1) construct absolute pressure monitoring station, using dracal USB pressure sensor

MS5611 data sheet

https://www.dracal.com/store/products/usb_bar20/index.php

2) in-dome temperature measuring system- start with Dracal USB temperature monitors Interested: David

3) infrasound sensors for Chile and Cambridge. 

4) contrail monitor/predictor system - resurrect ADS-B receiver, analysis of all-sky camera daytime images, exploitation of Antofagasta balloons

5) RFI monitor system - broadband antennas, logarithmic peak detector, low pass filter/peak hold, datalogger:  Interested: David

. 

RF signal combiner

RFI_board.pdf

6) dual band retroreflector attenuation monitor system- Dual LEDs at 470 and 850 nm broadcast to retroreflector at Gemini, measure dust attenuation, differentially. Frequency domain multiplexing. 

Adj frequency modules driving Thor labs black bricks, then a pair of fiber coupled diodes that span factor of two in wavelength. Typical Angstrom coeff is around 0.28, so expect ratio of (405/850)^0.28 = > 23% more attenuation at 405 than 850. Make differential measurement between broadcast and received ratios. Use splice-fiber launch combiner, through sm05 Thor labs collimator to beam expander aligned with receive aperture. Use 6-8 inch receiver plus eyepiece, expect around 7mm exit pupil. Really wants an S2281 Hamamatsu diode as receiver, to accommodate exit beam size. 

Beam divergence for 200 micron fiber is 200E-6/FL where FL is focal length. beam size at Gemini (distance of ~1km) is D= (200E-6)*(1000m/FL) = 0.2m *(1m / FL ). 
Longest focal length Thor Labs reflective collimator is 50mm. That gives D=4 meters if perfectly focused. Going through a Thor Labs reflective beam expander requires 3mm dia input beam. We're better off with an off-axis parabolic mirror. Edmund f/2 100 mm focal length would give us a 2m spot diameter at Gemini. Pointing accuracy would have to be 1m/1000 m = 1 mrad or a few arcmin.

Interested in working: Brodi 

Celestron telescope with manual eq. mount, use tiltmeter and rotary potentiometer for elevation and azimuth measurements. 

Celestron launch telescope

adjustable freq driver boards, robust against power failure

timer board, Amazon

Hilarious video documentation of timer board

measure elevation with 0-5V tiltmeter, single axis tiltmeter , tiltmeter manual. At 5V/180 deg we get 27 mV per degree, so need microvolt sensitivity. 

7) particulate sensor system, using SPS30 particulate sensor plus Dracal convert-to-USB unit. Interested:David

Particulate sensor, amazon link

IIC to USB adapter for particulate sensor

8) Development of axially symmetrical differential image motion monitor. RINGSS2020.pdf

9) overall in-dome monitoring system architecture, to incorporate atmospheric pressure sensor, distributed temperature sensors, dust/particulate sensor, aerosol attenuation, etc. 

Apple mac mini as wireless hot spot

directional wifi antennas

Minix Ubuntu computers

Project uses influxDB and Chronograf to store and monitor time series data, see for example https://www.influxdata.com/time-series-platform/chronograf/

We can bundle multiple USB devices onto a single cheap Ubuntu computer: (pressure, RH, temperature) unit plus particulate sensor.  

10) inline fiber-in, fiber-out combination of filter flipper, shutter, and chopper wheel. Parts list includes  -Interested: David
Thor labs MFF101 filter-flipper, beam center is 3.63 inches above base, needs 0.47 inch vertical spacer
Thor labs SH1 shutter, beam center is 3.80 inches above base, needs 0.3 inch vertical spacer
Thor labs chopper wheel- motor axis is 2.92 inches above base, wheel diameter is 4.00 in and midpoint in slit is 1.18 inches from motor axle so beam should be 4.1 inches from base. 
Pair of reflective SMA fiber optic collimators, larger beam size for output one. 

