That would probably be better, but I'm just replacing like with like in the lens kit just now.

I thought we wanted the straight up 1064 only coating on these (AR.33)?

According to Newport:

• R_avg <0.5% for AR.18
•  R_avg <0.25% for AR.33

 Gina has checked through the PO and Newport should definitely have received it, but they claim they did not.  This is the third time that I know of recently that we've had trouble with ordering this way from Newport.

---  The new order has now been put through and has arrived at Newport ---

I chased this up this morning.  Even though I made the order directly through techmart (not as a spot buy) Newport have not received it.  I'll put it in again.  The lenses will have the AR18 anti reflection coating for 1000-1550nm.

I'm ordering two each of the following:

KPX103 - f=175

KPX106 - f=200

KPX109 - f=250

KPX112 - f=300

KPX115 - f=400

KPX118 - f=500

KPX121 - f=700

KPX124 - f=1000

I'm ordering one each of the following:

KPX100 - f=150

KPX097 - f=125

KPX094 - f=100

KPX091 - f=88.3

KPX088 - f=75.6

KPX085 - f=62.9

KPX082 - f=50.2

KPC037 - f=-75

KPC034 - f=-100

KPC028 - f=-200

 What focal lengths/coatings

I ordered some lenses for the lens kit.

Servo diagram unclear. Requires photos and block diagram. New online diagram tool here.

Plot open loop gain - measure using HP/Agilent audio frequency analyzer.

Today, Frank, Alastair and I played with one of the fancy blue PDH servo boxes. It was shiny.

The cutoff frequencies weren't right, so we ended up bypassing one stage so that the overall transfer function was just one pole at 4.8 Hz, with a flat gain of 60 dB from there back to DC (without the boost--with the boost on, the circuit continues to integrate back to infinite gain at DC). This was already far superior to our old setup with the SR560s, because we had over twice the output range. We also switched to the PDA255 photodiode with a factor of ten increase in our error signal. Using this, we swept the cavity over the full output range and were able to finally measure our finesse, which is ~250.

To try and lock, we used a ground-coupled SR560 to trim the DC offset so that we were near a resonance peak, and then slowly ramped the gain on the PDH box until we saw a decent actuation signal. Unfortunately, changing the box's gain produces an additional offset, so we had to adjust both simultaneously. Eventually, we could lock for a short period before low-frequency oscillations pushed us out of range.

Next, we did two things: first, Alastair fashioned a nifty enclosure out of vacuum-grade foil and Kapton tape to block out air currents. We placed this over the cavity and it seemed to work quite well (it also helped block out visible light glare, and we could then see the spot image from the CCD camera without the lights off). Second, we fed the output of the PDH box through a voltage divider to an SR560, which we set to have one pole at 1 Hz with G = 1 (to steepen the rolloff even more beyond the low-frequency actuation range). This led to decent results; the cavity can now be locked with little trouble and will stay locked barring an earthquake or someone playing whack-a-mole on the other side of the table, but the lock itself is not very clean (oscillations near 25 kHz) and not much gain can be put on the temperature control at all before the loop becomes unstable.

Pictures will be uploaded tomorrow for fear of angering the elog gods again..

Changed the set point on the temperature in the lab to 61F from 64F.

When testing the performance of a loop in the range between 50 and 200Hz, I recommend using The Dark Knight Soundtrack - Track 11 as an acoustic noise source.

After improving the overlap between the two beams on the PDs and increasing the power incident on each PD, I closed the loop again.

The loop looks like this:

PD1---Demodulated_AC---SR560---Marconi_FM_IN---AOM---PD1

I played around with the SR560 settings and the Marconi settings. The spectra of phase noise are not calibrated although the saturation in the open loop signal indicates the number of counts that correspond to pi radians. The DAQ noise floor is just below 10^-2 counts.

I played around with the phase of the LO in the out of loop path so that the demoudlated AC signal on it was centered around 0V.

