Lab Journal F2024

[Jashcraf]

09/25/2024

First attempt operating the quasi-remote Derp thanks to @becca9808 @Ronald_Lopez (Ron what’s your GH Username). Things we did:

• Show RL how to turn on SuperK, Camera, and motorized stages
• Review current setup notebook Getting Started
• Changed the achromat in Derpy from $f = 50mm$ to $f = 60mm$ so that we can get before and out of focus
• Set the achromat to be in a near-telecentric condition with the Wollaston prism for easier alignment
• Tried to focus the camera manually, but it wasn’t clear if we were able to

Next Steps

• [x] #5
• [x] #6
• [x] #7
• [x] #8
• [ ] #9
• [ ] #3
• [x] #10
• [x] #12

[Jashcraf]

10/02/24

Delayed lab journal! On wednesday we tried to tackle an auto-focusing algorithm to get near pupil and focal planes. Unfortunately we kept running up against a problem where the camera sometimes wouldn’t see any flux. What we found after some poking around was that CRED2 was running near the pulse frequency of the laser (curse you fast camera) because we were getting so much flux we really pushed down the integration time.

We also found out that when the FPS is set, it appears to maximize the integration time CRED2.tint given the FPS. So, set the FPS before the integration time.

Summary of Changes

• changed the achromat in Depry from $f = 60mm$ to $f = 75mm$ for more margin before focus
• cried a little bit

Next Steps

• Add an additional manually-rotatable polarizer before the PSG for flux control

[Jashcraf]

10/09/2024

Delayed lab journal! This week we mainly introduced Briley to the laboratory and attempted to focus the camera on the newly-installed sample stage. It looks like we can't get there while maintaining the magnification that we want. We were able to use the additional flux control to manage the saturation on the detector though!

[Jashcraf]

10/16/2024

We are trying to do a calibration run of the DRRP. Set up a calibration routine and got all the way through to taking data. Can rotate from 0 -> 180 deg with 36 measurements in 3mins, very zoomy!

Installed Katsu to try and fit the data to a model to understand the orientation of the polarizers and waveplates. Katsu can do a forward pass in 200 microseconds on Rayleigh, so this should swan dive on a solution pretty quickly (if not necessarily accurately).

Next time we will try and actually do the fit.

Laser settings

• 4% Outyput power
• 50% wavelength power at 1782nm

[Jashcraf]

10/23/2024

Re-trying the DRRP calibration notebook, using the settings from above.

Here we just rotated the PSG retarder - interestingly, the two channels are not exact opposites of eachother

We had some difficulty with getting calibrated retardance values that made any sense, and found that if you take the difference of two images without moving anything, you see a difference in signal

The laser fluctuations strike again. Ronald had a good idea to take power measurement as a function of time and we see its pretty substantial.

[Jashcraf]

10/30/2024

Today we are going to try and calibrate with Q measurements so that we don't suffer from the intensity fluctuations like we did from the previous week.

Power Settings to try

• $\lambda = 1700nm$

Differential power measurement, Thorlabs 1550nm laser

Switching to the Thorlabs laser helps a lot!

Trying two sequential calibrations on raw intensity

Tried another calibration, and it got a lot of the quantities correct - except for the polarizer angles being off by approx 15 degrees

[Jashcraf]

@wcmelby were you ever limited by the SuperK fluctuations when doing your tests?

[wcmelby]

I was aware that the power fluctuates which can be a significant issue, but I didn't know a good way to monitor and correct for that while also taking measurements. Sometimes if the data looked really weird I would just redo it.

If you're using the Thorlabs broadband lamp, I would just caution that I had issues with several wavelengths, probably due to the filters. It had consistent outliers with retardance measurements around like 1620 nm that I never had an issue with using the SuperK laser.

[Jashcraf]

Ah I should probably go back in and clean up

We actually used a Thorlabs laser that Ronald found - it's pretty steady and quasi-monochromatic, so we didn't have any filter issues. Doing some honest-to-goodness Q data reduction should be fine to get us retardance though. I think if you do the Polar decomposition right you separate the diattenuation and retardance - and it's the Diattenuation vector (top row) that gets messed up when you have intensity fluctuations.

[Jashcraf]

Does it make sense to do an MCMC where we include the source variation?

[Jashcraf]

11/6/2024

Trying again with the Thorlabs laser for now, because we want to try a spatially-varying calibration. That cross-hatched pattern is suspiciously absent!

Power stability looking good

Power in each channel looking very symmetric :)

Here's a spatially-varying air measurement!

