I'm pretty sure some type of filter or lens would be required in front of the single pixel detector in order to make the measured intensity non uniform over the pixel detectors field of view. A gaussian field of view attenuation filter could be used or else a more complicated holographic patterned filter maybe in order to increase the effective correlation rate by stars crossing ridges in the filter and creating more intensity variance data to analyze! :)
Also perhaps a lock in amplifier stage could be integrated with this filter section.
I had an idea before for measuring the wavefront curvature of light from stars to determine their distance. I was thinking of using a theoretical super flat CCD detector that has a pixel sampling trigger fast enough to capture only the leading or trailing edges of a curved wavefront.
I think the intensity interferometer can do the same thing but is far more sensitive as it does direct optical interference. Has it been used to determine stellar distance based on the lights wavefront curvature, ie far off stars will have less curvature than closer ones, and will thus have a smaller temporal phase offset fed into the intensity interferometer.
You need at least two detectors to get one spatial baseline correlation.
And you need three to get the first good observable closure phase.
Well there is lots of arm waving going on but nothing of substance.
So long as the star remains unresolved at a given baseline it appears to be just as bright like a point source. Once the baseline is sufficient to resolve it then the high frequency components gradually drop off. The spatial frequency where this happens tells you the diameter of the star.
That would be a terrible way to do it at the mercy of a multitude of edge effects and hammered into the ground by airglow. There have been some slightly insane designs for gammaray scopes which involved rotating a venetian blind in front of detectors but it was a disaster and as far as I know never launched.
The only no optical designs that actually work are based on shadow masks constructed by quadratic residues which were used for surveys prior to glancing angle Xray mirrors being developed. It is basically a means to let more light through a pinhole camera. In an ideal one you need a sensor behind it 2x the pixel count linear dimensions of the shadow mask. Wiki article pretty ropey bur closest I can find:
Me too! I have a first edition of his book (was there any other?).
Sound like no reprints then. My copy is dated 1974.
Please tell me more.
The only way I know to get a decent phase observable is to have at least three (or more) detectors and baselines out of which you can then construct one good phase observable. COAST and others have perfected doing full closure phase interferometry in the optical/near IR. Golomb rulers lend themselves to this for small numbers of antennae.
It was sort of a partly-baked idea. The basic notion was to use a 2-D Golomb ruler based on antenna-coupled tunnel junction heterodyne detectors to get lots of Fourier components, but without the extreme amount of circuitry required to do a real synthetic-aperture system.
What I was hoping to do was to take, say, 10,000 ACTJ detectors on a wafer, illuminate with a CO2 laser, and look at the heterodyne output from the thermally-illuminated scene.
You can do that sort of measurement with photodiodes, but the problem is that the photon occupation number of thermal light is so small. That means that even if the source is direct sunlight, your heterodyne signal is really low unless you can maintain a really wide bandwidth. (*) Thousands of high bandwidth front ends were going to be a lot more work than just putting in a lens.
ACTJs are extremely fast at baseband--published results have shown RF outputs above 170 GHz, and that's basically the wiring. So I was hoping to use the baseband fourth-order product to interrogate the noise correlations in a bandwidth of a few terahertz, and make 10**8 low-budget intensity interferometers.
That would get me Fourier components of the autocorrelation of the image, just like the Narrabri telescope, but a zillion times faster due to the massive parallellism.
The problem is that all spatial phase information is lost, and has to be recovered before you can reconstruct an image. It should be possible to apply a variation of the Fienup phase-retrieval algorithm to do that, but convergence would probably be difficult to guarantee.
However, if one were able to use (say) 10 phase-sensitive receivers with adequate bandwidth, one would have 45 Fourier components with the correct phase, which seemingly could be used to seed the phase retrieval algorithm (exact mathematical details fuzzy here), to speed it up and improve its convergence properties.
But as I say, I never had a good enough tunnel junction process to be able to build one.
In my world that technique is called "closure", and it works great. (The 2-D Golomb ruler technique was invented by a pal of mine in grad school, Yoram Bresler, who's been at UIUC for ages now.)