MC1F2-AutoCADPDF.pdf

MC2000B-ChopperHeadAutoCADPDF.pdf

SH1-AutoCADPDF.pdf

MFF101-AutoCADPDF.pdf

11) Aux Tel collimation tool- use a combination of in-focus ellipticity measurements ( suspeceptible to telescope tracking errors) and out of focus donut centration ( pupil mask correction) to center the secondary. 
map out vector field of centration offsets and/or ellipticity in 2-d vs. secondary x,y and find intersection of lines from unit vectors using, for example

https://stackoverflow.com/questions/52088966/nearest-intersection-point-to-many-lines-in-python

12) sinusoidal intensity modulator

13) acoustic in-dome thermal monitor. In-dome seeing arises from temperature field fluctuations and structure. The acoustic travel time across the dome depends on line-integrated temperature and vector wind flow pattern. 
For a 10 meter path length, if we can measure 100 microseconds of arrival difference, what temperature difference does that correspond to? Wikipedia says

The time to travel 10m is dt=10m/330 m/s = 1/33 sec = 30 msec. With 100 microsec resolution we can measure fractional change in sound speed of 3E-3. 

Change is dc/dT=0.58 m/s per degree C. That's a fractional change of 1.8E-3 per degree C. 

Imagine an acoustic phase monitor that looks at a carrier's arrival phase to determine travel time. 40-42 kHz is a typical resonant freq for ultrasonic emitters and receivers. That's an acoustic wavelength of
330m/4e4 = 8.3mm. Say we have a 1 mK delta-T over 10m path length. The fractional sound speed changes by dc/c=0.5*dT/T = 0.5*1E-3/330 = 1 ppm. This is the same as fractional change in travel time. For a 10m path length the travel time is 3 msec so dT=1 mK induces a variation of 3 nsec in arrival time. How much phase change is that, at 42 kHz? dphi=2*pi*dt/T = 2*pi*3e-9*4e4 = 7.6 mrad of phase variation.  

Attenuation in air at 40 kHz is (http://www.sengpielaudio.com/calculator-air.htm) is 

See https://www.piezodata.com/200-khz-transducer/ for high freq resonant transducers. 

Bought four 200 kHz transducers, SMATR200H19XDA, from Steminc. about $40 ea. Impedance is evidently around 180 ohms. 220 pF capacitance. 500V p-p at 2@ duty cycle. 
I have a TS250-0 waveform amplifier from Accel, drives +-10V at up to 450 kHz at 6A rms, so it wants a load of a couple of ohms, and a SoundLazer card with a xmit array. 

SMATR200H19XDA transducer has C=220 pF and minimum of 180 Ohms parallel resistance. 

Bought Ultramic384 from dodotronics, ultrasonic microphone that goes up to around 190 kHz. 

From "microthermal measurements at Karasu" Sasaki et al 2008. These data imply an outdoors RMS temperature variation ranging from 0.3 to 0.003 K. 

Round trip travel time depends on temperature but not wind. Also, for acoustic propagation the Fresnel length is much larger. For L=1m and ultrasonic frequencies, Fresnel length is sqrt(2e-3*1)= 4mm*sqrt(L).
Options:

1) transponder with active return 
2) retroreflector- size should be 4mm*sqrt(L), but many wavelengths in diameter. For 10m path we want reflector of order 12mm in diameter.  

Bought from Steminc piezo: 
SMATR175D28PK

SMATR400HPKLR

SMATR200H19XDA

SMATR40H16XTLR

Kemo M043N power amplifier for low impedance speaker. 

Have 300W tweeter speaker with 25 kHz response

Kemo L010 tweeter with 60 kHz response, much lower power- 16V p-p into 4 ohms is about 30W

Tried an experiment- driving one of these speakers from Kemo amp, using laptop output. 

Using cardboard 3 inch on a side corner reflector about 2 m round trip path length

lessons learned:
- need to avoid multipath and saturation on transmission, with baffled emitter and no mechanical coupling to the microphone
- need to also avoid multipath on receive side
 - note that reflected pulses have opposite polarity, need to account for that for measuring time delays

sample data, 384 kHz data rate and 16 bits. Sent a sequence of transmit pulses like this:

Pulse width is about 2 usec. 

