I was very dubious of the previous measurement - there was something about the way the stabilization just suddenly engaged when I passed a threshold level in the gain that was very suspect. None of the DC channels were changing so I decided to have a look at the size of the 10MHz AC output from the OUT of LOOP PD on a spectrum analyzer. When the loop is open this peak is around -24dBm. When I close the loop with high gain it drops to -70dBm, then when I lower the gain by 10x this signal increases back to -24dBm. There's definitely something screwy going on.
I'm in the process of completely characterizing the fiber stabilization setup before we pull it all apart. I discovered the following things:
I've closed the loop again and it looks to be more stable than it was, at least by watching a time series of the out of loop sensor, but I think we could definitely improve it in a rebuilt version. I'll post some spectra tomorrow.
The attached plot shows the 570kHz sidebands on the in-loop signal that appear when I drive the 70MHz AOM close to the limit of the amplifier it is driving.
Here is the transfer function of the passive whitening filter I am using.
Z = 0.35Hz, P = 84.7Hz
Here is a schematic of the fiber stabilization layout before the dismantling due today. I have measured all the relevant powers and voltages and will append them shortly:
There was an additional 200x gain (~46dB) between V_B and V_A, where the measured transfer function below is V_B/V_A. Hence the loop transfer function is 200x the measured transfer function. See the experiment layout for an illustration of this.
Calculation of Loop transfer function:
Therefore, the loop gain, phi_out/delta_phi, = 2*K*F(s)*a_mixer*(-0.93dBm)/(i*f) ~= 2*(100Hz/Vrms)*(200)*(0.113Vrms)/(i*f) = 4520 Hz/f
This loop gain is plotted in the attached figure. It is within 10% of the measured value.
The following values were measured on the fiber setup - see experiment setup for diagram of the experiment.
Ref beam DC = 0.601 mW +/- 0.1uW (DAQ - PD calibration: assumed 770 V/W)
Ref beam DC = 0.504 mW (ThorLabs power meter)
Trans Beam DC = 1.67 uW +/- 0.02 uW (DAQ – PD calibration: assumed 770 V/W)
Trans Beam DC = 0.001 mW +/- 0.001mW (ThorLabs power meter)
RF Level = +6.36 dBm
Ref beam DC = 0.614 mW +/- 1uW (DAQ : PD calibration: assumed 770 V/W)
Ref beam DC = 0.468 mW (ThorLabs Power Meter)
(Re-calibrated PDs slightly at this point - multiplied DAQ input by 468/614)
Double-Trans Beam DC = 0.66uW +/- 0.03 uW (DAQ PD calibration: assumed 587 V/W)
Double-Trans Beam DC = 0.34uW +/- 0.04 uW (DAQ PD calibration: assumed 587 V/W) – with 70MHz AOM off, therefore this is scatter
RF Level = -0.93 dBm
The phase and frequency noise measurements for the fiber stabilization experiment. Includes open & closed loop, as well as relative sizes of contributions from intensity noise, shot noise and electronics noise. I think we need to build a better reference arm to stabilize the fiber against.
Boffo! Nice work lads.
- apparently PDFs work okay. Frank is reporting continual crashes from last night when uploading a graph of the particle counter in different formats from Firefox in Windows.
The guys are back this morning to reseal the vents. There are some green marks around the place but also what looks like new red ones over the top of the silicon they laid down two weeks ago.
I shutdown the 35W laser using the 'System Off' option on the touch-screen and then turned off the NPRO and removed the key. That key is now in the second drawer down in the first desk. The key for the Gyro NPRO was not in the laser.
I added some EPICS channels to the Hartmann sensor softIoc and then added these to be recorded in the frames.
I then killed the daqd process on fb1 so it would start afresh.
The 35W laser and the Gryo NPRO have been shutdown in preparation for tomorrow's conduit installation work. The key for the 35W laser in the lower draw in the first desk attached to a Thorlabs USB flash-driver. The key for the Gyro laser wasn't in the driver. The laser was in use this afternoon and since then someone has completely powered it down and removed the key, I'm assuming in prep for tomorrow morning.
I've taken the following items to the 40m.
They're all marked "Adhikari Lab".
We now have two temperature controllers in the lab:
A couple of weeks ago they installed a second temperature controller on the South Wall. This drives the HVAC heater that is above the stereo.