At this stage I need to add a whitening filter and amplify the signal before putting it into the DAQ. I think I'll try closing the loop with the DAQ output as well. Also, I don't understand why the out-of-loop signal appears to be saturating at lower frequencies when the loop is closed.

Note: FM Devn is the deviation in the output frequency of the Marconi per unit volt (rms) [Units: Hz/V]

Today was a day of much tinkering. 

Yesterday, we got the new triangular cavity aligned and got a decent error signal out, but were unable to keep the thing locked. Even worse, when it was "locked" it was oscillating like crazy, just in a uniform-enough way for us to call it a sort of sorry excuse for a real lock. Today, we leveled up from "marginally metastable" to just "metastable", in the sense that we are now able to get the cavity locked in the usual sense of the word, but not for very long.

Out of curiosity, Alastair and I decided to feed the error signal to each of our two SR560s (one for PZT, one for Temp) separately, so as not to piggyback one off the other for the temperature control loop. We realized after discussing it with a few people (Aidan, DMass, Rob) that this doesn't make matters much simpler because the stability of the overall scheme depends on the nature of the crossover point between the two actuation channels anyway. The bottom line behind most of the advice was to go back to basics, so we did; we removed the temperature actuation, and focused our efforts on the PZT alone.

It was pretty obvious that the PZT loop didn't have enough gain at high frequency (~200 Hz), because we were seeing a crap-ton of error signal there and the poles we had for our earlier cavity (see previous post) were much, much lower. I tried a variety of higher cutoff frequencies--still using just one pole--and nothing seemed to work. Eventually I moved to two poles, and finally I got the thing to lock smoothly at f_c = 300 Hz. Any pole frequency above or below that resulted in ugly oscillations during "lock". My suspicion is that with it any lower there simply isn't enough gain where our error signal is, while with it any higher our phase margin begins to dwindle.

In any case, this solved one problem. The new problem was that I was using the PZT to control everything from DC to ~300 Hz, and while it did fine at high frequency, the cavity wouldn't stay locked for very long before the SR560's output railed from lower frequency oscillations. In fact, I could actually sit there with my hand on the post of the end mirror, apply slight pressure to get the system near a peak, and once the PZT locked at high frequency I could simply add or release pressure on the post to keep the low-frequency drift of the PZT actuation signal from reaching either rail. Doing this, I could keep the cavity locked almost indefinitely. So, I then tried to feed back to the temperature control to keep low-frequency junk in check, only to find that long before I could apply anywhere near enough gain to do anything, the new addition seemed to totally wreck the stability of the PZT loop. 

"Why not put a bandpass filter on the PZT loop, instead of just two poles at 300 Hz, and then a lowpass on the Temp loop so that it just actuates at low frequencies?" you might ask.

"I would," I might reply, "but the SR560 doesn't let you have a bandpass while also having two poles at the high end". 

It seems that the only option is to use another kind of amp (or the DAQ), which will either let me (a.) put out a higher voltage, so that I can use broadband actuation on the PZT without railing at low frequency, or (b.) use a more complicated filter scheme, so that I can keep the PZT from actuating at low frequency, while also rolling off fast enough on the upper end.

Of course, there's still the possibility that I'm wrong about everything.

Added another setting to the memory of the Marconi - #103.

Same as memory setting #102 but with the added 100Hz signal - can be used to easily calibrate the output voltages of the mixers corresponding -pi/2 and +pi/2

See http://131.215.115.52:8080/AdhikariLab/248 for the other settings.

Earlier this week, Alastair and I decided to switch from our small (0.4m roundtrip) triangular cavity with three planar mirrors and an intracavity lens to an elongated isosceles triangle (altitude = 0.6m) with two planar mirrors and one curved mirror of R = 2m. The thinking here was to cut out the losses from the lens while also minimizing astigmatic effects caused by oblique incidence on the curved mirror. Of course, for the final gyro we will want to maximize our area for a given cavity perimeter, but we felt that this was a nice place to start on a ring cavity.

