We do, as of now, have one entirely working beam profiler. I believe we have borrowed it from Peter's lab.

Today we used it to profile the beam after realigning our setup (two or three times - so we actually have two or three profiles, but one of them is a razor blade measurement). We have much more bench space for the cavity now, and we can actually figure out what the beam is supposed to look like at different points, since we can get an accurate measurement of the waist size.

The beam is ~20% elliptical (estimate, we have yet to fit the data). It appears that the BeamScan software automatically compensates for ellipticity, so that on our readout we have one of the axes along the semimajor axis and the other along the semiminor axis. If we find out that we are wrong about this we will rotate the axes on the beam profiler head to get the correct orientation.

Pictures of new setup and beam profiles to follow - (tomorrow).

Friday 7th Aug 2009
2009-08-08 00:18:45 UTC - switched to DC output of PD2 in C2:ATF-QPD_SEG4_IN1_DAQ
Gain is 2000x on SR560. Loop is closed with 10,000x gain. So capture a longer spectra of intensity noise and convert to phase noise taking into account gain, transimpedance difference between AC and DC channels and phase difference at mixer(s). (933 725 940)

6:56:10 UTC PM - saved spectra with uncalibrated DC noise in it. (although it's divided by 10050)
    - also running realtime long term 0.003Hz bandwidth measurement over the weekend

Monday 10th Aug 2009
8:28:00 AM - Laser OFF to get zero level for DC noise measurement. The DC-OUT from PD2 (out-loop) has been plugged into C2:ATF-QPD_SEG4_IN1_DAQ for the weekend.

Minute average time series of PD2 DC-out for last two days is shown in the attached plot from dataviewer.

So in an attempt to impedance match the EOM, I took a ZFDC-20-1H Coupler (rated @ 30-400 MHz), and connected:

• The Wenzel Crystal (13 dBm @ 35.5 MHz) to a splitter, then one of these two signals to the In port, and the other to two open leads which I shorted with various resistors.
• A cable leading to the EOM input at the PSL enclosure to the Out port
• Both Coupling In and Coupling Out to an Oscilloscope.

I then looked at the RF signal on the scope for each of these, and tried to minimize the ratio of output coupling (a measure of my RF reflection) to input coupling (a measure of my input power) by eye. I couldn't seem to change the relative level of the RF signals by shorting various resistors around 50 Ohms in resistance to the T on the In signal, only attenuate them both. It seems to me this implies putting a 50 Ohm T at my input does nothing? Please let me know where this is wrong if it is.

Do I need to impedance match at my Oscilloscope with a 50 Ohm on a T?

We have to tweak the alignment of our PDH setup a bit, as well as change the cavity and profile the beam at different locations.

So we figure this is as good a time as any to stop using Aidan's laser and put in our one. We'll be working on that today, and we'll try to minimize the disruption this will cause to the fiber noise experiment.

Running out of disk space is sad:

1. Setup a cron to text Dmass and Aidan when the disk gets to 95%.
2. Talk to Alex and then get 2-4 TB of space. We don't need raid yet.

TL;DR WE SHOULD GET A NEW HARD DRIVE THAT IS BIGGER FOR ACTUAL LOOKBACK TIME.

• 4 2TB drives in raid? Redundant
• Maybe these for 220$ each? (options)

So me and Aidan were looking at the DAQ and trying to figure out our lookback because things weren't working. I deleted all the data in frames and kept the number of files.

In the current configuration we have:

• 60 files in /frames/full each of which store 1 hour of data.
• Each /frames/full/dataxx folder contains 225 *.gwf files 16 seconds apart each, for a total of 1 hour of data.
• These are each 16255955 bytes. So each frames data file is 3657589875. (3.66 Gigs)
• A total size of 204.3 GB

I did a "df -h" and found that the frames drive (or partition) is 275 GB

I did a "du -hs frames/trend" and found that we already have 90 GB of data written in the trend section. (this is bad...275 < 204+90)

du -hs /frames/trend/minute/raw = 30 GB

The /raw directories contained ALL the DAQ channels ever from the old OMS build @ 27 MB each. I deleted all these files. but they were immediately remade and are growing. Fixed?