Cheers
Phil Hobbs
(*) A couple of years ago, I collaborated on a very successful scheme to detect plumes of HF from clandestine nuclear enrichment plants, using heterodyne detection of sunlight with a tunable diode laser. The principle is heterodyne detected absorption spectroscopy. Works great, but you really have to vamp for bandwidth.
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
hobbs at electrooptical dot net
http://electrooptical.net
Just to be clear is this the apparent diameter or the actual diameter that is being measured?
The wiki article says apparent diameter, but I could see how it would be actual diameter, if at the edge of a star the distance to earth is increasing due to the star being a sphere, this light will have a phase delay to reach earth, it could be possible to have the two telescopes pointed at the same direction, and then slowly change the angle of one and sweep across the star and measure the rate of phase change to see the radius of the star that way.
(still confused!)
I think it is possible to use a coded aperture, the other message I sent mentioned "gaussian field of view attenuation filter" which is effectively an analog coded aperture which could still be calculated in software I think, but a chess board type coded aperture might be better, but either a continuous pattern or a grid it shouldn't matter as long as the aperture geometry/attenuation is taken into account in the data processing.
Also I think a single pixel detector with the coded aperture can be used to build up an image over time. It basically is substituting the 2D pixel array with a temporal array of pixel data which can then be correlated and reconstructed into a 2D pixel array statistical image with no diffraction limits still since no focusing lens is used. Since the aperture itself is coded and can discriminate light sources based on angles I don't think a 2D pixel array is required.
I think it would only have a chance in space possibly to reduce the noise floor (ie air glow as you mention) as much as possible.
The apparent angular diameter of the object on the sky. Essentially you can detect the transition between unresolved point object where brightness is constant and resolved where the thing looks like a disk and so the highest spatial frequency components are weaker.
Only if the star were a coherent emitter (and they are not).
You do get the OH maser region that does emit coherently.
The design of the right mask is incredibly important if you are to maximise the quality of the data with a given detector.
No. Gamma rays are heavily quantised and energetic. They were all scintillator crystals with a PMT in them and the ventian blind rotated above a second set.
Basically a rough heuristic if you are going to spin a satellite to collect your data is that you must have a good capable 1D imaging configuration to start with to stand any chance of making an image.
Here is a pattern from a mouse pad that might be a good high resolution coded aperture to use for the single pixel telescope assuming it is a totally non-periodic pattern:
formatting link
I think it would require quite a bit more software processing to handle the high resolution aperture pattern, but I think it has the advantage of narrowing down the potential solutions of statistical light source paths across the field of view compared to a simpler aperture.
Also for this statistical application the coded aperture doesn't need to have 100% attenuation on the dark pixels to make it effective for determining light source positions, maybe only 1% attenuation would be enough, since with statistical summing of light over time, a 1% difference should eventually become apparent in the data.
Maybe an aperture that also has variable attenuation on individual aperture pixels could be used as well, but that would probably require even more processing and maybe more even more accumulated data.
A bit of theory for how it could work, assuming a simple simulated 7 star field ie. the brightest 7 stars in the Big Dipper, moving across the aperture, each star will add its light to the sum of light on the single pixel detector based on if it is in the field of view and also depending on the attenuation of the coded aperture pixel the stars light is currently shining through.
For the simple case of one pass of all 7 stars, each star will have a unique attenuation pattern from one side of the aperture to the other side. The hard part for the software is to take the summed light changes over time, and decode all of the possible aperture passes could have caused the light changes, as well as how many stars were involved. These possible solutions would each have a probability associated with them. If the simulated star field is then passed across the aperture at a slightly different angle (ie to simulate the different position of stars in the sky seasonally) then the same
7 stars will generate another set of possible solutions for the number of stars and which paths across the aperture they took.
It isn't hard to imagine that with enough data like this and analysis software that eventually the number of possible solutions could maybe be reduced to the correct 7 stars and the correct paths across the aperture.
If that is true, then the next step would be to increase the sensitivity of the light detection so that real stars light could be detected.