We lined the thing up, and isolated the TEM_00 mode (visually using the CCD camera, and then by using Rana's patented ball driver handle method to sweep the cavity and maximize the peaks in transmitted power). What we found was that the impedance matching was terrible: our transmission line model suggested that we should get O(80%) drop in reflected power on resonance given our mirror reflectivity values, while the power drop we saw was all but nonexistent (even when we AC coupled the reflected signal so that we could crank up the scope resolution, the dips corresponding in time to the peaks in transmission were barely visible above noise).

At first, we thought that the reflectivity values we were using couldn't be trusted, since we weren't hitting our planar HR mirrors at exactly 45 degrees. We measured the transmitted light through each mirror, and the results were not far from manufacturer specs. The measured values were:

Y1-45S: T ≈ 1 e-4

Curved mirror: T ≈ 4 e-4

BS1: T ≈ 5 e-3

Pretty much spot on. (Note here that we were only using two Y1s and a curved mirror until this point).

So, we sought some guidance. Turns out that we were considerably underestimating the scattering losses of our mirrors, which can apparently be on the same order as our transmitted power (~100 ppm). We decided that we should replace the input Y1 mirror with a BS1, so that we had a "bigger hole" through which to inflate a balloon with "a bunch of other holes" in it. We did this, and--not unlike black magic--it worked; the power dips in reflection were very well defined, and the finesse of our cavity appeared to be PDG*. We hooked up the control loop, which was, from our old linear setup:

PZT loop: 1 pole @ 0.3 Hz, G = 10,000

Temp loop: PZT loop + 1 pole @ 10 Hz, G << 1 (variable voltage divider between SR560s)

With some minor adjustments to gain, this worked reasonably well. As of last night, we were able to continually achieve what might be termed a "marginally metastable" lock (didn't last forever, and when it was working it was constantly oscillating a set amplitude about resonance).

Today is a good day to play with gain settings, and to switch to using the nicer PDA255s.

Pictures to come.

*pretty damn good

the following 35W laser-system data is now available as (slow) EPICS channels (read access only so far):

[C2:PSL-AMP_MANUALMODE]
[C2:PSL-AMP_DBILOCK]
[C2:PSL-AMP_DCUROK]
[C2:PSL-AMP_ENABLETEC]
[C2:PSL-AMP_ERR]
[C2:PSL-AMP_OFF]
[C2:PSL-AMP_RESET]
[C2:PSL-AMP_DTEMPERR]
[C2:PSL-AMP_ENABLEPWRDOG]
[C2:PSL-AMP_SYSTEMOK]
[C2:PSL-AMP_LOWPWR]
[C2:PSL-AMP_CHILLERR]
[C2:PSL-AMP_DVOLT1]
[C2:PSL-AMP_DVOLT2]
[C2:PSL-AMP_PWR1]
[C2:PSL-AMP_PWR2]
[C2:PSL-AMP_PWR3]
[C2:PSL-AMP_D1TEMP]
[C2:PSL-AMP_D2TEMP]
[C2:PSL-AMP_D3TEMP]
[C2:PSL-AMP_D4TEMP]
[C2:PSL-AMP_D1PWR]
[C2:PSL-AMP_D2PWR]
[C2:PSL-AMP_D3PWR]
[C2:PSL-AMP_D4PWR]
[C2:PSL-AMP_DCUR1]
[C2:PSL-AMP_DCUR2]
[C2:PSL-AMP_XTALTEMP]
[C2:PSL-AMP_SECS]
[C2:PSL-AMP_MINS]
[C2:PSL-AMP_HRS]
[C2:PSL-AMP_DAYS]
[C2:PSL-NPRO_D1PWR]
[C2:PSL-NPRO_D2PWR]
[C2:PSL-NPRO_D1TEMPSET]
[C2:PSL-NPRO_D2TEMPSET]
[C2:PSL-NPRO_D1TEMP]
[C2:PSL-NPRO_D2TEMP]
[C2:PSL-NPRO_D1TEMPERR]
[C2:PSL-NPRO_D2TEMPERR]
[C2:PSL-NPRO_TEMPGUARD1]
[C2:PSL-NPRO_TEMPGUARD2]
[C2:PSL-NPRO_XTALTEMPSET]
[C2:PSL-NPRO_XTALTEMP]
[C2:PSL-NPRO_XTALTEMPERR]
[C2:PSL-NPRO_CURSET]
[C2:PSL-NPRO_CUR]
[C2:PSL-NPRO_NEMON]

