Strobed Imaging

Feb 21, 2022

A Canon camera with full sized 35mm detector and 300 mm lens has FOV of 8 degrees, corner to corner. See modified Canon EOS 5D Mark VI camera. 

Plate scale is (5.536 micron pixel/300mm) = 3.7 srcsec/pixel. So we'd need SNR of 10 or above to get centroid sigma of 0.3 arcsec, and SNR of 100 to get sigma centroid of 30 milliarcsec. 

If we make an array of optical fiber tips (perhaps with Thor labs collimators), assuming that source array is 10 meters away: 

max separation is 10m * sin(8 deg) = 1.4 meters between fiber tips. So for 10 away a 1m x 1m panel is good. 

If we stop down the input aperture to be about an inch, which is f/10 for the lens, then depth of field is large. Looking at f-number dependence of image motion can help determine where wedging is along the optical path. 

What is illuminated spot size for Thor Labs reflective collimator and 100 micron fiber?

This is kinda overkill. All we really need is a panel of LEDs, say a 4 x 4 array, pointed at the Canon camera. We can focus on the array, so de-magnification is focal length divided by object distance, or 10/0.3 = 33. So a 1mm x 1mm LED emitter shows up on the detector as 1/33 mm = 30 micron or 6 pixel across image. That's not so bad. Tip of a 100 micron dia optical fiber is ten times smaller, will be limited by lens MTF. 200 micron core optical fiber maps to 200/33 microns = 6 microns ~ 1 pixel so that's a good match. 

A group of 5 fiber splitters, each going 1:4, will give us an array of 16 synchronously-strobed sources. We can vary both f number and exposure time to explore seeing diagnostics. 

Need to make a plug plate with SMA connectors, with minimal scattered light. SMA threads are 1/4 inch, 36 threads per inch.  

Thor labs HASMA is what we want for this: 

If we drill out and tap a panel with 1/4-36 threads, 1/8" aluminum. we can put nut on back side. 
So also need 1/4-36 nuts and tap. 

McMaster has the tap:

Digikey has 1/4-36 nuts, at $3 ea! Part number is A104989-ND

Bought 30 of the nuts and one tap. 

3.2mm hole for SMA ferrule is 0.126 inches, essentially 1/8 hole. Ferrule is 1/4 inch long. 
So if we drill out a 1/4-inch panel with 1/8 inch holes. ferrule should insert with no extra scattered light. 

Remote-trigger for flash from Canon camera, wireless 

Bought a used Canon 300mm f/4 lens on Amazon for $600. That means biggest aperture is 300/4 ~ 75mm or 3 inches. For point sources 1m apart at 10m distance, when do those beams get to 50% overlap? Ans: 15 inches from the lens. 

We have 500mm mirror lenses for Canon cameras. The FOV is Diagonal: 5° Vertical: 2° 45' Horizontal: 4°. At a distance of 5 meters, edge to edge FOV on horizontal sees 0.35 meters, or about a foot. Displacement of 0.1 arcsec on focal plane is 500mm * 5e-7 = 0.25 microns. 
Closest focus seems to be around 2m. So with 2X Barlow lens, we get 1000mm focal length, f/12 or so, and FOV of 1 foot at 10m distance. How big is optical fiber image on focal plane? De-magnification is focal length divided by distance to object, so 1m/L(m) makes fiber tip image diameter 100 microns x 1/10 ~ 10 microns * (1000mm/FL) * ( 10m/L). We can avoid aperture averaging across input of lens by stopping it down. Need to know differential tilt vs. beam displacement, for r_o of order 10-20 cm. 

Distortion-grid source

1.8mm sized LEDs subtend an angle of 1.8 mrad from 1 meter, and 180 microradians from 10 meters away. That's about 40 arcsec. For an LED with a lens, we only need the forward-beamed part anyway.  

Edgelec LEDs are 1.8mm in diameter, with current-limiting resistor so it drops 12V at 20 mA so that's an effective resistance of 12/0.2 = 60 Ohms each. If 25 of them are operated in parallel, we need 12V at a current of 25*20mA = 500mA or 0.5A. 