Someone either added, or we never commented out all the epics channels in

/cvs/cds/caltech/chans/daq/C0EDCU.ini

in /cvs/cds/caltech.target/fb.

I edited this out and restarted the frontend, and it seems we aren't giving away 30 GB that we need anymore.

We also have redundant channels in the "NOT YET RECORDING TRENDS ABOUT THE PSL" in this file. Half are prefixed with C2, half are prefixed with C3. Once we get one of these halves to work, we should delete the others (probably in /cvs/cds/caltech/chans/daq/C2EDCU.ini) These are 5MB each and there are roughly 50 of them.

I don't know if/how the size of the /frames/full/dataxx/blahblah.gwf files scale with number of DAQ channels we have set, but we use 27 MB in /frames/trend/minute/raw for each DAQ channel we add.

I expect that something not nice will happen with the system when we try to write to the Nth frames/full file and run out of space. /frames/trend is now at 60 GB, and I didn't delete anything that was already there, so we should have enough space to get a full /frames/full cycle of 205 GB if this doesn't grow.

We adjusted the cable lengths going into IN-LOOP mixer in the fiber experiment so that the phase difference between the 80MHz signal of both cables was minimized (probably now < 5 degrees). This was done by looking at a long term average of these signals on the network analyzer. I suppose we could also look at the DC value coming from IF port of the mixer as well when the loop is closed.

Anyway, this allowed us to increase the gain by 500x on the error signal and we saw a huge stabilization in the noise on the out-of-loop PD.

Isn't it nice when things just work ...

(No spectra right now ... no DTT )

It appears that the frame builder has stopped writing to file again. Actually what I see if the following:

1. I can't retrieve any data from the last couple of days in dataviewer or DTT
2. I can see data from three days ago.

I'm not sure why this is. Perhaps the disk is full again. Restarting the frame builder didn't help so we need to start digging a little deeper I think.

Now that we have DTT fully up and running with AWG, I took a *good* open loop transfer function for the PMC control loop.

I am sure there are some things I haven't yet understood about my loop...

In the digital world I have:

• A 1 Hz pole for control
• Two 1Hz zeros and 10 Hz poles for antidewhitening to comensate for the Piezo Driver Box.

I emailed Alex and:

Hi David,
looks like you have switched from running om1/om2 pair to running atf.
When you do that you have to alter testpoint.par file:

[controls@oms param]$ pwd
/cvs/cds/caltech/target/gds/

A number of things made the past week somewhat unproductive, until yesterday. I was at a conference Mon-Wed. last week and was sick that Friday. This Monday I finished my second progress report (now on the wiki).

Yesterday we made major progress in the setup and alignment of the cavity. It was really helpful to have optics people (Zach and Frank) down in the lab, I learned a lot about alignment. We replaced the two lenses we had initially (one before the FI and one before the EOM) with one slowly converging lens which would place the waist in the EOM. Frank also said there was no need to have two steering mirrors before different optics, since our laser height is fixed. We were initially going to use a curved mirror for the high reflectance mirror at one end of the cavity, but Rana suggested that we would get a lock faster if we just used two flat mirrors with a lens in between them (since the finessed of the cavity doesn't really matter right now). We're also not going to worry about mode-matching just yet.

The cavity is now roughly aligned. It will need some fine-tuning, but we should be able to start trying to lock today. We're going to start with an analog system to get a feel for the parameters and just to get some sort of error signal, and then later we'll switch everything over to the DAQ. That's about all for now.

A working servo loop should not have such a low openloop gain like -60dB or -80dB in the working frequency band.
Please check the setting and the math.

Ultimately any measured openloop tf should be explained by the interferometer model, the filter response, and the actuator response.
(e.g. Ph.D thesis by M.Ando P.142 http://t-munu.phys.s.u-tokyo.ac.jp/theses/ando_d.pdf )

 The channels are set up in such a way that I am measuring on either side of my injection  point with no filters in between

Oh...This should not be the openloop transfer function. You must multiply the transfer function between two measurement points
unless those two points are just before and after the injection point.