Waveform looks like

roughly 7 msec separation between transmit and return signal corresponds to 2.3m round trip length. Detail of transmit and receive edges:

transmit edge:

echo edge, note parity flip:

At 384 kHz rate, time samples are 2.6 usec apart. The most reliable part of this is the leading edge. Seems pretty easy to get 2.6 usec time resolution, which is 2.6e-6/7e-3 = 4e-4 fractional travel time resolution. At 5.3 m/s per degree per meter, or a fractional change in sound speed of 16 e-4 per degree per meter of path length. Travel time changes by about 45 usec per meter per K. Over a 2m path length, it's roughly 100 usec per degree. So if we can resolve travel times to 2 usec, that corresponds to 20 mK. Not bad! 

These folks look good, for making a transponder as well as broadband pulse acoustics:

https://microacoustic.com/index.htm

BAT-1 system looks most appropriate. asked for quote on:

qty 4 BAT-1
qty 2 Q-amp
qty 2 V-pole
qty 2 of cables, adapters, as needed.

June 29, 2021

OK, there is clearly ringing in speakers when sent a pulse. Why not compensate for that by sending initial pulse concatenated with -1*ringdown? Tried that with good initial results!

anti-ringdown spliced on at x~400:

resulting digitized audio:

That works well! Can do compensated-launch waveform with arbitrary speaker, use BAT-1 for high fidelity transponder, and get clean peaks. Cool. 

Or.. just measure directly the thermal fluctuations with thermistors. 
Reference on balloon-borne Cn^2 measurements: https://play.google.com/books/reader?id=aMnzK6oZVdYC&hl=en&pg=GBS.PA4 

sensor and zero temp coeff resistor from Digikey:
Digikey_sensor_resistor_invoice.pdf

Thermometrics-NTC-TypeFP07 spec sheet

Measurement principle would be to apply sinusoidal excitation to a Wheatstone bridge circuit, and measure temperature difference between spatially separated sensors. maybe even shield one from the wind? Or not..
Both conductive and convective heat transfer from self-heating affect the measurement. This can be experimentally determined by looking at temperature change induced by bias current. Power dissipated is P=V^2/R where R=10 Kohms at room temperature  and V is excitation voltage, with dR/dT *(1/R) ~ - 4% per degree. For 1 Volt excitation, power dissipation is V^2/R = 100 microWatts.   
Thermal time constant can be determined by applying step function in current and measuring how the voltage changes, thermal time constant Tau_thermal determines temporal response of sensor. This acts as a low pass filter on temperature power spectrum sensitivity, and we want bridge excitation freq to be above that. The electrical time constant will depend on capacitive loading of bridge circuit. If there's 1 uF of capacitance, Tau_electrical would be of order 1E-6*1E4 ~ 1E-2 sec. Shielded twisted pair wire is about 50 pF per m so capacitance should be way lower. 

Why not make measurements at varying voltage drive levels and extrapolate to zero self-heating? Why don't people do that? 

Thermistor time constant in still air is claimed to be 0.1 sec, so it acts like a low pass filter with that time constant, for temperature fluctuations.  

Another idea- driven-air increases thermal coupling between sensor and ambient air. Setting aside the heating of the air by fan, driving laminar flow across it will shorten the thermal time constant. Here's a snip from AD590 spec sheet: 

So the thing to do is measure response time constant vs. airflow rate, with self-heating. Rise time indicates airflow, asymptotic value indicates temperature. 

typical values of C_T^2 are 10^-3, so that implies variance for 1m separation of 10^-3 K^2 so rms of tens of mK. For a 10V reference and 10KOhm 4% per degree thermistor, what voltage is that? 

resistance change is (10K)*(4e-2)*(0.01K)= 4 ohms. Voltage change is 1 mV. So we need voltage noise rms of 10 microvolts or less, in bandwidth of 100 Hz. 

