Cancelling the ghost image

Everything that concerns our open-source autocollimator
Post Reply
User avatar
Luc
Site Admin
Posts: 101
Joined: Fri Mar 17, 2023 9:18 am
Contact:

Cancelling the ghost image

Post by Luc »

The initial version of the autocollimator suffers from a ghost image which is 1/4th of the intensity of the main signal:
autoco ghost.png
autoco ghost.png (90.34 KiB) Viewed 6245 times
It originates from a reflection on the chrome-on-glass target which is reflective. Because of the beamsplitter, the ghost signal is attenuated by a factor (1/2)² (assuming 100% reflectivity of the chrome target -- it's a bit lower in practice).

This phenomenon is also well known with light reflecting on uncoated camera silicon sensor which has a ~30% reflectivity due to the refractive index of air/silicium. The problem is usually alleviated by inserting a neutral density filter in front of the camera such that reflected light goes twice through the filter, reducing the intensity by a factor (1/N)².

It should be possible to do the same in the autocollimator by placing a neutral density filter just in front of the chrome target and before the beamsplitter. Assuming a main gray level of ~200 (80% of saturation is a good rule of thumb to avoid non-linear effects of some camera sensors), the ghost signal should be around ~50 gray level. Decreasing it below 2 gray level would therefore require a filter with N=5 (OD 0.7). Thorlabs has such a filter (1") under the reference NE07B-A.

Note that the initial signal will be decreased by a factor of 5 so you will need 5x as much exposure for the same LED intensity. Using stronger attenuation is possible at the expense of larger exposure time. I wouldn't go above OD 1.7 because on typical camera sensor it doesn't make any sense to me due to their limited dynamic range -- noise level.

I did not test this solution yet but I'm pretty confident it will work. I ran out of budget for 2023 so I won't be able to test it before next year. Since this is a long time ahead, I wanted to share the hypothetical fix on the forum so that other people might eventually try before me and report if it works :)
Post Reply