Using the Fourier tools, I took a first attempt at getting a good open loop transfer function for my PMC.

The bottom right plot is my error signal in blue, and my noise injection in red.

It has been a brisk 72 degrees for a while (~1 week?) in lab. This is apparently the proper temperature for keeping graduate students awake, and lasers cool.

The Nokia wireless device was not able to connect to ws1, and for some reason, (probably related ?) we couldn't ssh into ws1 from the outside world either.

I reset the routers, and renewed the DHCP lease. This seems to have fixed the problems.

New IPs:

131.215.113.56 (for cds - same)

131.215.114.120 (for the outside world)

I have also sent Alex an email asking him to fix our awg/test points now that he can ssh in.

So I was trying to take some transfer functions with DTT using an analog sweep.

We don't have AWG or test points, so I was trying to do it with DAQ channels. I set number of A channels as 1, set the rest, then tried to run it.

I get the error "Readback channel unavailable (); Unable to setup test channels" 

I may be SOL to take any transfer functions without either AWG or the test points working if there is some sort of handshake missing.

I am measuring the power of my PMC transmission with a Thorlabs PDA10CS. I have one after a 50:50 Beamsplitter on the output.

I measured 603 mV through a 2.0 ND filter on a 0dB gain setting with an oscilloscope (1MOhm load). This should be half my transmitted power.

I calculate: 0.603 V *(1A/1.5*10^3V)*(1W/.7A)*100 (for ND filter) = 58 mW...

I measure again with a 10 dB setting for sanity:

1.94 V * (1/4.75*10^3)/\*(1/.7)*100 = 58 mW....

BUT when I put a power meter in the same part of the beampath, I get a reading of 30 mW. My input power is 90 mW...I think I am getting at least 2/3 transmission, but am confused by my photodiode reading. It seems to imply I am getting more power out than I put in. (hmm.)

• ARE THERE ANY TRICKS TO GETTING THE POWER METER TO BEHAVE?
• OR: IS THERE SOMETHING STUPID I AM DOING IN MY CALCULATION

I also took some spectra of the tranmission with a new alignment and a new lock.

We added an out-of-loop PD (New Focus 1811) on Friday and measured the added phase noise of the fiber. This allowed us to take a demodulated output from one PD and feed it back to the VCO (Marconi) to actuate on the frequency driving the AOM and use the new PD as witness measurement of the noise reduction.

The attached spectra show a reduction in the measured phase noise (well, it's not really phase noise until it's calibrated) for a variety of different FM Deviation (transductance) settings (Hz/Volt) on the Marconi. The largest value we could get, 800kHz (per Volt?), gave us the most suppression of the fiber noise.

I'll attach a layout of the electronics, including all the gain settings on SR560s, in the next day or so.

The next thing we need to do is to make sure that all the cables we have carrying the RF signals are the right lengths to give the correct phases at each mixer and at the AOM. Simultaneously, we need to calibrate everything very carefully now.

(I'm curious as to why the signal was much larger at an FM Devn of 100kHz. Also, I'm not sure why the 800kHz spectra saturates at a lower level than the free-running noise. I checked the DC output of the PD with an without the feedback and it appeared to be unchanged.

Morale of the story: we need to fully understand the system we have before we move any further forward.

The war is now at an end. Aidan is now free to use the Marconi, and I am using the SRS DS345 we picked up to generate 13 dBm of 30 MHz to lock with.

In other news, Koji & I discovered that there were two dewhitening filters in the piezo driver box which were not being compensated. I put in anti dewhitening (1:10 and 1:10) at the last filter bank before the DAC, and a 1 Hz pole (:1). I was then able to lock with high gain (see attached spectrum). Koji says "it looks like your UGF is around 400 Hz" based on the spectrum. I have included a PDF for more detail.