Convective power flow seems to scale as P~Delta-T*v^(0.45), or roughly sqrt(v). That does indicate we need to be in the small-self-heating regime in order to not be confounded by convective effects. 
Seems hard to measure both, reliably.  

June 25 2021.  Initial thermistor tests:

wired up 9V battery and 10K thermistor from Digikey, as Wheatstone bridge. "Dissipation constant" for these thermistors is 0.050 mW/C. So only 50 uW generates 1C of self-heating! At 9V bias how much power is dissipated? P=V^2/R = (4.5V)^2/1e4 = 2mW. Note this is close to max power rating of 6mW! So we need way less power into the thermistor. It's acting as much as an anemometer as a temperature sensor. Higher resistance thermistors would help. Ordered 150KOhms at 20C thermistors to cut this down to 100 uW at 4.5V across the device. 

Note can use second harmonic content of response under AC drive to see self-heating? maybe a combination of dc and ac drive, take fundamental to second harmonic ratio?

Plot of wheatstone bridge output for enclosed thermistor:

Plot of wheatstone bridge output with thermistor in air:

This is a combination of air currents changing self-heating equilibrium, and ambient temperature variations. We need to isolate these two. 

Imagine we seek temperature sensitivity in the mK regime. This means we need self-heating to raise the temperature by no more than 10 mK, which implies I^2R dissipation so that dT=P*(1C/50 uW) < 0.01C so that
P < 0.5 uW. At 150KOhms that means V =  sqrt(5E-7*150e3) = 0.25 V. 

Beam-tilt monitor. 

Can use position-sensitive photodiode with a lens to deduce wavefront tilt. If we're looking for 0.1 arcsec tilt that's 5e-7 radians. With a 1m focal length that's a spot displacement of 0.5 microns. 

3 ways to get dome seeing characteristics:

1) differential temperature measurements, measure dT_rms in some bandwidth. Can actually do this with even a single sensor, look at autocorrelation statistics. 
2) air turbulence with acoustic anemometers, take power spectrum of travel time differences. Commercial unit claims 0.01m/s sensitivity, so 
3) scintillation and differential image motion with an optical system. CMOS camera we bought has 60 degree FOV with 6mm focal length lens. Presumably 8 bits of A/D so best for centroids not scint. Could run defocused for scint. however. Pixels are 2 microns .  Size is 7.18 x 5.32 mm, diagonal is 8.93mm. For more dynamic range we could use one of the Canon cameras, bigger sensor and 12 bit A/D. We could stop down the aperture to get rally small pencil of light. Could use stable halogen bulb and use focal plane for chromatic discrimination, out of focus to avoid pixel Moire pattern effects. Short exposure time would work, don't need fast frame rate. 

use a bar with an array of 7 optical fibers all driven from common, stabilized Halogen source (thor labs one, good to better than ppt). Compare scintillation as seen from inside and outside dome, and on Aux Tel. (Needs Hartmann screen in front of Aux Tel.) How far apart for 1 arcmin on focal plane? 1 arcmin is 3e-4 rad so if it's 10 m away, spacing is 3mm. 

But if we use a 10 micron pinhole at the filter position, effective aperture at pupil is about (17m/0.2m)*10 microns  =  85 microns and so spots should be really small! But we'd need strip chart readout or (hah!) shutter at the source! 

LocalSeeing.pdf

Canon flash closes a switch to trigger external flash. Maximum applied voltage of about 5V seems OK. 

We can use an RF remote flash controller to trigger external LED pulse, like this:

If LED forward voltage is about 1.2 V, and MOSFET is low impedance, then about 4 Volts drops across the LED series resistor. 

Or just send the signal that goes to MOSFET gate directly to Thor Labs fancy LED controller. Need to measure how long Canon camera keeps switch closed for shutter trigger. 

Proof-of-principle test:
Canon 6D Mark II camera can use EOS to T-thread adapter, then Thor Labs TMA2 adapter to accommodate SM1 threads. Add pinhole to make pinhole camera. 

