PROCEEDINGS OF SPIESPIEDigitalLibrary.org/conference-proceedings-of-spieNext generation optical designmethodologyCroce,AlessandraAlessandra Croce,"Next generation optical design methodology,"Proc.SPIE11287,Photonic Instrumentation Engineering VIl,1128709(2 March 2020);doi:10.1117/12.2544265SPIE.Event:SPIE OPTO,2020,San Francisco,California,United StatesDownladed From:https://www.spiedigitallibrary.orgconference-proceedings-of-spie on 04 Mar2020 Terms of Use:https://www.spiedigitallibraryorgfterms-of-useNext Generation Optical Design MethodologyAlessandra Croce*Zemax Europe Ltd.,8 Riverside Business Park,Stoney Common Road,Stansted Mountfitchet,CM24 8PL,United Kingdom*alessandra.croce@zemax.comABSTRACTAs the requirements for high performance optical systems become more and more demanding it is increasingly importantthat every stage of the optical design process is tuned to ensure that manufactured systems meet specification.Traditionallythe design steps of optimization and tolerancing have been treated separately which could result in a large divergencebetween the optical performance of the nominal,or unperturbed,system and that of an as-built system with realisticmanufacturing and assembly errors.This divergence can lead to lower product yields or the requirements for tightertolerances and hence increased costs.In this paper we demonstrate how including the effects of tolerance defects in theoptimization step,through Zemax OpticStudio's new High-Yield Optimization,can result in systems with higher as-builtperformance at potentially lower cost.We will present a case study showing the utility of this technique on a high-performance imaging system (such as may be found in medical or consumer electronics applications).The results of theHigh-Yield Optimization will be compared to those of a system designed to an identical set ofspecifications and using thesame realistic tolerances with the traditional method.A detailed analysis of the resulting design forms will be performed,and the key considerations and improvements of the High-Yield Optimization discussed.Keywords:Optical design,Optimization,Tolerancing,Zemax OpticStudio,Production yields,As-built performance,High-Yield Optimization,Desensitization1.INTRODUCTIONThe benefit of accounting for system tolerances in the early stage of an optical design is well known.Including the effectof realistic manufacturing defects in the optical system's optimization phase has indeed the advantage of excluding over-sensitive designs,that would then be difficult to produce and assemble.To help optical engineers design more easilymanufacturable systems,Zemax has recently implemented in OpticStudio a new optimization method.High-YieldOptimization is based on minimizing the angle of incidence of rays at each optical surface,which can be shown toeffectively reduce the sensitivity of the system to optical aberrations induced when tolerance defects are applied.Thetechnique uses ray angle as a fast,numerical approximation for the rigorous sensitivity to tolerance defects,and can beemployed to achieve less sensitive designs without incurring the limitations and drawbacks of other desensitizationtechniques.High-Yield Optimization is of general applicability,as it is not based upon first order optics,system symmetryor assumptions about small tolerances.It can be easily integrated in typical design workflows and is appropriate to useeven in the earliest stages of the optical design.Previous literature!has shown how the High-Yield Optimization can beused in conjunction with global algorithm techniques to find less sensitive design forms starting from plane parallel platesof optical materials.In this paper,we focus instead on further improving a pre-optimized system and on analyzing theimpact of the High-Yield Optimization on its sensitivity and performance yield.Proc.of SPIE Vol.11287 1128709-1ference-proceedings-of-spie on 04 Mar 20202.CASE STUDY2.1 MethodologyWith the purpose of investigating the effect of the High-Yield Optimization on a system's sensitivity,two identical copiesof the same OpticStudio file were used:one is optimized for Best Nominal Performance and one for ImproveManufacturing Yield.Both options are now available as optimization goals in OpticStudio's optimization wizard,as shownin Fig.1.Merit Function EditorWizards and OperandsMerit Function:0Optimization FunctionBoundary ValuesCurent Operand (1)☐Glass Min:30Max1e+03X Weight:Rings:3Edge Thickness:0YWeight:Arms:60Type:RMS0Max1e+03CentroidOptimization GoalAAdd Favorite OperandsSave SettingsLoad SettingsReset SettingsFigure 1.The new Optimization Goal section (red box)of the Optimization Wizard allows users to choose between the traditionalBest Nominal Performance and the new Improve Manufacturing Yield,which helps desensitize the design.Best Nominal Performance represents the traditional optimization methodology,aiming solely to improve the systemperformance based on the chosen image quality criterion.Improve Manufacturing Yield represents instead a newoptimization methodology in OpticStudio.When Improve Manufacturing Yield is chosen,the wizard automatically adds,in addition to appropriate operands for the image quality criterion,HYLD operands for each surface in the design2.Thesenew operands apply a penalty term whose magnitude is linked to the ray angle of incidence/exitance on the specifiedsurface.HYLD operands are added for each ray for the chosen Pupil Integration and are all optimized towards zero.Theirimportance with respect to the image quality criterion operands can be scaled using the Weight setting,which is userdefinable and whose impact on the resulting system sensitivity will be discussed below.Due to how optical aberrations areinduced by the non-linearity of Snell's law,limiting the angle of incidence at each surface typically makes the design lesssensitive to both manufacturing and assembly perturbations.Note that HYLD operands use cached ray tracing information,so their presence has typically a negligible effect on the merit function calculation speed and,consequently,on the speedof the optimization.For the sake of simplicity,in the following discussion the file optimized for Best Nominal Performancewill be referred to as BN,while the file optimized for Improve Manufacturing Yield will be referred to as HY.2.2 System description and specificationThe system chosen for this case study is a laparoscope,which is a high definition rigid endoscope typically employed forminimally invasive diagnosis or surgery.A laparoscope represents a promising candidate for this investigation as the High-Yield Optimization has proven particularly effective in other complex designs with a high number of surfaces'.The startingpoint of the investigation is an older laparoscope file that needs to be reoptimized to meet the more stringent specificationsProc.of SPIE Vol.11287 1128709-2d From:https://www.spiegitallibrary.org/conference-proceedings-of-spie on 04 Mar 2020