Strange Non-Uniformity from Array of LED Light Sources

  • 30 June 2021
  • 6 replies

I am having a strange issue with a 4x4 array of LEDs using Cree XP-E2 source file placed inside a square mirror tunnel with LEDs on one end and detector on the other.

The issue is that there is a slight gradient on the detector and I can't figure out where it's coming from. ZAR attached.


Best answer by David 1 July 2021, 00:03

View original

6 replies

Userlevel 6
Badge +4

Hi Andrey,

I think the assymetry is in the source file. I rotated all the sources by 90 degrees and got this:

I then put the a single source in a file and looked at the centroids in position and angle space. They are not centered, with the Y direction being the worst. I attach that file.

Hi David,

Thanks for checking, I thought maybe the source files had an angle to them but wasn't sure how to check.

From your check, what do you think should be the tilt angle that I should apply to the source object?

Userlevel 6
Badge +4

A good way to find that would be to open the single source ZAR I attached. Make tiltx and tilty on the source variables. Put a weight of 1 on the ang x centroid and ang y centroid in the merit function. Then optimize. It will determine the tilts needed to make the two centroids zero.

Edit: When I did this I got a Tilt About X of 2.629 degrees as correcting the centroid.

David, would you mind posting your ZAR for how you found tilt about x? I'm new to ZEMAX and not familiar with Merit Editor.

Thank you


From manual optimization, trying various numbers, I got TiltX=1.2 and TiltY=-0.32, roughly.

Your TiltX=2.629 is way off.

Userlevel 6
Badge +4

Hi Andrey,

I attach the ZAR file. Non-sequential merit functions are more diffcult to decipher than sequential merit functions because the meaning is built into the parameter values of the NSDD operand, which are a bit complicated. The merit function was built up using the merit function wizard, which is easier than doing it manually.

You will see in the merit function multiple instances of NSDD operands which measure the centroids in position and angle space. Only the angle space operands have non-zero weights.

Optimization, with x and y tilts variable on the source, reduced the angular centroid values to effectivey zero with Tilt About X of 2.658 and Tilt About Y of -0.285.

When these tilt values are used in the original file the result is this:

There are many good resources for becoming more familiar with optimization and the merit function. For example:

How to design a singlet lens, Part 3: Optimization

How to optimize non-sequential optical systems

Kind regards,