Before collimating, the beam will have diverged a bit to a radius of 5cm, the range of the wavelengths is 1400-1600nm and I don't want to lose more than 30% of initial power.

I'm asking in general how to collimate it, but my idea was use a parabolic mirror, then focus all wavelengths into a very short optical fiber, then get the light back into vacuum. Would this work? Is there a better way?

  • 1
    $\begingroup$ What do you define by 'collimate' in this situation? You can't have a freely propagating beam remain the same radius - see Gaussian modes. $\endgroup$
    – Jon Custer
    Mar 24, 2022 at 21:51
  • $\begingroup$ I was thinking about making the beam straighter for a certain distance. So not perfect collimation. $\endgroup$
    – Toad
    Mar 24, 2022 at 21:56
  • 2
    $\begingroup$ OK - now, why put it through a fiber? Have you consider the divergence from exiting the fiber? Anyway, seems mostly a question of optical engineering to me, since you seem to have a specific application in mind likely with specific performance needed? $\endgroup$
    – Jon Custer
    Mar 24, 2022 at 21:57
  • $\begingroup$ I thought if you put it through a fiber all wavelengths divergence will become equal at the exit of the fiber. One of the things I don't understand as you mention is how the y beam will diverge at the exit and thought the divergence would be low. Does it obey $\theta = \frac{\lambda}{\pi d_{0}}$ where d0 would be the radius of the fiber? $\endgroup$
    – Toad
    Mar 24, 2022 at 22:05
  • 1
    $\begingroup$ It gets complicated by what the modes are for each wavelength in the fiber and how long each mode travels in the fiber. There are a variety of modeling options out there to determine what the fiber may do to help/hinder your situation, although I've lost touch with what the 'best' ones are these days. $\endgroup$
    – Jon Custer
    Mar 24, 2022 at 22:08

2 Answers 2


Good lasers have Gaussian beams. Gaussian beams are never perfectly collimated, but larger diameter beams have a small beam divergence. To get a larger diameter beam, use a beam expander.

  • $\begingroup$ Do you know if the beam expander would have issues expanding 100 wavelengths? I thought these would all focus differently due to different wavelengths having different optical paths through lenses. $\endgroup$
    – Toad
    Mar 24, 2022 at 22:43
  • $\begingroup$ Many questions about optical components and how to best use them can be answered by looking through the RP Photonics Encylopedia or the related Buyer's Guide $\endgroup$
    – mmesser314
    Mar 25, 2022 at 14:01
  • $\begingroup$ Well, I might say "a laser designed for quality rather than quantity" . In my wild youth :-) I worked on a project that used a corner reflector to ensure the laser (pulsed) stayed internally aligned. The output had lots of power but with a hexagonal array of intensity blobs $\endgroup$ Mar 25, 2022 at 14:21
  • $\begingroup$ @Toad now you are asking about the spectral dispersion of the lens. There are many ways to alleviate that. $\endgroup$ Mar 25, 2022 at 14:22

The answer kind of depends on the details of your light source.

For wavelength division multiplexing your 100 different wavelengths could each be produced by an individual laser. They could then be sent through a single mode optical fiber and then your could collect the light coming out of the optical fiber. That ligh does each wavelength will be fairly nice almost Gaussian spot that will diverge. You could put a collimating lens where the spot is 5 mm and get a collimated beam. It would act pretty much like a Gaussian beam.

There are some details, each wavelength will travel at a slightly different speed through the fiber due to waveguide dispersion and material dispersion.

Each wavelength will have a little bit of chromatic dispersion due to the lens, unless you choose to use a special lens.

But in the big picture, if you start out with a nice beam, you can usually use a lens or a set of lenses to collimate the beam, and each wavelength can be treated as a slightly different problem.

If instead you have incoherent light, rather than having nice electromagnetic modes to propagate, it is a little different. For example you wouldn’t get much light into the single mode optical fiber. But you could do something like put two apertures separated by a distance and have a roughly collimated light. The diffraction angle will depend slightly on the wavelength, but could still put a lens or set of lenses to expand or focus the light or to collimate the light. Same as before the chromatic aberration of the lenses and type of lenses can matter some, but big picture it would work.

  • $\begingroup$ What I was most concerned about is chromatic dispersion as you mentioned. Is there a way to solve it for 100 wavelengths? Or will the error produced by chromatic aberration be negligible and can therefore be ignored? $\endgroup$
    – Toad
    Mar 25, 2022 at 8:36
  • $\begingroup$ It depends on what you are trying to do and your requirements you can use an acromat lens and have very little chromatic dispersion from the lens, but the different wavelengths will still diverge over distance at different rates. For short distances it might not matter. $\endgroup$
    – UVphoton
    Mar 25, 2022 at 11:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.