940 is the only kind of fused silica in the COMSOL model library.
  • NIST SRM 739 Type I is one of many orientations/variations for Corning 7940. I don't know which, if any of the options, is the kind we're using.
  221   Thu Jun 30 13:18:40 2011 Dmass, FrankUpdateCavityupdated mirror quotes

we talked to a couple of vendors today and relaxed the specs a little bit. Four of them will quote us within the next 24h.

Specs are now:
Silicon mirror, 1m cc, 30min wedge
L/10 @ 1.5um
SD20-10 both sides
<3nm rms
(annulus only if possible on short time scale)

  220   Wed Jun 29 12:21:50 2011 dmassDailyProgressLab WorkRFPD Measurements

Quote:

Did things today - will elog more details and plots after sleep.

  • Checked out the LB1005 extensively to make sure it behaved as I thought it did (gains and corner frequencies included) - it did.
  • Brought 2 RFPDs to the 40m to check them out
    • Took apart both PDs and inspected them
    • 61MHz PD (formerly REFL1 @ LLO)
      • Used Jenne laser to take transfer functions of the transimpedance gain in the RF path
      • Looked at LLO elog to see if I could find what mods have been made, a search for REFL1 over the whole LLO elog yielded nothing shiny
    • 25 MHz PD (formerly AS5)
      • Had a funny rattle which I couldn't truly localize - this was something on the PCB/PD assembly, I took it apart as far as I dared, couldn't find the source of the rattle, then put it back together
      • The nut on the SMA connector for the RF came loose, so I twisted the RF wire off when I was screwing on one of the SMA F-M adapters (the extenders used because the RFPD housing has its holes in the wrong place)
      • Soldered new stranded wire in place of the old broken one

Plots which will exist shortly (e.g. data exists and is saved):

  • A couple transfer functions through the LB1005 to understand its knobs
  • Measurements of the LB1005 closed loop behavior up to 1 MHz
  • Diagram of direct diode current modulation
  • Transmission, Reflection, and error signal in the "lock diode to cavity" setup using an integrator

 Make schematic drawing of self delay homodyne freq noise measurement w/ components on it

Make short Michelson for optical delay self homodyne meas

 We got a couple RFPDs as described in elog:210

Using the Jenne Laser setup at the 40m,

JenneLaserSetup.png

  • I measured the ratio of the transfer function through the RFPDs over the transfer function through the New Focus 1611.
  • The cable delay (which was small because it shows up differentially) was cailbrated out by putting the RFPD SMA cable into port 1 of the splitter after the New Focus 1611 in place of hte 50 Ohm terminator.
  • I think it is reasonable to expect the New Focus transimpedance amp to be relatively flat, and the PD itself is supposed to be GHz

The 25 MHz PD has a burn spot on it (~200 um in diameter), and its dark current is high

  • Dark current was estimated by looking at the DC coupled output (which I think is 50 Ohms) on a voltmeter
  • Vdark = 15 mV, R=50 Ohms -> Idark = 300 uA
    • Koji pointed out that this could easily be some offset in a busted op amp in the gold box

Edit - Added Rai W's measurement of REFL1 from Livingston Elog

The labels for the noise plot are switched - the 25 MHz PD is the one that should have the higher 25 MHz noise

Attachment 2: AS5TF.pdf
AS5TF.pdf
Attachment 3: REFL1TF.pdf
REFL1TF.pdf
Attachment 4: RFPDDarkNoise.pdf
RFPDDarkNoise.pdf
Attachment 5: RaiW-1101179576.pdf
RaiW-1101179576.pdf RaiW-1101179576.pdf
  219   Wed Jun 29 12:18:24 2011 dmassDailyProgressLab WorkRFPD Measurements

Quote:

Did things today - will elog more details and plots after sleep.