i've checked some of them but not all, so feel free to check all channels you new and let me know if something is not working or should be changed. i also have to add the units for each channel...

in order to have access to the laser data i had to change the network a bit. the laser and the windows computer (T42 notebook) are now in the real-time segment (as all other machines). IP-adresses are assigned via DHCP, might change this to fixed ones later...if there are any problems please let me know...

You should try just leaving FM4 on all the time.  Since the purpose of this FM is to make the loop 1/f everywhere from 1Hz to several kHz, the loop should be stable over very large gain ranges.  This makes locking much easier.

Filt 1 integrates.

Infinite gain at DC.

There's constant input.

Came in and found the PMC_PZT output trying to deliver an obscenely large number of counts to the DAC.

Not really sure where the issue is coming from - looks like the filters in the attached module. Anyway, just set a limit on what that module can output for the time being. Will leave this one for DMass to figure out.

Frank was talking about rebooting the frame builder last night. That shouldn't have affected the front end though. But if we do want to reboot the front end will all our settings be saved and restored automatically?

I improved the alignment into the recombination beam splitter using the TV camera to overlap the beams in the near and far fields of the splitter - essentially making them as colinear as possible. As  fixed the position of PD2 so that it was at the focus of the focussing lens and optimized the position of the beams on the two photodetectors.

I remeasured the AC and DC outputs from the photodetectors. The results are below:

overlap = AC@80MHz/(2*sqrt[FiberDC*RefDC])

This leaves the beam size as the culprit for the relatively poor overlap. The reference beam size is noticably bigger than the fiber beam size on the CCD. I'll see if I can improve this ...

I had left some of my output filters off on accident from doing a cavity sweep. This should be fixed.

Sometimes scopes are in their 10x probe mode which gives a factor of 10.

Spotted a ton of accumulating overflows on the DAQ - specifically the output for DMass's PZT on his PMC.

It looks, from the EPICS screen, like that channel is trying to output a lower number of counts (~ -90000) than the DAQ can reasonably drive. (+/- 20V = 64K counts). We should try and set upper and lower limits on the channels. I'm not sure why this isn't set automatically in the RCG.

Also made a note on the ATF wiki ...

Seems to be a non-fatal error though.

Whenever I come in to make a measurement, more often than not DTT is nonresponsive. This is fixed by

telnet fb0 8087

daqd>shutdown

But it seems this should not be happening, and may have other negative consequences for us.

While trying to calibrate my channels, I noticed a factor of 10 difference between power according to a PDA10CS from thorlabs (as measured on a Hi-Z scope) and the power according to the newport wand I was using.

Turns out the Tektronix TDS1001B oscilloscope was giving me traces that are a factor of 10 higher than they actually are.

I double checked with one of the LASCAR DC power supplies, and got a reading on my scope of 150V when I put 15V in.

I tried pressing the "default setup" button, but I still had the extra factor of 10.

I have pulled it and left it on the bench with an "NFG?" sticker on it.

I used a Newport 818-ST-UV power meter wand to measure the mode matching of the PMC.

Someone had calibrated the wand for use with its built in OD3 screen, so I think it was fine to 400 mW on the head, since that is 400 microwatts at the detector. 

• I measured an offset in the wand of 20 uW
• I measured a transmitted power of 246 mW with the PMC locked
• I measured a reflected power of 392 with the PMC unlocked
• These numbers resemble what I measured with the big fat large power head
• So I think this means my mode matching is 246/392 = 63%

I may need to do some overlap integrals to figure out if this is reasonable coupling for our elliptical input beam.