Spacing between terminal blocks is 0.375" or 9.5mm. 

Bought laser drilled pinholes on ebay for $10 ea, in 100 micron hole with 3/4 inch OD holder that is 0.6mm thick (qty 30) , and 300 micron in 1 inch diameter holder also 0.6mm thick (qty 10) also 200 micron pinholes in stainless disks that at 3 mils thick, 20.5mm dia. (qty 40) for $6 ea

Plan is to mount the precision apertures into the recesses in the panel, and hold them in place with 4-40 screw and large washer. Then glue or RTV to hold small LED up against the pinhole aperture. Efficiency should be aperture area divided by LED emitting area, or about (0.1mm/1.8mm)^2 ~=  3e-3. 

If LED is about 50% optical efficiency, we get 2V*20mA*0.5 ~ 20mW optical output. include pinhole and we get 60 microWatts of optical emission. At 10m distance and 60mm lens aperture effective diameter and 2pi of solid angle, we catch a fraction (60e-3/10)^2/6 = 6e-6 of the output. 

So energy flux into aperture is around 0.5 nW per source, at 10 M. At 600 nm that's a photon flux (https://www.kmlabs.com/en/wavelength-to-photon-energy-calculator) that should be plenty, even for 1 millisec exposure time.  

12 position terminal is Digikey 283-4021-ND‎. ten position jumpers are WM20055-ND‎. 

Fastners: 

McMaster has black 4-40 large diameter washers, 

Fancy McMaster socket head black with integral (small) washer

LED array assembly steps and parts list

Assembly:

install blankoff disks in all 16 holes, using washers and 4-40 screws to hold them in place

install terminal strips on side opposite to 3/4 inch recesses, qty 2, using 10-24 bolts

gently glue qty 16 LEDs into mounting holes, being careful to not get any epoxy on the front lens. Try to make gentle contact with blankoff plate, and keep LEDs centered and vertical . Keep anode wires on one side and cathodes on the other

wire up "+" side of all LEDs to one terminal strip, "-" to the other, so they are all wired in parallel

install capacitor across terminal strips, use MOSFET for ground interrupt. Run three conductor cable (+12, gnd, on/off) 

run cable to waveform generator 

connect waveform generator to Canon synch output 

Make our own holes w surgical needles? 

MDS precision surgical instruments:

Needles are sold here: https://www.wpiinc.com/var-3184-nanofil-needles 

Diabetes lancets come down to gauge 33, which is 209 microns. https://www.amazon.com/AgaMatrix-Lancets-33-Gauge-Count/dp/B00186SI4U

gauge 32 is 230 microns. Those are readily available. 

Foil options:

Assuming we'll poke through the foil with gauge 33 needle with OD of 210 microns, the Thor Labs foil would give us 4:1 hole aspect ratio. 

Assembly:

Poke holes in Thor Labs foil from 'top' while it's on top of a neoprene backing. Flip it over and use razor blade to cut off hanging chads. Then sandwich the foil between a diffuser on the back and clear glass or plexiglass on the front. 

March 20, 2022. 

Bought Thor Labs foil roll, 32 gauge lancets, neoprene backdrop, frosted and clear acrylic. 

Diffuser-illuminated mask:

Canon lens performance-

DYI telephoto lens

https://sot.com.sg/telephoto-lens/

Works best if diverging lens has half the focal length of the objective, with T  slightly larger than second lens focal length. Focal length is then twice the length. One meter focal length has 5 microns per arcsec plate scale. We can do this with a 500mm focal length objective, -250mm focal length second lens, and about 300mm interlens spacing. Diverging lens needs about half the diameter fo the objective.

Bought 50mm diameter -400mm FL 50mm dia, -2000, -1333 focal length 65mm dia negative meniscus lens, Rolyn optics, from ebay. It wants 800mm focal length, 100mm diameter objective. Bought 900 mm FL, 102mm dia telescope objective from Surplus Shed. So it's running at f/9. 

