I don't recall much about the specifics of OCT ATM.
However you mentioned a broadband 'non coherent' collimated source and a beam splitter, and the source being coherent over 'some distance'.
Those are probably all relatively key qualifiers.
Any source that is sufficiently collimated and looked at over short periods (e.g. single photons) is pretty self-coherent. At any moment of time some individual atom of the source is emitting a given wave-packet of a photon or a few photons in a given direction. You can beam split that and interfere that wave packet with itself and it'll be fairly self-coherent over a coherence length that is a reasonably small fraction of the distance corresponding to the duration of its wave-packet photon pulse.
The overall bandwidth of the source will be typically much wider than the actual bandwidth of any indivudual wave-packet it emits, though those will have some bandwidth to them as well.
There will be low coherence between successive wave packets, so over a time and over any appreciable distance the source will be non-coherent. Distinct wave packets emitted in different directions will generally be non-coherent.
There's nothing really stopping one from taking an incoherent source, passing it through a collimator, narrowband interference filter (maybe 2nm or less), polarizer, beam splitter, and getting a very low intensity beam that's pretty coherent, probably as much coherent or moreso than many general purpose diode lasers which aren't designed to have long coherence lengths.
It really just depends on looking at classical optics transitioning to quantum optics on a continuous basis, and coherence, polarization, wavelength, et. al. as not absolute boolean sorts of things so much as statistical and analog sorts of things that may be partial / fuzzy / statistical in nature.
Look at the COAST synthetic optical aperture telescope project, for instance, they're using the partial coherence of starlight received by many small mirrors over hundreds of meters of distance to interfere and generate imaging information about the source. Obviously (being distant starlight) the source is broadband and incoherent in origin, but extremely well collimated and really coming in just a photon wave-packet at a time over microsecond type time-scales.
Using the partial coherence and interference of received starlight vs. interferometer arm length for interferometry to measure stellar source 'size' is also a common technique for many many decades.
Confocal microscopy does (seemingly) a similar type of depth-resolved scan using more classical optical design. If you analyze the lenses as creating a reinforcing image over a certain range of wavefront depths based on the wavefront coherence of the source and reflected beams you'll see that it's about the same thing.
Originally posted by: Stiganator
I think I'm getting a bit confuzzled with my optics course.
Is this right?
For OCT to work there must be coherence over some length (the constructive interference) and everywhere else you get some amount of destructive interference). The source itself must be incoherent and collimated. It helps to have a broad bandwidth, since increased frequencies allows for increased temporal resolution.
Two different bandwidths can't be completely coherent, right? Because if they are oscillating at different rates, they will rapidly become incoherent and then return to one point of coherence etc etc. ?
Facebook
Twitter