Having checked the wand versus a PDA10CS with ~590 uW on it, I found agreement between the Newport wand and the PD to within 5%.

 This previous entry showed a reduction in fiber noise once our loop was closed.

Rana commented that the noise floor of the New Focus 1811 PD seemed too high relative to the signal. After some measurements of the power we calculated the overlap between the two interferometer beams was substantially smaller than it should've been, hence dramatically reducing our 80MHz signal and raising the relative noise floor of the 1811 PDs. 

After making the spectra, but before performing this calibration, we moved the out of loop PD around a little, so its numbers are a little different from what they would've been during the recording of the spectra. The in-loop PD calibration should be good though.

* 1811: DC Transimpedance = 1V/mA, Responsivity = 0.75 A/W 

This suggests that PD2 is not quite at the focus (at least for this measurement - not sure for noise spectra) and the alignment onto PD1 needs to be improved.

Next we measured the AC Levels out of the photodetectors. We also measured the incident power in the two beams, calculated from the DC output voltage when one of the beams was blocked. From this we could calculate the overlap between the two beams. It is very poor.

* 1811: AC Transimpedance = 40V/mA, Responsivity = 0.75 A/W 

 Clearly we need to improve the alignment of the two beams at the output of the interferometer in order to decrease the relative size of the electronics noise to much lower levels.

What I helped was only ... this

With Koji's help, I figured out that my problem with the prism was an insufficiently steep angle of incidence. I now have split the beams nicely. The prism is uncoated, so I have visible reflection losses (which I am dumping with razor blades).

I turned up my input power by a factor of 4 (to 400 mW) and turned down my loop gain accordingly. I seem to still not be saturating the refl PD (thankfully)

Using a lens with the wrong AR coating, I focused the 532 onto a Thorlabs PDA10A. I swept the temperature with my Newport 3040 temperature controller and fit a Sinc^2 curve to it.

My fit tells me that what the Newport 3040 readback of my RTD calls "37.42 C" is my optimum temperature operating point. This is a 2-lead measurement with a prefabbed calibration curve on the resistor, so that should not be used as an absolute number.

Using the responsivity and gain of the photodiode, I estimate that I have about 290 microwatts of green. This is after some loss due to reflection at the prism face, and some loss due to reflection at the UV coated lens I'm using to focus the green.

Leaving the laser on, taking more long measurements.

Next up: Calibrated RIN at each point in the path

Running a long one overnight. PSL table is hot.

Using a photodiode on the PMC transmission I optimized:

• The alignment of common and differential DOFs of the steering mirrors into the PMC
• The mode matching (waist size and position) by moving the (temporary?) mode matching lenses

I eyeballed the SHG Oven to be collinear with the beam into it, and aligned into the aperture via knobs and looking at the transmitted spot.

I adjusted the position of my f = 5 cm "mode matching" lens right before the oven by looking at the brightness of the green spot transmitted. This was easy to do.

• I emailed Raicol (the crystal makers) and they told me they had no data for damage thresholds of PPKTP.
• N.B. on the above: Kirk says 1W in 30 microns is probably fine. This is something under 100 mW in 25 microns, so I should be OK.

I tried to eyeball where the optimum temperature of the oven (and thus PPKTP) was based on the brightness of the spot.

I played around with the Thorlabs Equilateral Dispersing Prism that I bought some time ago in an attempt to split the beams, but was unable to get it to do anything intelligent

• I tried most orientations I could think of, wiggling the thing around
• I tried both S and P (I have a 45 degree QWP in the beam path with pickoffs)
• I was not able to get any green or 1064 transmitted through the second face.
• Maybe there is something non intuitive about this, or maybe I missed what the thing was supposed to do...

I also played with a short pass filter but apparently this doesn't reject past 800 nm or so. I see no stock filter which rejects 1064 and passes 532.

