In a recent post I described some simple acrylic lenses I made using a simple press-molding technique. The methods were crude, but the results weren't too bad. Also, I had designed an assembly containing four minor variations of these lenses and submitted that for fabrication over the holidays. The injection molding step was a bit of an experiment- rather than using a full optic-grade firm, which would have cost us well into the five figures to try, we used U.S. based Protomold, who was able to create this mold in two weeks and make 100 assemblies (400 lenses) for a bit more than $2k. I again selected acrylic as the resin material for these lenses.

The picture above shows the parts as they came back (top and bottom side). Below shows a close-up of two lenses cut out from the above assembly.

 

The real test, of course, is that image quality. I mounted these lenses onto some of our image sensor chips using the same methods as that discussed in the above-quoted recent post, painted on an iris, and sealed the chip up. Below is a picture of me waving at the camera in 32x32 resolution.

 

I also took another picture of my backyard with a different chip and a different setup at 90x90 resolution. The field of view was roughly between 70 and 80 degrees, thus the pixel pitch was less than one degree. The image quality in this latter picture was not as good. Two factors probably contributed to this- First the finer pixel pitch could have exceeded the limits of the optics, second my method for removing fixed pattern noise was less accurate in this setup. Right now I do not know which of these two factors dominate.

One comment- There was in fact some shrinkage in the lens, on the flat bottom part that gets placed onto the chip. However this was small and easily filled in with the optical adhesive, which has almost the same index of refraction as acrylic.

One lesson learned regarding the injection mold design: There are four slightly different lenses in the above assembly. The difference is in the total thickness, with sequential lenses different by 25 microns. (It turned out this difference was moot compared to the varying thickness due to the amount of adhesive used.) This was to allow me to experiment with variations to compensate factors such as shrinkage and enlarging of the mold through polishing. However I made the mold family perfectly symmetrical (other than the small variations in lens thickness)! When I got the parts back, it was hard to find out which lens was which! Fortunately I found the sprue (where the plastic charge gets injected into the mold) and with careful eyeballing under a microscope, identified the lenses. But the lesson learned is that I should have added a slight marking or asymmetry to help me identify right from left.

Overall I am pleased with the results. For pixel pitches of about two degrees per pixel and up, this technique is adequate. Two degrees per pixel may not sound like much, but many flying insects have this type of resolution and do quite well. It may be that with the right iris and better fixed pattern noise cancellation, I could get the sharpness down to one degree or pixel, but this will have to wait.

Here again is the link to a zip folder containing the Alibre files for the mold: CYE_LensMold_Untested.zip

Views: 1870

Comment by Sky Monkey on January 11, 2011 at 10:05pm
Fascinating.  Thanks for sharing the process.
Comment by Lew Payne on January 14, 2011 at 10:39pm
I've read some interesting papers on this... Optic Flow Coding in Flys and Maps of Fly Eyes.  I also started some preliminary design involving pixel clusters for optic flow use.  I'd love to read more about this... would you mind sharing some of the sources and/or research papers from which you got your information?  The mind is a hungry thing, you know...

Comment

You need to be a member of DIY Drones to add comments!

Join DIY Drones

Groups

Season Two of the Trust Time Trial (T3) Contest 
A list of all T3 contests is here. The current round, the Vertical Horizontal one, is here

© 2018   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service