The original temperature controller (West Wall) was also upgraded to have a fancy new 1960's-style mechanical display of the setpoint-needle on the front type. Maybe in 50 or 60 years we can get a digital controller in here.
At the moment the second controller has not been calibrated to match the original controller.
the network connection is down again, so our router has to be restarted ...
whoever will go to the lab on Monday first plz powercycle the linksys router...
I fixed two machines in the TCS lab to have static IP addresses on the local network.
The Athena DAQ CentOS box: 'tcs_daq' 10.0.1.34
The CentOS workstation: 'tcs_ws' 10.0.1.25
Frank, please add these to the network topology diagram you have ...
I killed and restarted the daqd process because I wanted to add some 16Hz TCS channels to the frame builder. These are from the Athena DAQ box.
I edited the following files:
Looks like Hurricane Dmass.
I was trying to understand why the mode matching through the PMC was so bad (I got 30 mW transmitted with 120 mW input).
So. Who knows where the wincamD is?
added: pics of PMC REFL before and after locking, in order
Totally possible that my mode matching is just that crappy? Maybe.
I did take beam scans of the mode going into the PMC just before aligning to it for the final time. I can do so again to get real profiles and characterize the H1NPRO
Here are some scans of the beam with the winCAMD. I had a reflective ND filter on for these (hence some of the fringing).
1) User directories have been moved into /users. There is now the awesome /users/abrooks/aidan/Aidan/ directory.
/users links to /cvs/users
(syntax to do this was "ln -s /cvs/users /users")
ln -s /cvs/users /users
I added some Hartmann Sensor channels to the frames and restarted DAQD on fb1 to include them.
See here for details ...
Visitors to the lab ...
The ants are back again and it's only going to get worse over summer.
We have ant poison in the TCS lab. Feel free to use it.
I've moved all the old medm directories into an archive directory. (/caltech/medm/archive)
I've also gone into the /cvs/cds/advLigo/fe/ and moved all the old models into an archive directory (/cvs/cds/advLigo/fe/archive).
I saved the existing ATF model as atf.mdl.ctrl and built a modified version for some temporary TCS real-time work for aLIGO. The new ATF model is saved as atf.mdl.tcs (and currently as atf.mdl).
The new model compiled and was installed. It still runs all the GYRO stuff (which was all unaltered) but I replaced the defunct CTRL block from the doubling experiment with a new TCS block.
- should the old model need to be replaced, this can be done by copying aft.mdl.ctrl to atf.mdl and compiling with fb0:/cvs/cds/advLigo$ make atf install-atf
The C2ATF model with aLIGO TCS controls is now running correctly on fb0.
I followed the standard troubleshooting instructions in ATF eLog 124 to get the model to run. All the channels can be access from the C2ATF_TCS_MASTER.adl medm screen in /cvs/cds/caltech/medm/c2/atf/
We're now saving 14 channels of data to frame builder. They are:
[C2:ATF-TCS_AOM_OUT_OUT_DAQ] - 4096Hz
[C2:ATF-TCS_AOM_SET_OUT_DAQ] - 16Hz
[C2:ATF-TCS_CHILLER_SLOW_GAIN_OUT_DAQ] - 16Hz
[C2:ATF-TCS_CHILLER_TEMP_SET_OUT_DAQ] - 16Hz
[C2:ATF-TCS_ISS_IN_AC_OUT_DAQ] - 4096Hz (in-loop ISS PD)
[C2:ATF-TCS_ISS_IN_DC_OUT_DAQ] - 16Hz (in-loop ISS PD)
[C2:ATF-TCS_ISS_OUT_AC_OUT_DAQ] - 4096Hz (in-loop ISS PD)
[C2:ATF-TCS_ISS_OUT_DC_OUT_DAQ] - 16Hz (in-loop ISS PD)
[C2:ATF-TCS_LSR_HD_PD_OUT_DAQ] - 16Hz
[C2:ATF-TCS_LSR_SLOW_GAIN_OUT_DAQ] - 16Hz
[C2:ATF-TCS_LSR_TCS_LSR_TEMP_OUT_DAQ] - 16Hz
[C2:ATF-TCS_LSR_PWR_MTR_OUT_DAQ] - 16Hz
[C2:ATF-TCS_PZT_CTRL_SENS_OUT_DAQ] - 16Hz
[C2:ATF-TCS_PZT_OUT_OUT_DAQ] - 4096Hz
Another (incredibly) useful suggestion:
Always add the filename used to generate plots to your plot ... so when you post/print it, you can easily determine the source file at later date.