The Red, Blue, and Green are the plots of PMC Transmission with different loop gains. I chose green (14000) to be my operating point.

When I swapped out the Marconi for the DS345 I took another spectrum of transmission with 14000 gain. That is the brown curve. I am now happy using the DS345*

*subject to change without notice

Spectrum is PMC ERROR SIGNAL WITH VARIOUS LOOP GAINS

I grabbed some lenses from the nether regions of the lab, and have managed to get the PMC relocked. I did a cursory alignment, and 10% transmission seems to be my max for now. I haven't optimized my mode matching because I am changing lenses tomorrow. I spent most of the night trying to optimize my alignment and getting a good number for transmitted power.

I was still:

• Saturating my photodiode on the PMC reflection by not dumping most of the beam. I put a 90% splitter in front of it w/ a dump
• I am sure there are more things.

 Each time I changed something I put in 90 degrees of phase via cable into my mixer to make sure I haven't unintentionally fubared the phase

The PDA10CS from Thorlabs appears to have a peculiarity, or I am just confused. The Spec Sheet says that it should have a 10V max output for a Hi Z load. This appears to not be true for its 0 dB gain setting. It rails at 5.44V with zero dB of gain, and 10.something V at higher gain settings. This caused some modicum of confusion.

It was plugged into a Tektronix 1001B which is high Z (I measured to double check - 1MOhm).

Is it fine to use ND filters with my photodiodes? It seems fine, but I don't know if there is somet hing I am missing there.

So it turns out that I was putting my 35MHz signal from the Marconi (13 dBm before a spilitter) without any impedance matching and probably causing all kinds of havoc with my RF reflections.

I changed the setup so I have a 50ohm resistor T'ed into the input for the EOM at the PSL enclosure. My error signal is now 10 mV rather than 1ish without the 560.

*** check RF phase ***

We inserted an SR560 with G=+100 as a preamplifier for the mixer output so that we can have a ~100mV error signal for a cavity swing.
The error signal of ~mV with the offset of ~mV (presumably coming from both the ADC and the mixer) is a totally wrong approach.

Ah, you don't need the 50Ohm termination at the ADC input any more as we are handling only AF signals there.
It just make the signal amplitude half.

I called upon Koji to assist me in my woes.

We got a lock by turning off the digital filter. We could not lock with it on at any gain we tried

I will take a transfer function with the 35670A and try to get at the real transfer function.

Sure would be nice to have AWG

The attached time series shows measurements of the power in the s and p polarizations (via ThorLabs PDA10CS PDs) for different input HWP settings. This just shows the raw data from the experiment  described here ...

I spent some time tonight realigning and relocking (attempting to relock) the PMC.

I got to a point where I did not understand the behavior of my error signal. I did the standard things...

• Changed cable length by pi/2
• Eventually realized that I had done this all before and looked at my old elog to not do the same dumb things again
• Same setup as this elog entry
• Change the offsets muchly
• Change the gain
• Change the sign of the gain
• Make sure just a single pole @ 1Hz is toggled on
  

I get a behavior where my transmitted beam would oscillate in power, seemingly about the center, and my error signal looked as below

When I say error signal I mean "Thing that should be my error signal but for some reason is not" this is the raw signal I am inputting into the DAQ - so this is what I pass to my filters then to teh PZTChange the offsets

Aidan and I matched the polarization of the beam to the polarization-maintaining fiber.

To begin with, we sent the output beam through a polarizing beamsplitter and directed both the transmitted (P-polarized) and reflected (S-polarized) light into photodetectors. We oriented the output fiber axes such that the reflected (S-polarized) power is minimized.

Next, we wanted to orient a HWP at the input of the fiber such that fluctuations in the reflected (S-polarized) output beam are minimized. This would mean that the polarization of the incoming light is matched closely to the fast axis of the fiber, because fiber length fluctuations don't cause polarization changes in the output beam.