  • Checked out the LB1005 extensively to make sure it behaved as I thought it did (gains and corner frequencies included) - it did.
  • Brought 2 RFPDs to the 40m to check them out
    • Took apart both PDs and inspected them
    • 61MHz PD (formerly REFL1 @ LLO)
      • Used Jenne laser to take transfer functions of the transimpedance gain in the RF path
      • Looked at LLO elog to see if I could find what mods have been made, a search for REFL1 over the whole LLO elog yielded nothing shiny
    • 25 MHz PD (formerly AS5)
      • Had a funny rattle which I couldn't truly localize - this was something on the PCB/PD assembly, I took it apart as far as I dared, couldn't find the source of the rattle, then put it back together
      • The nut on the SMA connector for the RF came loose, so I twisted the RF wire off when I was screwing on one of the SMA F-M adapters (the extenders used because the RFPD housing has its holes in the wrong place)
      • Soldered new stranded wire in place of the old broken one

Plots which will exist shortly (e.g. data exists and is saved):

  • A couple transfer functions through the LB1005 to understand its knobs
  • Measurements of the LB1005 closed loop behavior up to 1 MHz
  • Diagram of direct diode current modulation
  • Transmission, Reflection, and error signal in the "lock diode to cavity" setup using an integrator

 Make schematic drawing of self delay homodyne freq noise measurement w/ components on it

Make short Michelson for optical delay self homodyne meas

 We got a couple RFPDs as described in elog:210

Using the Jenne Laser setup at the 40m:

JenneLaserSetup.png

  218   Tue Jun 28 23:46:53 2011 FrankLab InfrastructureGeneralGremlins?

I cleaned the lab yesterday as we obviously have a dirt problem in the lab as the sticky mat gets dirty within minutes if you walked in the lab and over the mat over an over again.
Attached a picture of just the big stuff i collected in the first half of the lab. I replaced the mat yesterday and took a picture today: people still don't wear booties! Is this so complicated?
And remember, no food and drinks in the labs, but that's a general thing.

Remember we all contribute to that. The more we carry in the more we have to mop!

dirt1.jpgdirt2.jpgfood1.jpg

 

taking the current data into account a tried to extrapolate how the lab will look like end of this year (95% confidence interval). Here the result:

mess1.jpgmess2.jpgmess3.jpg

I also tried to find some evidence who is responsible for the mess or even better, some eyewitness.
So i called Eyewitness News but they couldn't help me as they had no team on campus and so they had no footage. As usual you don't find people who have seen anything .
Finally i've asked a lonely stranger down in the dungeons but he couldn't help me either:

eyewitness.jpg

 

As it obviously does not work how we handle the situation at the moment i propose that we come up with some rules from now on:

  • lab cleaning EVERY week (from now on until we see some progress)
  • clean your stuff BEFORE you bring it into the lab. When you see that the stuff has a layer of dust on it, CLEAN IT (outside the lab)
  • no cardboard unpacking in the lab (this produces a lot of dust. Simply do it outside the lab)
  • no food and drinks in the labs
  • if you throw shit on the ground, pick it up and don't wait until your mother does it for you

  [DYM: I nominate this for elog of the year]

  [ZK: Seconded]

  [JD:  I would like to know who that's a "photo" of in the end-of-the-year prediction]

 

 

  217   Tue Jun 28 22:24:14 2011 JennyDailyProgressSimulationRadiation in COMSOL

I built this model today:

6-28_cavity_shield_geom.png

Info:
    •    2D axisymmetric model
    •    120 K cryo cavity and 120 K shield
          ⁃   cavity radius: 6 cm
          ⁃   cavity height: 10 cm
          ⁃   dx=1 cm (thickness of shield AND distance from outer radius of cav. to inner radius of shield AND distance from top and bottom of cav. to top and bottom of shield)
    •    All materials: silicon, bulk
          ⁃   emissivity: 0.9 for shield, 0.1 for cavity
          ⁃   everything else: from material
    •    Physics
          ⁃   surface-to-surface radiation
          ⁃    ambient temp: 120 K
          ⁃    no heat flux from boundaries via conduction
          ⁃    conduction within cavity and shield
          ⁃    both cavity and shell initially at 120 K
          ⁃    step function
          ⁃    outside boundary of shield set to 120.1 K at time t=0
    •    Results: time-dependent
          ⁃    probing temp of inside of cavity
          ⁃    time constant tau= 3.72*10^5 s = 103 h

The time constant DMass calculated was 640 h for similar dimensions and material properties... The reason for this discrepancy is still unclear.