text(0, -0.12, [mfilename('fullpath'), '.m'], 'FontSize', 6, 'Interpreter', 'none', 'Units', 'normalized')
Depending on the orientation and margins of the plot, you may have to play around with the Y coordinate of the text command (= -0.12 in this example) to get the text to fit on the image without interfering with the X-axis plot label.
RXA: Use IDfig.m instead.
function tt = IDfig(varargin)
% IDfig puts the name of the calling function and the date on the right side of a figure
% there must be a current axis, and it must be called from an m-file
% replaces IDplot, which has a name conflict with something in the sys-ID
% toolbox. BTL July 17, 2007.
% if called with an input string, that string will be appended to the message,
% eg IDfig('data from tf_120104_1.mat')
% if called with an output argument, it returns the handle to the text object it created
FB0 was unresponsive from the network. I am trying a hard reboot of it.
I've attached a template PID servo Perl script that we used during the eLIGO days for temperature control of the TCS chillers. We can adapt this to offload the PZT voltage offset to the temperature of the NPRO lasers.
# Chiller PID Servo
# Adapted by Danny Atkinson 2009-05-20 from:
# PID Servo for PSL-FSS (Slow)
# Tobin Fricke 2007-01-09
# Current SVN version
# $Id: PIDChillerServo.pl 937 2010-05-04 16:42:01Z daniel.atkinson@LIGO.ORG $
One in the TCS Lab. One by the ATF door. I have more ant bait available in the TCS Lab.
I set up most of the two micron fiber optics as described in T1600146. I discovered that (a) our 90:10 splitters were accidentally ordered with
FC/PC connectors, not FC/APC connectors [but Iíve already spoken to Thorlabs to arrange a swap] and (b) we definitely need fiber trays to
handle spools of extra fiber.
Other than that, the basic layout looks good.
Before the holiday break, I set up the first stage of the TMTF. The basic concept is to use a fiber Mach-Zender with unequal path lengths as a ruler for the frequency noise of the laser diode.
The existing set up is as follows in the block diagram. The actual set up is shown in the attached photo.
The output from the laser diode is sent into an isolator. This is then passed to a 90/10 splitter. The 90% port is dumped and the 10% port is sent into COUPLER A (nominally a 50/50 BS for the input into a MZ). One path of is connected directly into COUPLER B. The other is first connected into a 2m long patch cable and then into COUPLER B. One output from the coupler is sent into a Thorlabs S148C InGaAs photodiode based power meter.
None of the fibers are polarization maintaining.
After connecting everything and turning on the diode, the power meter showed an output power than ranged from 11.25uW to 14.8uW. I could clearly see the output fluctuate through a sine wave as made one up longer by gently heating the patch cable. Additionally, changing the diode temperature induces a wavelength change. By adjusting the temperature set point, I could make the output go through fringes.
Given a 2m long path length difference (DeltaL) with a refractive index of 1.43, the accumulated phase difference is:
Phi = 2 Pi DeltaL/Lambda
The change in phase difference with changing wavelength is:
Dphi = -2 Pi DeltaL/Lambda^2 * DeltaLambda
= -2 Pi DeltaL/Lambda^2 * (DeltaLambda/DeltaT) * DeltaT
= 450 radians/K
For the 10K thermistor, one degree corresponds to 462 Ohms, so we expect around 6.5 Ohms to correspond to 2 Pi radians (or 3.25Ohms corresponding to a peak-peak value of 3.55uW or 0.31V from the power meter analog output). Very gently changing the thermistor set point showed the output of the MZ going through fringes at roughly this rate.
The output of the MZ slowly drifts between maximum and minimum over the course of twenty or thirty seconds. I decided to try to lock the output of the MZ at low frequencies using the analog output of the power meter, a crude analog integrator and the input to the temperature controller as an actuator. I also used a DC voltage source to lock to a point half way up the fringe.
