Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Toward a paraxial pre-design of zoom lenses

: Milde, T.; Zimmermann, T.


Betensky, Ellis (Ed.) ; Society of Photo-Optical Instrumentation Engineers -SPIE-, Bellingham/Wash.:
Zoom lenses V : 10 - 13 August 2015, San Diego, California, United States; Fifth SPIE Conference on Zoom Lenses
Bellingham, WA: SPIE, 2015 (Proceedings of SPIE 9580)
ISBN: 978-1-62841-746-3
Paper 958005, 15 pp.
Conference on Zoom Lenses <5, 2015, San Diego/Calif.>
Conference Paper
Fraunhofer IOSB ()
lens design; nonlinear optimization; variational methods; zoom lenses; NAT

Optimizing the power distribution of fixed and moved lens groups as well as the motions of the latter, is typically a challenging part of the whole zoom lens optical design task. Once, the merit function is formulated to optimize an initial approach, the paraxial moving equations are solved implicitly in local optima. Hence, finding local optima becomes an ill posed problem when these equations cannot be solved uniquely for certain zoom configurations. Furthermore, an inappropriate initial power distribution can lead to large overall lengths, sensitive lens groups, small zoom ranges, induced aberrations and much more disadvantageous effects. From these reasons it appears as a logical consequence to first consider a paraxial pre-design of the zoom lens. This paper shows how first order aberrations, centering sensitivities as well as all common paraxial requirements can be formulated as a merit function for finding power distributions and (zooming) air spaces. In particular, the benefit of formulating zoom invariants as constraints in order to apply the Sequential Quadratic Programming (SQP) is shown. Based on a variation approach, an optimizable characteristic is introduced for control of the uniqueness of the moving equations. Global optimization methods like e.g. Differential Evolution can be used to obtain an initial paraxial approach. This approach can be improved using the SQP or the Damped Least Squares (DLS) method. Finally, the generation of an initial real lens system is described.