Hi Fatemah

Here’s a few thoughts for you.

1. Leave the weights of the default merit functions alone. That means everything after the DMFS operand should not be edited.
2. I tend to set the weights of my own operands (those before the DMFS) at 1 and see what I get. The first place to start increasing weights is the field and wavelength editors, not the merit function editor. So if you want to trade some on-axis performance for better off-axis performance, increase the weight on the off-axis fields. 
3. Rebuild the merit function after changing the weights or the number of field and wavelength points, so the new data is built into the MF.
4. In general, weights of 100 should be avoided. I liken this to shouting at the optimizer  It is giving you the best compromise between all the various targets you have given it, and if a weight of say 5-10 is not giving you what you want, it’s more likely that the design itself is working as hard as it can, and your best resort is to increase the number of variables by adding surfaces or aspheric terms.
5. Gaussian Beam Paraxial Size, GBPS, will usually optimize at a different location to RMS Spot, as the RMS spot accounts for aberrations and GBPS does not. If you put conflicting requirements like RMS Spot and GBPS in the merit function, you’ll end up satisfying neither, but get some average that smears the two requirements together.
6. The FICL results should converge as sampling is increased. Make sure the RMS wavefront error is a quarter of a wave or less before using FICL.
7. The best way to optimize for encircled energy is to make sure the RMS Spot is somewhere around the Airy disk (say within a factor of two or so) before trying to optimize. Use the Encircled Energy Analysis feature to estimate the sampling needed.

Hope that helps,

• Mark

Thank you a lot Mark. Indeed helpful.

Regarding weigth selection, I was following the spectrometer implementation which selected 10 or 100 in many cases for different boundary conditions. Therefore, I was confused about the scale or how to choose the number. 

https://support.zemax.com/hc/en-us/articles/1500005578862-How-to-build-a-spectrometer-implementation

Fatemeh

Hey Fatimeh,

I don’t know why the author of that article used such large weights, but I stick to my original comments. Set pre-DMFS weights to 1, and modify only when you have reviewed the results of the optimization. Large weights may well be necessary, but shouldn’t be your initial setting.

• Mark