So a good design point is 620mm setback spacing. Beam diameter is 102mm * (1-620/900)= 31 mm, less than size of diverging lens. Full focal length is 3 meters, gain of 3000/900= 3.33. Length went from 900mm to 1.5m. 

Behind-your-head reflected imager:

Flat mirror should subtend FOV of camera + lens. If we use 500mm mirror lens plus 2X converter, we have 1m focal length. This has the advantage of not needing long distance flash setup- it's local to camera. Plus we get 2x path length and if effect goes as L^3 that's 8X the signal strength. 

FOV for 500mm is 2.5 x 4 degrees, and half that for 1000mm. So 500mm lens with 2X telesconverter has FOV of 1.25 x 2 degrees, or 2.35 deg on diagonal. Total size subtended across diagonal for separation L is L*4.12E-2, or 40 cm *(L/10m). That's only 2 inches. 

Concage mirror version has wider transverse separation so less correlated image motion. 

angle subtended by 1.8mm LED is 1.8mm/20m = 90 microradians or 18 arcsec. Angle subtended by 200 micron fiber is 10 microrad or 2 arcsec. 

Local strobe control

"PC" connector on camera is effectively an open-collector NPN transistor, biased into conduction when camera wants to trigger the flash. 

Sparkfun schematic: Optoisolator-v12.pdf. Ordered 10 of these optoisolators, good to have around.

Possible Fiber spacing/arrangements:

1) circle of diameter D ~ 20 inches, in a single panel. This gives directional information, by looking at elliptical distortions. 

2) 1-d or 2 orthogonal 1-d arrays, log separation

3) rectilinear 2-d grid 4 x 4, with 4 on each panel

4) quadrants, each done with different spacings of 2 x 2 arrays, ranging from 1 inch apart (overlapping

5) Meximize "uv plane coverage", like an interferometer. Biggest diversity in baselines for all N(N-1)/2 = 120 pairwise separations. Do this two ways, assuming and not assuming isotropy. Needs some kind of annealing algorithm. Good job for a CS undergrad!
Define range of desired separations. Start with one maximally close pair, say 1 inch apart. Add next to maximize L1 distance from all existing spacings, that means go to max separation along the same line. For subsequent ones, do a raster search in x and y and compute sum of abs(separation differences) Find one or more locations that maximizes this, and place it there. Iterate until all 16 are placed.  

Wrote a first-cut program for maximum uv plane coverage: annealer.m

If we image a fiber tip with 300mm lens, how big is minified image of fiber tip? For 300mm lens, spot size is 200 microns * (300mm/L). So if it's 3 meters away, spot is only 20 microns. If it's 30m away, spot is 20 microns. For 300mm lens, 1 arcsec is 1.5 microns. So it might be good to use a 2x extende, get 600mm focal length so 1 arcsec is 3 microns. 

Sensitivity function calibration (June 26, 2022 Stubbs)

Consider two emitting optical fibers that are imaged by ~500mm f/8 mirror lens. 

discussion of intersection of two circles:
https://diego.assencio.com/?index=8d6ca3d82151bad815f78addf9b5c1c6 

Wrote a crude beam simulator that takes a Gaussian-tilted wavefront and moves it along this optical path. For the figures below, z=0 is at the sources and z=L=10m at the camera lens. Plots show differential image motion and common mode motions of the source centroids. 

Lower plot shows that index perturbations near the sources are weighted 10X as much as ones near the camera. This is a combination of aperture averaging and beam convergence. 

MATLAB simulation code: PhaseSim.m

Schleiren imaging. 

April 1 2022. 

This (at Tonry's suggestion) seems a promising thing to do.

Bought qty 2 of 203mm diameter spherical mirrors on EBay, 1280mm focal length. That means a plate scale of 1 arcsec being 6 microns. So pinhole source needs to subtend an arcesc at most, 5 microns would be good. Mirror is 1280/203 = f/6 so we'll get some spherical aberration. 