Inside center of cavity:

6-28_cavity_shield_probe1.png

Next I'm going to change the dimensions, materials, and temperature conditions to more closely reflect the room temperature cavity on which I can take real measurements. Then I'll work on fitting the simulation point-probe output in Matlab and try to identify the both the radiative pole and the pole for conduction through the materials.

Also I'm going to go through the radiative heat transfer calculations and see if the COMSOL output for the time constant matches what I'd expect.

  216   Tue Jun 28 20:19:34 2011 DmassLaserLab WorkLB1005 Testing

Rana suggested that I test the New Focus LB1005 to make sure it behaves as I think it should.

To test this, I used the following setup:

LB1005Test.png

I used both a short and a 1M Hz (R = 3.5k C = ?) as Ztest for closing the loop.

Integrator Closed Loop Behavior:

  1. I set the loop shape to "integrator"
  2. I closed the loop through Ztest
  3. I put sine waves of various frequencies into the error point
  4. I look at the sine wave as a peak on the HP4395 through the error monitor (Zout = 50 Ohms)
  5. I turn the gain knob all the way down to see the free running level of the sine wave
    • I checked that this was the same as turning off the loop
Gain Freq (Hz) Free Running (dBV/rtHz) In Loop (dBV/rtHz) Suppression (dB)
 5.6  300  -35.7 -59.2  23.5
   30  -36.7  -80.9  45.2
 10  78k  -59.8+20  -72.5+20  12.7

We get the extra ~20 dB suppression we expect at 30 Hz. At a gain of 5.8 (1.74 = 20 dB) we would expect a UGF of ~3kHz, so far this jives.

 P-I Closed Loop Behavior:

  1. I put in the 1MHz Hz pole (as Ztest)
  2. I set the PI corner to 1MHz
  3. I repeated steps for the Integrator Closed Loop Behavior
Gain Freq (Hz) Free Running (dBV/rtHz) In Loop (dBV/rtHz) Suppression (dB)
4.1 10k -55 -79 24
  1k -55 -99 44
         
5.8 1k -52 -120  68
  10k -52 -99.9  48
  100k -52 -79.9  28
  1M -52 -54.5  2.5

 

2M

-52

-52.2

 0

 The box seems to work as expected up to 100 kHz

  215   Tue Jun 28 20:07:52 2011 FrankUpdateCavityMirrors

We might have a problem with the current vendor for our Silicon mirror substrates.
According to the Caltech export control guys this company is currently blacklisted, so we are not able to buy from them.
However, i've called that company today and they say that they are not blacklisted and can do unlimited business with anyone and our people might have an old list and are not up-to-date.
The have been on that list a few years back when the were accused of illegally exporting defense technology. However this case is settled and they paid a small fine (link).

I've forwarded some contact information to Rod who will forward it to our people on campus so that they can get in touch with their attorney to clarify this. That's what they offered me today.
He was sure that they will figure that out on a very short timescale. I will wait another day or two as we don't have a real alternative at the moment (with short lead time) and would have to start the whole bidding process again.
In parallel i will make a list of companies i didn't ask already and then we start with those. We will also talk to all the others to see if we can get something with less quality earlier.

