Hi Mark Nicholson,

Thanks for the reply. 

Hi, how do you account for the resolution of the compensator? Example - I want to use decenter of a lens to compensate, but it is hard to adjust anything to better than 7µm accuracy. How would this error be accounted for? 

Hey David,

The 2 ways I would account for this finite accuracy is:

• Change the Criterion > Cycles to a finite amount.  A compensator is really just a variable that gets optimized after all the other tolerance operands (perturbations) are applied.  When Cycles is set to Automatic, then the optimizer will keep on perfecting the compensator until the merit function value doesn’t get better, which can typically be in the 10th decimal place.  You’ll need to determine the number of cycles based on how tight/loose your tolerance operands are.
• Save your Monte Carlo runs and post process the results.  Once you have all the MC runs saved, you can create a ZOS-API script that loops through all the saved files, performs a 7um X/Y decenter on the compensator, and then plot the performance of the system.  The saved MC files will give you the “perfect” compensation for your perturbed system and your post processing will give you a distribution for your 7um accuracy.  Conceptually, this would be similar to running a Through Focus MTF analysis but you’re changing the compensator’s decenter, not the focal plane’s location.