Assume 40mm from pinhole to focal plane, so geometrical image is roughly same size as pinhole. Best-image is with 150 micron pinhole if spacing is 40mm:

plot of spot size vs. pinhole diameter:

For a 150 micron pinhole a distance of 40mm away, source is unresolved for D/L  < 150 microns/40mm, or D < 4 cm (L/10m). So maglites would work fine. 

Autocorrelation comparison of atmospheric scintillation from 

Effects of time averaging on optical scintillation in a ground-to-satellite atmospheric propagation, Toyoshima & Araki, Applied Optics 39, 1911 (2000)

from N. Perlot, D. Fritzsche, Aperture-Averaging Theory and Measurements, Proc SPIE 5338, 2004. 

Canon 6D Mark II has 5.73 micron pixels. 

at f/8 the diffraction limit is about 4 microns. Say we had 1000mm focal length. One arcsec of beam deviation gives 4.8 microns of focal plane displacement, that's a full PSF. 

So to see image motion with Canon camera, we need a long focal length. The 500mm mirror lens might be good enough, but it's unlikely to make diffraction-limited images. Better to use a 6 inch RC telescope. 

What about the scintillation idea? The 50mm lens from Zeiss can focus at 0.5m. That gives a FOV of around 0.35 meters in horizontal direction. For simplicity, assume a diffusive screen 250mm x 250mm. 

How far way must laser diode be to fill that screen? L*5E-4=0.25, so L=0.25/5E-4 =  500m. 

For photographic diffusive screens, one step in f-stop progression is a factor of two in  flux. Standard sequence is 1, 1.4, 2, 2.8, 4, 5.6, 8, 11, 16, 22...

We can use fast steering mirror from Newport with strobed source, and Canon-triggered LED using SRS DG645 delay generator. 

Beam-diverger for few-mm wide beam: beam divergence is beam width divided by effective focal length. If beam size is a few mm, a 1m focal length gives few mrad of divergence. 

can use two lenses of equal focal length to tune spot size, separation determines effective focal length. 

So if f1=-f2, the first two terms on the right cancel, and f=f1f2/d

Say f1=-f2=100mm. 

parts for adjustable-beam-divergence unit:

----------------------------

Clay telescope on science center:

https://sites.fas.harvard.edu/~astrolab/claytelescope.html

16 inch DFM telescope

imager is 2048 x 2048 at 0.82 arcsec/pixel, 13.5 micron pixels. So focal length is 0.82 x 4.86E-6 = 13.5e-6/FL ; FL=3.38 meters. Diameter is 16 inches or 0.4 meters so f/# is 3.38/0.4 = f/8.45.

Illumination with 6 inch beam will be somewhat vignetted by pupil, but I don't think that matters. 

Beam deviation is roughly half the window wedge angle:

 

Assume we use 2 inch apertures, that has diffraction limit of about 2 arcsec, will have to do. 

SM2 lens tubes are 2.20 OD so 2.25 clearance will be tight, 2.30 probably better. That means a triangle with 2.30 sides. SM2 threads are 2.035-40 plate thickness of 0.125 is fine

2 inch collimation tester should be fine, need SMA coupled laser.  

Vixen rail system for Omegon RC telescope. - 44mm Celestron/Vixen = CG5 1.75" plate is a standard. 

ThorlabsProjectorParts.pdf

Dovetail adapter for mounting telescope:

Bottom width of 1.75" would be good. 