  214   Mon Jun 27 15:28:43 2011 JennyDailyProgressSimulationMaking COMSOL solution match analytical solution part II

 

I ran this new simulation of silicon and foam. Here's a picture of the model (again with the foam in red and the silicon in blue):

disk_si_foam.png

I changed the dimensions to two cylinders, each of length 1 mm and each of radius 10 cm for two main reasons:

1) A short, wide silicon cylinder has a lower thermal resistance than a long one. The approximation in which I neglect its resistance becomes more accurate.

2) I didn't want to make the foam cylinder so long that its thermal resistance was so large that the time constant of the system was on the order of days. These dimensions gave a time constant (in my pen-and-paper calculations) of 43.6 seconds.

 

My first run popped out a time constant of 50 seconds. That's a 13% difference from my calculations.

Then I noticed that the solver was only updating the solution every ~10 seconds. I told it to update every second instead and got a time constant of 46 seconds, a 5% difference.

 

In summary:

Percent difference between calculated time constant and COMSOL time constant

  • Si w/o foam, computer-generated timesteps: 40%
  • Si w/ foam, computer-generated timesteps: 13%
  • Si w/ foam, max timestep of 1 second: 5%

 

Here's a plot of temperature at each end of the silicon cylinder. disk_si_foam_probe.png

It just looks like one line because there is essentially no temperature gradient over the cylinder. Both ends are at the same temperature to the 4th decimal place.

  213   Mon Jun 27 10:26:50 2011 JennyDailyProgressGeneralMaking COMSOL solution match analytical solution

I’ve made a basic COMSOL model.

I only modeled conduction through the silicon cylinder.

 

Here’s my cylinder: cylinder1.png

 

The length is 50 cm and the radius is 10 cm. The entire cylinder was initially at 120K. I set one end boundary to 130K at time t=0 and thermally insulated the other boundaries. I ran a time-dependent simulation and recorded the temperature at the other end of the cylinder (in the center of the circular boundary) as a function of time.

 

I generated this plot:cylinderprobe1.png

 

The blue trace is temperature at the center of the circular boundary at which I set the temperature. Obviously it stays at 301 K throughout the simulation. The green trace is temperature at the other end. This simulation produced a time-constant of 1500 seconds.

 

ANALYTICAL SOLUTION:

As a first approximation I treated the cylinder as a 1-D heat problem assuming the cylinder could be modeled as a resistor and a capacitor in series. The solution to this problem T(x, t) assuming the initial/boundary conditions T(0<x≤L, 0)=300K and T(0, t)=301K is (1-exp[t/(RthCth)])+300 with Rth=L/(A*k) and Cth=rho*V*Cp

 

The time constant for the system is tau=RC=L^2*rho*Cp/k. In my problem L=0.5 m, rho=2330 kg/m^3, Cp=703 J/(kg*K) and k=163 W/(m*K). This gives a time constant of 2512 s.

 

Looking at the simulation, the time constant is 1500 s. At first I wasn’t sure why the values didn’t match up. Doubling the length of the cylinder did quadruple the time constant in my simulation as expected. Changing mesh size, and radius of the cylinder did not change the time constant (as expected).

 

I realized that the problem was that since the silicon is the only material in the system, it produces the dominating thermal resistance and thermal capacitance, which cannot be modeled as a lumped resistance and lumped capacitance one after the other. The cylinder does not heat up uniformly, as shown in this plot of temperature gradients across the cylinder at time t=1500 s.

cylindergradient1.png

Instead the system is more like an infinite number of little resistors and capacitors R—C—R—C—R—C—…. This problem is harder, so instead I think it makes more sense to change my initial model so that the simple approximation works.

 

To ensure that my simulated solution and my analytical solution match up, I am choosing to change the COMSOL model so that the approximation in which I treat it as a single resistance and a single capacitance is valid.  I’ll add a piece of insulating material at one end of the cylinder. Since foam has a low thermal conductivity, and a low specific heat, I can treat it as a resistance with no capacitance and the silicon as a capacitance with no resistance. Adding the foam should make the time constant of the system much larger, and it should make the analytical approximation more closely match the COMSOL output.