Up Next/Soon:

• Make some calibrated measurement of PMC transmission (within 1% or so)
• Split the 532 and 1064 and get a measurement of my SHG efficiency
• Sweep the temperature of my oven and get a phase matching curve

Uncalibrated phase noise spectra from free-running interferometer + noise source.

We increased the AC voltage from the PDs by +33dB on PD2 and about +25dB on PD1. This was done by improving the alignment onto the PDs (giving about 2 and 10 dB for PD1 and PD2 respectively) and by increasing the power by about 15 times.

The AC and DC outputs are now going directly into the DAQ with no filtering or gain. The channels are as follows:

• PD1-DC = [CH17] = C2:ATF-QPD_SEG1_IN1_DAQ
• PD1-AC = [CH18] = C2:ATF-QPD_SEG2_IN1_DAQ
• PD2-DC = [CH19] = C2:ATF-QPD_SEG1_IN1_DAQ
• PD2-AC = [CH20] = C2:ATF-QPD_SEG1_IN1_DAQ

We also cleaned up some of the BNC cables around the PDs. 

I added in some boost to the PMC Loop. The loop is now locked with a net:

Poles: 10 25 0

Zeros 150 50

And some gain to make my boost stages actually boost. zpk(150,10,15) and zpk(50,25,2) are my "boost" stages.

My locking algorithm:

• Get close to the fringe and let the system spaz out using my 72 Hz PMC PZT pole for locking
• Engage FM4, which is a zero at 72 Hz and a pole at 1 Hz
• Engage my Boost stages
• Engage my integrator (zero at 1 Hz and a pole at zero)

Results of the loop with the boost on and off (with slightly different gain settings) are attached in pdfs

RED: Boosts on, integrator on, gain slightly lower to move my UGF (~ factor of 2)

GREEN: No boost, integrator on

BLUE: Some old spectrum of the PMC Transmission light. I don't *believe* that this calibration has changed, but could be wrong.

PD1_IN1 is my PMC Transmission

ISS1 is my error signal

 never mind.

Going to order 15x 2.5" pedestals from New Focus. Will submit order this afternoon.

What does the "rules for all servos" book say about where integrators belong?

Rules for all Servos:

1 - Compensate the whitening in the first filter modules.

2 - Compensate the dewhitening / actuation (e.g. PZT) on the output modules.

3 - Use FM4 to make the servo 1/f.

4 - Put the boosts, etc. into FM1-FM3. FM3 should be the first boost you turn on, FM2 the second, etc.

I talked to Rob and now understand what was happening. I was engaging changes to the gain downstream of the integrator, after it was engaged. This naturally caused some (notsosmall) problems.

I will change the filter modules accordingly, and expect to have nice noise spectra for my error signal tomorrow.

I made some changes:

I am now using an SR560 as an amp for my input to the ADC. Settings:

• Low noise
• Gain = 100
• DC Coupling

Also it turns out I was locking the PMC with a 1 Hz pole and a 72 Hz pole.

The 1 Hz pole was from my digital system, and the 72 Hz pole was from my PZT on the PMC.

I measured the capacitance of the PMC PZT to be 221 nF.

The piezo driver box has (according to schematics pulled by Rana) a 10 kOhm resistor.

This gives a 72 Hz pole so...

I put a 0 Hz pole and a 72 Hz zero (in zpk parlance "72:0") into the filter bank. Locking was swift after turning this on when close to the carrier.

I also tried to put in a "boost" stage:

A 24 Hz Pole and a 48 Hz zero with gain of 2. I looked at the phase in foton, and it wasn't much - my open loop transfer function seems to imply that I have it to spare

I played with the ramp time some, but every time I tried to click on this filter, I lost lock. I do not understand/know what is wrong yet.