The integrator used 1.68MOhm resistance with 10uF capacitor to yield a transfer function of 9.5mHz/f. The total loop gain, including the 2kOhm/V response of the temperature controller input and the 0.31V/3.25Ohm response of the power meter readout of the MZ to a thermistor set point change was estimated as 1.8Hz/f.
This was enough to lock the MZ output to a steady-state point about half-way up a fringe. Without a proper photodiode, there was no real spectrum to be measured.
The next step is to set up some proper photodiodes to do the readout out and locking.
We want to set up a new Cymacs in the ATF. It would be much cheaper if we don't have to order new ADC and DAC cards.
I pulled the old IO chassis (which contained the old ATF ADC and DAC cards). Unfortunately. they are PCI-X (we think) which will not fit in the PCI-E slots in the new Cymacs.
We figured we might be able to leave the DAQ cards in the old chassis and continue to it if we can hook it up to a new Cymacs. So we pulled FB0 and extracted the board that expands the BUS out to the IO Chassis (see below). At the moment, we're investigating if this board will work in the new Cymacs. It is, at least, PCI-E.
I have set up a dedicated Windows 7 machine in the TCS lab to be the OPC server for the Newport Temperature sensors.
The OPC server is now running and broadcasting the EPICS channels for the temperature, humidity and pressure for these sensors. Currently, the channel names have the prefixes listed below, but we'll likely change them to actual lab names.
C4:PEM = TCS Lab
C3:PEM = CTN Lab
C6:PEM = CRIME Lab
I forgot to note this last Friday. The temperature data from the labs is now being written to file in a somewhat brute force fashion:
camonitor [channel_name] > file.dat
This is still running, so at the very least we should be getting a coarse record of the lab temperatures. Eventually we want to set up a DAQD and framebuilder for this. We should be able to convert the current data into GWF frame files when that happens.
I measured the intensity noise of the Eblana EP2004-D laser diode. I confirmed that the intensity noise is above the dark noise of the photodiode but I have not, as yet, made any attempt to optimize the photodiode response (or analytically determine its expected noise floor).
Power measurement (power meter)
Laser > isolator > 90/10 split > 90 port > Thorlabs power meter (596 microW)
Power measurement (photodiode)
Laser > isolator > 90/10 split > 90 port > DET10D > 50 Ohm terminator > SR560 [100x gain] > 3.159 V [DC]
Therefore, the voltage across the 50Ohm terminator on the photodetector is 0.0316V. The current is then 632 microAmps. The responsivity of the DET10D is 1.2A/W at 2000nm. Therefore, this is approximately 530 microWatts (about a 13% difference with the power meter measurement).
Intensity/dark noise measurement:
Laser [OFF] > isolator > 90/10 split > 90 port > DET10D > 50 Ohm terminator > SR560 [1E4 gain, high pass filter at 0.3Hz] > SR785 [DN measurement]
Laser [ON] > isolator > 90/10 split > 90 port > DET10D > 50 Ohm terminator > SR560 [1E4 gain, high pass filter at 0.3Hz] > SR785 [INT NOISE measurement]
The results from these two measurements are plotted below. The SR560 filtering was undone to refer the measurements back to the DET10D/50Ohm output and both results were divided the DC voltage at that point (31.6V).
1 Hz: 1E-5 per rtHz
10 Hz: 1.5E-6 per rtHz
Input File: ../spectra/srs003.78d
Measure Group: FFT
Measurement: FFT 1
Num of extracted Points: 801
Start Freq: -0 Hz
Span: 100 Hz
FFT Lines: 400
Averaging Mode: RMS
Averaging Type: Linear / Fix Len
Input File: ../spectra/srs001.78d
Measure Group: FFT
Measurement: FFT 1
Num of extracted Points: 801
Start Freq: -0 Hz
Span: 100 Hz
FFT Lines: 400
Averaging Mode: RMS
Averaging Type: Linear / Fix Len
Johannes and I took the phospor coated CCD camera (from the cryo lab) down to the ATF to see the response to 2004nm. We tested it was working by shining an incandescent light onto it and confirming that we could see a signal on the monitor. Then we took the output of the 2004nm laser diode and put about 600 microWatts directly onto the CCD camera and ...
We saw no response at all. Nothing - on any of the gain settings.