 

Here’s the new model: cylinderfoam1.png

 

The red is foam and the blue is silicon. I may change the dimensions from what's shown but this is the general picture. I will update soon with my findings.

 

  212   Fri Jun 24 19:18:23 2011 FrankThings to BuyCryoCryogenic G10

list if sources for cryo G10

http://www.jjorly.com/g10_nema_grade_sheet_material_price.htm

http://www.cryogenicg10.com/

http://www.spauldingcom.com/cryogenic.html

http://www.smallparts.com/phenolic-g-10-glass-grade-sheet/dp/B003SLF01C/ref=sr_1_1?sr=1-1&qid=1308646606

  211   Fri Jun 24 13:10:32 2011 FrankMiscGeneralDevelopment of a Sub-Hz laser system for QUEST@PTB

1685f6f2ee.jpg

c0dada6559.jpg

 

 http://www.quantummetrology.de/quest/index.php?id=222

quest_ptbposter.pdf

  210   Fri Jun 24 03:04:12 2011 dmassDailyProgressLab WorkWork

Did things today - will elog more details and plots after sleep.

  • Checked out the LB1005 extensively to make sure it behaved as I thought it did (gains and corner frequencies included) - it did.
  • Brought 2 RFPDs to the 40m to check them out
    • Took apart both PDs and inspected them
    • 61MHz PD (formerly REFL1 @ LLO)
      • Used Jenne laser to take transfer functions of the transimpedance gain in the RF path
      • Looked at LLO elog to see if I could find what mods have been made, a search for REFL1 over the whole LLO elog yielded nothing shiny
    • 25 MHz PD (formerly AS5)
      • Had a funny rattle which I couldn't truly localize - this was something on the PCB/PD assembly, I took it apart as far as I dared, couldn't find the source of the rattle, then put it back together
      • The nut on the SMA connector for the RF came loose, so I twisted the RF wire off when I was screwing on one of the SMA F-M adapters (the extenders used because the RFPD housing has its holes in the wrong place)
      • Soldered new stranded wire in place of the old broken one

Plots which will exist shortly (e.g. data exists and is saved):

  • A couple transfer functions through the LB1005 to understand its knobs
  • Measurements of the LB1005 closed loop behavior up to 1 MHz
  • Diagram of direct diode current modulation
  • Transmission, Reflection, and error signal in the "lock diode to cavity" setup using an integrator

 Make schematic drawing of self delay homodyne freq noise measurement w/ components on it

Make short Michelson for optical delay self homodyne meas

  209   Fri Jun 24 01:16:35 2011 FrankHowToSimulationMultilayer Control of an Optical Reference Cavity

found this one while looking for publications on thermal modeling of refcavs:

http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=5447814

http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=4416756

done in comsol see here:

Simulation of a Multilayer Thermal Regulator for an Optical Reference      http://www.comsol.com/papers/7236/

Ospina.pdf

Ospina_pres.pdf

  208   Thu Jun 23 19:33:27 2011 Warren JohnsonNotesmaterial propertiesspecific heats of 3 materials

For thermal design, we need to know specific heats of materials.   I looked up the basics in the cryogenics book by Guy White, 3rd edition.  You need to do some simple calculation to get the needed numbers for the entire temperature range.   I did this for the three dominate materials in the cryostat: copper, aluminum, and silicon, and plot the results below.   The programs that do it are attached.   

specific_heats.png

 

Attachment 2: specfic_heats.m
% specific_heats.m
%    to get the specific heat of a solid, we can use Debye's model 
% for the specific heat due to lattice vibrations.
%   Only three ingredients, 1) the universal debye function, 2) the debye 
% temperature for the solid, and 3) the atomic mass (to convert from moles
% to grams.   Results are plotted for three elements.  The results checked 
% against wikipedia values for copper and aluminum.  