To find the amplitude of fluctuations in the reflected (S-polarized) output, we aimed a heat gun onto the fiber, lengthening it by multiple wavelengths so that the photodetector output reached clear minimum and maximum values. We recorded the difference in these values.  I will soon attach a plot of the reflected output amplitude fluctuations vs HWP angle.

Attached are photographs of the current input and output layouts.  The angle with minimum output fluctuations is 356.4 degrees.

We put a HWP immediately before the input to the fiber to match the polarization to the fast axis of the fiber. We can see the effect of this in the contrast of the fringes in the demodulated AC output of the interferometer. As the fiber expands and contracts the output polarization becomes more or less circularly polarized depending on how much the input polarization is mis-aligned from the fast axis. Below is a time series of the demodulated output of the PD over about 45 minutes. We need to optimize the position of the HWP but that's a job for tomorrow.

Aidan and I generated power spectra of uncalibrated phase noise, reference only intensity noise, fiber only intensity noise, DAQ noise, and dark noise using the DTT to create a noise budget.  We made a shot noise calculation and imported it onto the same plot.

The intensity noise of each arm seems to be limited by dark noise. The spectra are attached.

Our next goal is to match the polarization of the fiber arm beam to the fiber's fast axis.

 The current demodulation setup:

Aidan and I are acquiring a beat signal from the interference of a reference beam and a frequency-modulated beam. The demodulated output can be seen on THIS ENTRY.

I have attached a photograph of our setup below. The reference beam is a pickoff from the 1W NPRO, and the zeroth order AOM beam is blocked by an iris.

Aidan will shortly post a schematic of the electronics, and we hope to start actuating on the laser frequency later today.

Connor and I have set up an 80MHz optical beat on the photodetector in the fiber setup, (one beam is passed through an AOM to freq. shift, is passed through the fiber and then recombined with a reference), demodulated the output from the PD and acquired it in the DAQ.

A 3 hour+ time series of the demodulated output is shown below. The variation in the maxima and minima over the 3 hours is most likely due to fluctuations in the output polarization state from the fiber due to the input not being completely matched to th fast axis of the PM fiber.

We're going to insert a HWP before the fiber to adjust the input polarization so that it is aligned with the axis of the fiber.

Aidan and I wanted to know the Mach-Zehnder reference beam power as a function of the angle of a half-waveplate which, coupled with a polarizing beamsplitter, attenuates the beam. Between the half-waveplate and the Mach-Zehnder arm, the reference beam is attenuated by a 50/50 beamsplitter.

Here is our data:

Angles (degrees from vertical of fast axis): [30, 36, 42, 48, 54, 56, 58, 60, 62, 64, 66, 68, 70, 72, 74, 76, 80, 86, 92, 98, 104, 110]

Power (mW): [397, 328, 237, 146, 70, 47, 31, 17, 7, 4, 2, 7, 14, 26, 41, 60, 110, 197, 291, 372, 430, 459]

I've attached a plot of our data with a least squares fit of a cos^2 function parametrized by amplitude, frequency, phase shift, and offset.

 I checked the relative stability of each of the two mode matching solutions here by perturbing the lens positions and focal lengths.

 I used the Anderson formulas <add link> to calculate the power lost to first order into higher order modes. I then altered parameters (lens positions and focal lengths) as noted in the following tables and compared changes in power loss.

I choose the .5m and .5m solution.

 For lenses, I plan on using a plano-convex lens for as lens 1, and a biconvex lens as lens2.

The 1W NPRO beam is elliptically polarized. We use a quarter-waveplate (QWP) to linearize the polarization.

7 days ago, I made some measurements of the extinction ratio of the beam for various orientations of the QWP (attenuating the beam with a half-waveplate and polarizing beamsplitter) to find the QWP orientation at which the light is linearly polarized. The minimum ratio occured when the fast axis was between 10 and 15 degrees from the vertical. I've attached a plot of the data, and I will fit it to a curve as soon as I work out which function should describe the data.