I think there would be basically thousands of terabytes++ of data required to be collected from the single pixel detector though before there would be enough data to correlate, and the processing and memory requirements would be very high too.
Noise filtering in hardware and software would also be critical, lock in amplifier, and software low pass filtering decimation etc.
If the aperture was an LCD then the 1% attenuation pixels could be turned on and off, or the whole pixel array blacked out periodically to do the lock in amplifier function.
Also having an active coded aperture might be a way to increase the effective aperture coded resolution.
Ok I hope that idea is crazy enough for anyone crazy enough to entertain it! :)
Oh ya.. I guess if anything it would be a Doppler frequency effect of the edges of the star having a relative frequency shift based on the star's diameter and rotation rate, but that would be a miniscule effect to try to measure and determine the star's diameter and rotation rate from, unless there is a "frequency interferometer" that can measure super-accurate spectral line differences between light from the edge of the star and from the mid point.
Also another method that would work with bright stars to detect their positions quickly with a single pixel telescope could have a LCD panel that runs a pattern algorithm using feedback from the single pixel detector, and the algorithm would turn pixels on and off to block light and eventually find the solution combining the minimum number of pixels required to be transparent while still having maximum detected star light. Those pixel positions would be the star positions.
The LCD panel could also block all but a single stars light out which would be useful for spectral analysis etc.
Also if the LCD panel was placed far enough away from the single pixel detector, ie in space 100 miles distance between the LCD and the detector, the LCD panel angular size could approach the size of a star that is aimed at viewed from the detector, and the star itself could be selectively resolved over time. It might be a problem to block other light sources and narrow the field of view of the single pixel detector to the stars angular size, maybe a long tube would suffice :D
It was sort of a partly-baked idea. The basic notion was to use a 2-D Golomb ruler based on antenna-coupled tunnel junction heterodyne detectors to get lots of Fourier components, but without the extreme amount of circuitry required to do a real synthetic-aperture system.
What I was hoping to do was to take, say, 10,000 ACTJ detectors on a wafer, illuminate with a CO2 laser, and look at the heterodyne output from the thermally-illuminated scene.
You can do that sort of measurement with photodiodes, but the problem is that the photon occupation number of thermal light is so small. That means that even if the source is direct sunlight, your heterodyne signal is really low unless you can maintain a really wide bandwidth. (*) Thousands of high bandwidth front ends were going to be a lot more work than just putting in a lens.
ACTJs are extremely fast at baseband--published results have shown RF outputs above 170 GHz, and that's basically the wiring. So I was hoping to use the baseband fourth-order product to interrogate the noise correlations in a bandwidth of a few terahertz, and make 10**8 low-budget intensity interferometers.
That would get me Fourier components of the autocorrelation of the image, just like the Narrabri telescope, but a zillion times faster due to the massive parallellism.
The problem is that all spatial phase information is lost, and has to be recovered before you can reconstruct an image. It should be possible to apply a variation of the Fienup phase-retrieval algorithm to do that, but convergence would probably be difficult to guarantee.
However, if one were able to use (say) 10 phase-sensitive receivers with adequate bandwidth, one would have 45 Fourier components with the correct phase, which seemingly could be used to seed the phase retrieval algorithm (exact mathematical details fuzzy here), to speed it up and improve its convergence properties.
But as I say, I never had a good enough tunnel junction process to be able to build one.
In my world that technique is called "closure", and it works great. (The 2-D Golomb ruler technique was invented by a pal of mine in grad school, Yoram Bresler, who's been at UIUC for ages now.)
Cheers
Phil Hobbs
(*) A couple of years ago, I collaborated on a very successful scheme to detect plumes of HF from clandestine nuclear enrichment plants, using heterodyne detection of sunlight with a tunable diode laser. The principle is heterodyne detected absorption spectroscopy. Works great, but you really have to vamp for bandwidth.
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
hobbs at electrooptical dot net
http://electrooptical.net
--- SoupGate-Win32 v1.05
* Origin: (1:249/999)
--- Synchronet 3.15b-Win32 NewsLink 1.92
SpaceSST BBS Usenet Fidonet Gateway
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.