TopEndOmegon154.pdf

DovetailMountBar.pdf

TopEndOmegon154Bracket.pdf

Schleiren imaging:

precision tilting rotary table:

https://www.travers.com/precision-tilting-rotary-table/p/65-800-210/?utm_term=&utm_campaign=GSN+-+Items+-+Machine+Tool+Accessories&utm_source=adwords&utm_medium=ppc&hsa_acc=5580074735&hsa_cam=8659863797&hsa_grp=92538857168&hsa_ad=407073320798&hsa_src=g&hsa_tgt=pla-915098267277&hsa_kw=&hsa_mt=&hsa_net=adwords&hsa_ver=3&gclid=Cj0KCQjw5PGFBhC2ARIsAIFIMNdbk9DKLCc1377sWlT_0h5k5CIA9Tn6AbT0pXlhRslj8LEIsQzOLCcaAuy0EALw_wcB

manual pan head

https://www.amazon.com/Proaim-Professional-Cameras-Carrying-P-OGR-H/dp/B01M0U26D4/ref=cm_cr_arp_d_product_top?ie=UTF8

camera wedge plate:

https://www.proaim.com/products/proaim-big-multi-angle-levelling-wedge-plate

30 arcmin wedge implies around 15 arcmin beam deviation, just about right for Clay telescope. Can always use 2 if need be, to reduce deviation. 15 arcmin is 4.4 mrad of beam divergence. 

If beam goes 10m then separation increases by 4cm or about 2 inches. We should be able to use 10 inch flat to redirect the beam if that's a useful way to do this. 

Three apertures of 2" diameter should make ~2 arcsec diffraction limited spots. 

Parts list for strobed imager 

SONY fast frame rate camera , 960 fps

https://www.wimarys.com/sony-rx10-manual/

Ken Rockwell manual https://www.kenrockwell.com/sony/rx10-iv-users-guide.htm 

----------------------

DFM telescope on Science Center. 

0.4m diameter primary
plate scale is 0.82 arcsec per 13.5 micron pixel. That implies a focal length of 13.5E-6/FL = 0.82*4.86E-6, or FL=13.5/(0.82*4.86) = 3.38m, or f/# = 3.38/0.4 = f/8.5. 

Back focal length is 4.35 inches from back flange. 

Back flange bolt circle, 3/8-24 bolts (12) on 6,8,10,12 inch dia bolt circles. 

Dispersed projector. Oct 5, 2021, CWS

One interesting thing to consider is sending dispersed but collimated light into the input of a telescope, with a beam divergence that has the wavelengths of interest span the detector. This would allow a filter-in vs. filter-out comparison, allowing for a way to monitor filter transmission in situ. 
A nice calculator for prisms is https://lightmachinery.com/optical-design-center/more-optical-design-tools/prism-designer/

For AuxTel we don't need to do this, a broadband collimated source will do. For LSST with roughly 10m focal length and a single detector of 4K x 4K with 10 micron pixels, a single chip spans 4K * 0.2 arcesec = 800 arcsec = 13 arcmin. This is about 0.21 degrees

So we want the beam separation between 350 and 1100 nm to be about 0.21 degrees. For a grating (note second order problems) we'd want a deflection of 0.3 degrees at 1100 nm. Grating for that is 7.2 lines/mm. 

A prism that accomplishes this (for BK7) is this:

So prism wants a wedge angle of 3.5 degrees for 10 m focal length and 40mm wide sensor. 
For STAR DICE system guess at focal length of 0.8m * 8 = 6 m or so. Our CBP has focal length of about 1m. That wants wedge angle of 26 degrees: 

So we can use the Risley prism trick with a pair of wedges with 20 or so degrees of angle

Thor labs wedge prisms:

https://www.thorlabs.com/newgrouppage9.cfm?objectgroup_id=147

What kind of spectral resolution will this provide? A monochromatic source of diameter D produces a beam divergence of D/FL where FL is focal length. Say we use 200 um core multimode fiber. What's beam divergence vs. focal length? 

theta=41 arcsec * (1m/FL). 

Optosigma makes wedged prisms of the kind we need for this, up to 100mm dia

https://www.optosigma.com/us_en/optics/windows-substrates/wedged-substrates/wedged-substrates-WSB%20-%20WSSQ%20-%20WSSQK.html 

Cheap forestry wedge prism notation: BAF 10 is 104.18 arcmin or 1.74 degrees. BAF 20 is 147 arcmin or just under 2 degrees.