10:10AM : need to align AOM and check RF drive into it.
    AOM - wants < 2W RF Power Level
    Amplifier gain [ZHL-1A http://www.minicircuits.com/pdfs/ZHL-1A.pdf] - 16dB
    Maximum power output (dBm) = +28dBm = 630mW

14:00: Have AOM aligned. 3Vpp amplitude 80MHz signal to power amp gives ~63% diffraction efficiency [-1 order: 100+/-3mW, total: 160+/-3mW]

16:40: -1 order is aligned into fiber - coupler is very touchy and sensitive to misalignments. looks to be good transmission though. 37mW incident, 7mW transmitted - about 19%!

Update:

Today I aligned the laser beam through the EOM with something resembling a normal shape at the output. This was tricky. The powermeter isn't giving me very reasonable readouts, the aperture on the EOM is impossible to see with the IR viewer, the EOM is in the middle of the table where I can't really reach it, and it's very close to a focusing lens. The point being that I may have to tweak it a bit, but I really don't want to have to redo the alignment of that particular optic. So no earthquakes for a while.

A note on the power meter: It's been registering anywhere between -6 mW and 11 mW with no beam on it at all. I've been zeroing it before putting it in the beam before each use, but I don't know how much I trust it.

The beam shape coming out of the EOM  is still a bit funny - with one bright beam, and then a very faint ring, so it looks like a ring you would wear on your finger with a diamond on top. I am attributing this to the fact that the beam is slightly larger than the EOM's aperture; I have set it up so the center of the EOM coincides with the beam's waist. I'm also (according to the power meter) getting a high attenuation inside the EOM, but whether this is correct or normal I haven't yet figured out.

We need another mount for a beam splitter, this will go at the input to the cavity. We are also still in need of a quarter waveplate, and Aidan says he has a spare that he won't need to use for a few weeks. Also, CVI doesn't specify the damage threshold for the cavity mirrors for cw lasers, only for pulsed sources. Alastair has sent them an email requesting this information, and once we know that we can send the beam into the cavity at appropriate power levels.

Our lens positions for mode matching into the PMC seem to be based ON LIES. I cannot locate my old data on the first PSL scans and fitting. I have now learned a very important lesson in imitating Aidan.

Since my old "measurement" (quoted number) seems to disagree with what I have found by roughly a sign, I am operating under the assumption that I was simply wrong in the linked entry.

I have plotted our old mode matching solution with the new beam parameters.

I have also found 2 mode matching solutions for the "average" mode coming out of the PSL, one with two 500mm lenses, and an alternative in case we don't have them around.

If these do not suffice for our locking needs, we will need to deal with the ellipticity of the beam. I think it will be fine for now based on this.

I have included plots of the two offered mode matching solutions, as well as the picture of what we were using for coupling into the cavity.

Connor and I have been assembling the FS experiment. The following diagram shows the layout of everything that is aligned at the moment.

We've got the AOM installed and have it aligned to give -1 order when driven at 80MHz (attached image shows main 0th order and -1 order to the right. Not sure what spot up the top is).

I've added our medm screens from the ATF to the 40m SVN. They can be found in 

https://nodus.ligo.caltech.edu:30889/svn/trunk/medm/c2/atf/

As title says ...

This happened mysteriously and had absolutely nothing to with me. The fact that I was the last person to open the filing cabinet before this happened is circumstantial and beside the point.

Will get the lockshop onto this in the next couple of days. In the meantime, just try and exercise your clairvoyance.

Connor and I got the beam aligned into the fiber today. After a little optimization of the fiber coupled XYZ I'd estimate we were getting maybe 10-15% transmission. I'm pretty sure we can improve this with some work. We caculated the input mode (which I will reproduce elsewhere) but it's looking for a mode size of approximately 6.6 microns diameter. With a 10X microscope objective (f = 16.5mm) we want the intput beam to the coupler to be around 3.4mm diameter - which we have.

We mounted the AOM and inserted it into the setup. We should be able to get an 80MHz frequency shifted beam through the fiber tomorrow measure the fiber noise. That same setup can be used in a PLL to suppress the fiber noise.