clear

... 44 more lines ...
Attachment 3: debye_func.m
function c_v = debye_func(T_D)
% given normalized temperature T_D = abs temp T divided by debye temp
%     (valid  for .033 < T_D < 10 ) 
% return the specific heat c_v, in units of J/mol K

% load table of debye function, 
% copied from Experimental Techniques in Low Temp Physics, 3rd ed.,
%   by Guy K. White, Table D, page 314.
DT = [
    .0333 .072
... 29 more lines ...
Attachment 4: test_debye_func.m
% test_debye_func
%    check that interpolation routine is working.
clear

% load in table of debye func 
DT = [
    .0333 .072
    .04 .124
    .05 .243
    .0625   .474
... 37 more lines ...
  207   Wed Jun 22 22:57:13 2011 ranaElectronicsGeneralSuperconducting THz Bolometer Readout

Really cool thesis on low noise amplifiers:

Thesis

  206   Wed Jun 22 10:22:54 2011 FrankUpdateVacuumvaccum pump station shipping date changed

i got confirmation that the pump including all options and adapter will be shipped August 5. They don't have any in stock at the moment.
I hope it won't delay anything as the cryostat will be delivered first week of August as well.

Will update the project file on the svn.

  205   Wed Jun 22 08:07:30 2011 FrankUpdateCoatingscoatings from REO

Unfortunately, REO is not making low value or small quantity coating runs anymore. After their internal meeting this morning the sales guy contacted me to let us know that they won't do small coating runs anymore.
This is what i got from them:

We do have the capability but our business model has shifted to focus on recurring optical opportunities or large high value programs, as we focus our limited engineering resources.

  204   Wed Jun 22 07:49:56 2011 FrankUpdateSensorsCryo temp sensors - slight delay

the shipping of the cryo PT sensors from Heraeus will be delayed a little bit.
They still didn't ship the sensors as the still didn't receive a correct PO.
Our Techmart database is out of date and contains a wrong company name.
So i'm waiting for the change request paperwork being  done before we can resend the order.
They also don't collect sales tax and need proof/conformation that we do not resale the parts or pay sales tax somehow.
I hope we can figure that out by end of the week. Parts are still in stock.

  203   Mon Jun 20 19:29:44 2011 FrankUpdateVacuumall parts ordered

Warren checked the vacuum system this morning, so i placed all orders for the remaining parts.
The vacuum pump is the longest lead item with 3-5 weeks (still before the cryostat arrives). Everything else is in stock and should be here end of the week.

  202   Sat Jun 18 13:17:14 2011 FrankThings to BuySensorshigh temperature Germanium sensors

this company makes them: http://www.adsem.com/

The price for regular thermistors SE-1M is ~$40 each. The price for the resistance value in the range of 0.5-1MOhm at 77K is about $10 more, lead time 3 weeks.
With calibration they are about a factor of two more expensive, but i think we don't need absolute calibrated ones as we don't know the exact temperature we need and we have out PT sensors which follow a standard curve (
DIN43760 and IEC751). Alpha is about -2.3%/K for the NTCs which is much higher than any other sensor at 125K.
Minimum order is $100.

Below the datasheet for the SE-1M. They don't have any datasheets on their website.

SE-1M_NTC_thermistors_data_sheet_196_6-2011.pdf

  201   Fri Jun 17 23:21:58 2011 FrankCryostatDrawingsvaccum system

here a picture of what i think the vacuum system could look like. There is a flex hose missing from the valve on top of the 6-way cross to the tee attached to the other valve (the valve for the outer insulating vacuum).
There are two gauges, one for the insulating vacuum (down to 10-5 Torr), mounted on a tee, which we don't need (the tee) but i wanted to see if there is enough space to add one just in case we want to add another feedthrough over there.
The other is a better model which can measure down to 10-8 Torr. There are also two 24pin connectors on the cross and an open port for the needle valve for the exchange gas. The two open ports are connected to the LN2 reservoir.

Full_system.jpg


Will add the schematic for it later.

 

ELOG V3.1.3-