On the Astrosurf website here, someone has designed a deceptively simple echelle spectrograph for viewing solar spectrum. Briefly, light from an optical fiber falls on an echelle grating and which is cross dispersed with a prism.

amateur spectrograph

In theory, an echellogram has a range of wavelengths (free spectral range) in various orders as follows:


I tried to make a crude version of this (without an optical fiber or a collimating lens) for educational purposes. enter image description here

With a white LED (from an iphone or simple LED flashlight, or a LED bulb) shining on the echelle at a proper angle, one can generate a pattern as follows, which looks impressive and looks like a true echellogram. Unfortunately, this success is rather short lived. The moment we use a fluorescent lamp with a slit infront, we just get repeating images, which shows that this is not an proper echellogram! Does anyone know why this is happening? A senior spectroscopist has also tried it and no matter what orientation of slit or blaze direction (against or towards), grating angle, grating orientation... the result is just a repeating image.

Questions: a) Is it just because there is no collimator in our crude version? Why is an LED working and line source not? I tried collimation with a lens, but there was no improvement (repeating lines were still there).

b) If this were a proper echellogram of a line source, such as a fluorescent bulb, one should see one purple line in first row (higher orders), two green lines, two yellow lines (Hg doublets) and a few red lines. I cannot understand why there is a repetition of the same wavelength say purple line or the two green lines?


enter image description here enter image description here

  • 2
    $\begingroup$ @uhoh, I have tried almost all angles and inclinations and so has my senior spectroscopist friend (independently). It has been a couple of years of this "side" educational project so you can imagine all permutations and combinations of angles and orientations. This repeating image does not go away. I have a feeling that this is just a collimation issue that parallel rays are not falling on the echelle. I don't have a collimator and a simple lens in not working. $\endgroup$
    – AChem
    Sep 3, 2020 at 12:33
  • 1
    $\begingroup$ Did you find an arrow as mentioned here? Did you try from the opposite side? The "staircase" profile only works from one direction, as shown in images in this answer and here and here. BTW you really must use a slit or a cheap multimode fiber bundle and probably you'll need to also collimate. Optics is challenging and it usually fails if things are done in a less-than-correct way. $\endgroup$
    – uhoh
    Sep 3, 2020 at 12:49
  • 1
    $\begingroup$ @Uhoh, I have also tried a slit. Yes, the grating has a pencil arrow. If you stand a night far away from a mercury lamp, you can see that if the angle of holding is right, the echelle grating becomes suddenly very bright with a beautiful spectrum. Holding a transmission grating crossed with it gives the same multiple images.When you are the "right" angle of incidence and diffraction, the diffraction pattern is markedly bright. If you use it in the "wrong" light and holding direction the image is not bright at all. $\endgroup$
    – AChem
    Sep 3, 2020 at 13:11
  • 1
    $\begingroup$ @Uhoh, If you have an echelle, you can also try. Worth the demo. I think we need to try a point source and a lens in front to collimate it. $\endgroup$
    – AChem
    Sep 3, 2020 at 13:12
  • 2
    $\begingroup$ Oh, Thorlabs has both "cage mount" and "tube mount" systems that work very nicely and come in different sizes! You can get fiber connectors and lens mounts that fit in either one thorlabs.com/newgrouppage9.cfm?objectgroup_id=69 This would allow you to hold the lens and fiber at a fixed or an adjustable distance. If you can get a physical paper catalog and look through it (it's hundreds of pages) you can see a wide variety of different mounting systems. $\endgroup$
    – uhoh
    Sep 5, 2020 at 21:57


You must log in to answer this question.

Browse other questions tagged .