Calling it a "really fast camera" elides much of the actual difficulty. We're trying to tag individual wavefronts of light at different telescopes, record them, and then play them back at a central "correlator" with the appropriate delays so that the waves come to a focus.
For a wavelength of 1.3 mm, we'd want the time tagging to be better than a quarter of the wavelength at least - say 0.3 mm. The speed of light is 300 mm/ns (a foot per nanosecond is the shorthand beloved of circuit and chip designers). So, for 0.3 mm, we're going to have to get down to a wavefront tagging accuracy of 0.001 ns.
No clock is going to get there, but if we can get ~close enough, we can use a procedure called fringe fitting to determine the clock corrections by looking at the wavefronts. (Does it line up this way? How about this way? How about now? Yes, it's as laborious as it sounds, but computers, eh.)
This is all in the calibration of data, before we do the Fourier inversion to create images - the magic of radio interferometry is that we can record the signal to disk while preserving phase. Optical photons can not be recorded and played back with phase preserved - optical interferometry has to split up the photon streams and send different parts to be correlated against streams from other telescopes, so you run out of signal quickly. Meanwhile, we can record radio waves at the 27 VLA dishes, say, and play them back for correlation on all 27*26/2 = 354 baselines, no problem. That's why radio VLBI is a thing, but not optical VLBI.
Even as a professional radio astronomer, the underlying physics is deep and almost magical.
> That's why radio VLBI is a thing, but not optical VLBI.
Hi, long baseline optical interferometrist here who specializes in modeling and image reconstruction.
To set the record straight, long baseline optical interferometry really is a thing. At present there are two optical interferometers operating in the USA and one under construction: Georgia State University's Center for High Angular Resolution Astronomy (CHARA), and the Navy Precision Optical Interferometer (NPOI), and New Mexico Tech's Magdalena Ridge Optical Interferometer (MROI, under construction). Europe operates the Very Large Telescope Interferometer (VLTI) in Chile. Australia has the Sydney University Stellar Interferometer (SUSI). Optical interferometers have been around for a really long time. Michelson famously measured the diameter of Betelgeuse in December 1920. The first image from an optical interferometer was of Capella produced by the University of Cambridge's COAST telescope in September 1995.
The key difference between VLBI and optical interferometry is that we must combine the light from each telescope in real time, rather than recording the RF data to disk and forming the interference patterns later using correlation. Our interference patterns are recorded on high speed cameras, extracted, calibrated, and then stored as OIFITS files. These files are then later reconstructed using a variety of methods, including Markov chain processes and regularized maximum entropy.
Except for the CLEAN deconvolution process, the methods used to reconstruct images from the EHT data are identical to what optical interferometry has been doing for decades (see https://iopscience.iop.org/article/10.3847/2041-8213/ab0e85, Section 2.2.2 for references to literature). The maximum entropy process used for optical interferometric image reconstruction was, in turn, developed for MRI image reconstruction.
Don't get me wrong, I am not attempting to trivialize the result of the EHT team. The effort involved is monumental and the result is astonishing. In fact, I suspect my facial expression was very similar to Katie Bouman's now famous photo when I first saw the image. Then my jaw hit the floor when I found that some of my work (Baron, Monnier, Kloppenborg 2010) was cited in their imaging paper! However, my first inspection of the "eht-imaging" and "SMILI" repositories has yet to reveal anything new or novel that is not regularly employed by optical interferometrists.
Because you'd also like to know the phase of the waves. If you get unlucky and get just DC offset of the waves (think of it as the 0 crossing of a sine wave) then you've no idea what the phase is. The peak could be before the 0 crossing and the minimum could be afterwards, or it could be the opposite (pi offset in phase). Granted, this is unlikely to occur, you're much more likely to not run into these scenarios. Ideally, you want 4 pieces of data per wave, the 0 crossings and the max/min, for each wave. From that you can get the amplitude and the phase for certain (again, luck is involved)
Can you not make a coherent quadrature detector? I was under the impression that those were pretty common in optical communication systems. Edit: nevermind, 230GHz, so millimeter wave but mixers and oscillators exist at that frequency, are the noise figures just too high to use?
Luck isn't involved as long as you can ensure your sampling noise is uncorrelated with the incoming wave, either by direct insertion of dither or characterization of environmental noise. Regardless, your point still stands, just a fun tidbit.
> the magic of radio interferometry is that we can record the signal to disk while preserving phase. Optical photons can not be recorded and played back with phase preserved
Why is that for the optical photons, when it’s “deeper” than just a higher frequency as you answered elsewhere?
I can’t speak to any physical limitations that this poster seems to be speaking to (I only studied physics theory so I’m not too sharp on the details of lab / experimental devices / apparatuses) but I would guess intuitively that a fundamental limitation is that radio tends to be more coherent, so you aren’t relying on individual photons but rather aggregating a bunch of photons to measure a wave. In contrast, optical light tends to be incoherent (unless from a laser), so you have to measure individual photons and they aren’t really correlated with each other (so interferometry doesn’t work).
For a wavelength of 1.3 mm, we'd want the time tagging to be better than a quarter of the wavelength at least - say 0.3 mm. The speed of light is 300 mm/ns (a foot per nanosecond is the shorthand beloved of circuit and chip designers). So, for 0.3 mm, we're going to have to get down to a wavefront tagging accuracy of 0.001 ns.
No clock is going to get there, but if we can get ~close enough, we can use a procedure called fringe fitting to determine the clock corrections by looking at the wavefronts. (Does it line up this way? How about this way? How about now? Yes, it's as laborious as it sounds, but computers, eh.)
This is all in the calibration of data, before we do the Fourier inversion to create images - the magic of radio interferometry is that we can record the signal to disk while preserving phase. Optical photons can not be recorded and played back with phase preserved - optical interferometry has to split up the photon streams and send different parts to be correlated against streams from other telescopes, so you run out of signal quickly. Meanwhile, we can record radio waves at the 27 VLA dishes, say, and play them back for correlation on all 27*26/2 = 354 baselines, no problem. That's why radio VLBI is a thing, but not optical VLBI.
Even as a professional radio astronomer, the underlying physics is deep and almost magical.