Need a mirror mount that allows tip-tilt adj. Thor labs PY004 will support the mirror mass of about 1.5 kg. Also got Newport model 36 tilt-rotation stages. Now need 8 inch mirror mount

Also got from Surplus Shed 76mm diameter -1250mm FL negative meniscus lens. Can add that to one side of Schleiren system to get bigger plate scale. Mirror is f/6 and so this would be 6*3 = 18 inches upstream of focus, which is 1280 - 18 x 25.4 = 450mm behind the focusing mirror. So optical train would be M1=1280 FL then space of 450mm then -1250mm negative meniscus. That gives

Focal length of 3.8m and total length of 3m. Plate scale is then 18 microns per arcsec. 

How much does collimated beam diverge? Not much, 5 microns per meter 

Radius of curvature of Aux Tel primary is 6605.99 mm:

Aux Tel beam annulus is 0.5m across. The center of curvature of the primary is a distance 6.6x2=13.2 meters away from axis intersection. . Could we do this? Top of met tower is 17.11 meters away, horizontally: 

Secondary-to-primary spacing is 2.7 meters. Baffle OD is 0.637m diameter. From the met tower, the primary subtends an angle of 1.2/17 = 4 degrees. Annulus spans 0.5/1.2 of that, still over a degree. Won't fit on detector. 

Schleiren references

2017-A review of recent developments in schlieren and shadowgraph techniques-Settles-Hargather_Meas._Sci._Technol._28_042001.pdf

1954-Shadow and schlieren methods-Beams-Physical measurements in gas dynamics and combustion.pdf

In astronomical use:

1979MNRAS.188..249L

March 26, 2022 summary

Assume 1000mm or 1300mm focal length for imaging lenses, either Omegon telescope or 500mm lens with teleconverter. One arcsec maps to around 5 microns in focal plane. Claimed resolution of Thor labs lateral effect sensor is submicron, so this is the right regime. 
Canon 5D sensor has 8.2 micron pixels (meaning 8.2 microns contains multicolor Bayer elements, really half that for single-color pixel). FOV of Canon camera with 1000mm lens is 2.5 degrees from corner to corner. 

FOV for 1000mm lens and 35mm sensor, FOV = 0.43 * (L/10m). 

Tried poking 32 gauge holes with needle, into Thor labs foil, worked great. 

Assume source is 20m away, with reflector if need be to place source panel near the camera. That means we can use up to 1m size panel. 

Thor labs Solis LEDs are good potential light source for this. Thor Labs LED controller will drive it with a pulse using external trigger.

Thor labs foil roll is 1 foot across, so making 12 x 24 inch panel seems sensible and manageable. 

Solis LED from Thor labs has a few Watts of output power, sort-of collimated output beam (70mm FWHM at 700mm distance). Output aperture is just under 50mm. Might work best to just remove the collimating optics? 

Edmund optics part number #32-248 is so-so quality (4-6 wave) flat mirror. Each reflected beam spans one half of lens input diameter in size. Am assuming figure errors on that mirror are much larger scale and that image degradation is minimal. 

April 22, 2022 gphoto2 attempt

gphoto2

Conclusions: 

Array of optical fibers (16) driven by a common incoherent source, Thor Labs LED at a few hundred mW. Lens with 30 lines/mm resolution gives around 30 micron spatial resolution. For that to be 2 arcsec requires a focal length of at least 30e-6/L=10E-6 so L=3m. Yuk. 
If we use a Omegon OTA as the front end, we get focal length of 1370mm. 

35mm detector (Canon camera) has FOV of 35mm/1370mm = 1.5 degrees corner-to-corner. 

1:1 image with collimated light for 1 arcsec FWHM implies source size of 5e-6=D/1370mm so D=5E-6*1370mm = 8 microns. 

Or... low divergence diode module with 0.5 mrad divergence and 3mm beam size ends up with outgoing beam divergence of (3/154)*0.5 mrad = 10 urad or 2 arcsec.