I set up a measurement of the DET10D QE at 2004nm. I supplied the Eblana 2004nm fiber-coupled laser diode with 45mA of current. I first measured the output with the Thorlabs power meter and then measured output with the DET10D photodiode. Both systems are fiber-coupled (although both fiber couplers are screwed on rather than glued on).
The power was 0.34mW = 3.425E15 photons per second
The photodetector output was run through a 50 Ohm resistor which was then run through a SR560 with 100x gain at DC. The measured voltage from the SR560 was 2.183V. Hence the photocurrent was 4.36E-4A = 2.7E15 electrons per second
Therefore, the QE was 0.79 at 2004nm. This agrees with the manufacturers curve. However, I notice that there:
I've done a more thorough examination of the Eblana 2004nm intensity noise spectrum. Unfortunately, there is something weird going on with the detector which I haven't resolved yet. The set up was:
500 micro-Watts > DET10D > 1497 Ohm transimpedance resistor (RT) > SR560 (AC measurement settings: HP filter, 6dB @ 0.1Hz, Gain = 1000x)
DC measurement: SR560 Output = 0.962V (gain = 1) > current = 6.4E-4A > 535E-6 Watts electrical > 677E-6 Watts optical (need to check this!)
However, when the intensity noise and the dark noise are measured, I find the dark noise (laser off) is way above the NEP, and well above the JN, the shot noise, the SR560 input noise (measured and from manual) and the expected NEP.
The DET-10D is an extended InGaAs photodiode which is biased either using a battery or an external supply. On the schematic you can see that the diode is directly connected to the BNC cable.
According to the datasheet, the NEP is 1 pW/rHz. The shot noise in 600 uA is sqrt(2 * 1.6e-19 * 600e-6) = 14 pA/rHz, so we should be fine if the datasheet is accurate for f < 1 kHz (which I doubt).
14 pA / 600 uA => RIN = 2.3e-8
Where is this dark noise coming from???
The input referred noise of the SR560 is ~5 nV/rHz above 5 Hz. With the 50 Ohm resistor, that gives us an equivalent current noise of 100 pA/rHz (much worse than anything else in this circuit). It is also ~5x larger than the thermal noise of the 50 Ohm resistor.
You should use a wire-wound resistor instead of a sketchy terminator, and it should be ~1-3 kOhms. To keep from saturating the SR560, you'll have to AC couple it.
If the diode shut resistance was really low (like 10 Ohms), then this might explain what we're seeing. Unlikely, but the detector is behaving like it has a very low resistor in parallel to the transimpedance resistor.
I've installed Debian 8 on the new ATF Cymacs and called the machine FB4.
Following the instructions on the ATF Wiki, I've installed the ADVLIGORTS-CYMAC package. However, I've not built the framebuilder, RTS or DAQD.
I ran the 1550nm laser through the silicon piece at 11mW incident power. The laser current was set to 100mA and the output power was measured on the Thorlabs 2W power head. The camera was approximately 250mm from the silicon and I did my best to image the front surface of the silicon.
The camera exposure time was set to 200ms. The didn't seem to be any other gain settings available.
I took 100 images with the laser on and then 100 images with the laser off. I average these and took the difference - which is shown below. You can see a faint horizontal light-saber line around Y-coordinate = 170 pixels.
The data from the averaged images is in the attached MAT file.
Some more plots of the data, including a mean of the columns in the image of the silicon. The silicon piece is about 60mm in width.
Yesterday the leak was worse. And there was a smell of foul effluent in the lab. And some brown water by the bottom of the rack. This was later discovered to be drain water rather than sewage.
The custodians cleaned up the clear water. I filed a Service Request and the plumbers came out and cleared up the brown water with disinfectant. They subsequently spotted a leak in the pipe itself - about 8 feet up the pipe. This is scheduled to be fixed. Eric G informs me that there is duct tape on leak right now. I spoke to the service center and they're going to do a permanent repair (replacing the pipe) sometime in the near future. If the pipe leaks again in the interim, they said to contact them and they'll expedite the replacement.
[Aidan, Gabriele, Eric]
We have wrapped the two optical tables in plastic. The flammables cabinet has been shifted away from the pipe on the West wall. The computer rack has been disconnected from several cables so that it is clear of the south wall drain pipe.