Top Plot:

• Green is new loop (72:0 engaged), Red is old loop.
• ISS_1 is the (uncalibrated) error signal - multiplied by 100 (SR560)
• PD_1_IN1 is the (uncalibrated) transmitted PMC power

Bottom Plot  (pdf)

• Red and blue are old weird loops
• Green is my current loop

My 1/f now appears to be 1/f!

I took the top SR560 from Alastair's side of the other table, as all 4 were taken up by that table, and Alastairs were the only ones which were off, so I assumed it would be easier than looking through all Aidan's cables to figure out his setup.

I want to use one SR560 indefinitely as a low noise preamp of my error signal into the ADC. If Aidan and Alastair both need 2, we may need to procure more.

I also trimmed the offset with a glasses screwdriver so that it roughly read 0 mV.

The old setting was: -56.0mV as read on a tektronix TDS3032 with settings:

• DC Coupling,
• No filters,
• Terminators on channels A and B
• Invert off.
• Gain = 10
• Low noise mode

The cavity is locked! With some aligning trickery this morning and guidance from Rana this afternoon, we were able to stabilize the laser. Here are some pictures:

Tomorrow we have to work on getting it a bit better (and by that I of course mean maximizing the gain, starting at the top of the hierarchy). Anytime the temperature rails, the cavity goes slightly off resonance. But we can block the beam, hit the table and shake the mirrors and it stays on lock. =)

I've ordered the following:

Mini Circuits
--------------
- ZX05-1MHW-S+ (mixers) x4
- SHP-100+ (High pass filter) x4
- SLP-1.9+ (low pass filter) x4
- ZFSC-2-1W-S+ (splitter) x2

DMass and I have been using the Marconi a lot recently, each with different settings. I saved these into its memory and they can now be recalled easily by doing the following:
1. Press [RCL]
2. Type setting number
3. Press [ENTER]

Settings
DMass 35.5MHz - memory# 101
Aidan 80.0MHz - memory# 102

They are shown in full in the photos below ...

BeamScan with rotating slits does not compensate the ellipticity but gives you two orthogonal slices of the beam spot
through its slits on the drum. This means that BeamScan gives the same kind of information as that of razor blades.
Both are not the true 2 dimensional scanning.

In general the angle of BeamScan should be aligned so that you can get the maximum ratio of the widths
along the semimajor and semiminor axes.  cf. If you have the vertical/horizontal astigmatism, for example,
BeamScan rotated by 45 deg gives you the same widths for orthogonal axes.

It is not surprising that you saw the astigmatic beam. An LWE Misers has an NPRO (Non-Planar Ring Oscillator)
as the master oscillator. This is a ring cavity with non-right angle reflection on a curved surface. This means that
the cavity has different geometries for the vert and horiz directions. BTW, the ring cavity of the laser gyro experiment
will have the same issue.

The attached figure is an example of the astigmatic beam of Miser (LWE 50mW) and its correction by cylindrical lenses.

You just can confirm that the astigmatism is along with the vert/horiz axes by rotating BeamScan by 45deg as described above.

I was testing the Wenzel to compare it to the other oscillators we have, and managed to hook up the leads backwards after retwisting the pairs. I cooked the something that was connected to the power input terminals. Picture coming soon. I will not be using the crystal for the moment. Maybe everything will be fine when I swap out new components.

Used DMass's bidirectional coupler setup and measured the reflected power from the lab AOMs for a sweep between 30 and 110MHz. Also tested a 50Ohm terminator and an unterminated setup to calibrate the equipment.

Results shown in the attached plot.

Fixed! The key is now hanging in the second bay from the right above the main bench.

Noise budget from out of loop PD with 10,000x gain on SR560. Loop gain isn't yet measured.

In the current setup we have only one in-loop photodiode and are therefore limited by intensity noise - only it won't be apparent in the out of loop detector. But with only a single PD the true frequency noise will be limited by the intensity noise. The next set up will reduce the coupling of intensity noise to frequency noise. (Im too tired to write out why it is now, but it's pretty straightforward).

Next step is to double pass the fiber, include a second AOM and switch in the new fiber coupler.