Interestingly, it looks like the source fluctuations we see when taking consecutive difference images

It looks to me like we are limited by coherence here - these residuals look speckley!

[Jashcraf]

Went and computed the standard deviation of the mean-subtracted Jones pupil elements. They are shown below in the figure title. I suspect that the artifacts we see are largely due to variations in the source -- all the more reason to try and make corrections based on intensity!

Related, I also wanted to try Q data reduction on the same dataset and normalized it to M11. We should expect zeros from the top row of the Mueller matrix.

The data is noticeably less source-noisy :) but interestingly, there appears to be a bias term remaining which probably corresponds to errors in our estimation.

[wcmelby]

11/8 Today I tried to tried to take a measurement of the laser stability by taking images every 4 minutes for 5 hours. It didn't work (program closed after 30 minutes for some reason) and the data before that was not good. I used the camera at T=-40C, exposure time=2ms, fps=490, and I had the usual screen tearing issue. Here's a few examples of what that looks like:

I'm going to hold off on this task until we figure out the tag pixels or some other way to prevent/toss out the bad images.

[maxwellmb]

Yikes. That's no good. Were you taking one from every 4 minutes, or a burst of frames? Were they all bad?

[wcmelby]

Every 4 minutes, I took a burst of 5 frames (so that I can take the median later). From each burst, some were bad like this and others were fine, it seems to happen randomly

[Jashcraf]

I actually haven't seen the screen tearing before - I wonder if it doesn't happen as much at lower FPS? I've been using an FPS of like 10-100 when trying Mueller polarimetry experiments

[Jashcraf]

11/13/2024

Goal for today is to try and replicate the results above with the SuperK laser, then try a data reduction.

Regrettably, we continue to contend with the power fluctuations in the SuperK

Let's take some data in spite of this and try to normalize the fluctuations in power. In principle, with a dual-channel polarimeter we may get this for free by summing the channels.

Raw data after the Wollaston

Then had a thought that I perhaps should have been doing this entire time. I don't think we need a photodiode to monitor raw intensity - we just get it from the Wollaston prism. After all, Stokes $I$ can be found from

$$I = P{0^{\circ}} + P{90^{\circ}}$$

Where $P_{\theta}$ is a power measurement made through a polarizer at some angle $\theta$. Normalizing the data from the figure above produces a much cleaner measurement of power. Note that there is a Malus' law factor of 2 missing here, but calling it Normalized power is probably sufficient.

The model fit is quite good

Below are the calibrated model parameters. I tend to not agree with the PSG/A Polarizer Angles (PSG/A Pol Angle), but those won't be very as long as they are close enough. I believe they will basically be specified by the achievable null.

I would note that the PSG Pol angle and PSA Pol angle are 82.4 degrees separated, meaning that they are nearly orthogonal as expected. This makes me realize that this entire calibration procedure is rotationally-insensitive, so it will produce the same results for a global rotation of all of the optics. I think we can fix this by asserting the transmission axes of the Wollaston prism

Before asserting Wollaston transmission axes

PSG Pol angle: 69.20108944366167
PSG WVP angle: 57.46694657480928
PSG Ret: 91.16750629765133
PSG Inc: 0.9991693351136111
PSA Pol angle: -13.18298767026058
PSA WVP angle: 25.200648661216114
PSA Ret: 90.79044195477512
PSA Inc: 1.0002421757122035

After asserting Wollaston transmission axes

PSG Pol angle: 82.38342868078286
PSG WVP angle: 70.65055506982013
PSG Ret: 91.16811177751207
PSG Inc: 0.9991596277130934
PSA Pol angle: 0.0
PSA WVP angle: 38.38271705434254
PSA Ret: 90.79029489217537
PSA Inc: 1.0002469458528314

Awesome, all of the starting angles just get rotated by that 13.2 degree offset we saw in the previous data! Does it hold up for the right channel as well?

Model fit for right channel is also good

After asserting Wollaston transmission axes It looks to me that it approximately recovers the same data to within a degree, which is encouraging.

PSG Pol angle: 82.46119328781535
PSG WVP angle: 70.47722578408252
PSG Ret: 90.14513768447357
PSG Inc: 1.0008096596659906
PSA Pol angle: 90.0
PSA WVP angle: 38.40650840863234
PSA Ret: 89.6329510261102
PSA Inc: 0.9996547595162676

Will follow up with Air measurement - I was seeing funky things and I think it's because I have the wrong angular offset or something, need to investigate further.