Optical System Design

The meaning of optical system design varies depending on whom you ask. In my view, it’s more of an ongoing process than a single task. A process is something you execute continuously, using feedback to refine and improve each stage of development. On the other hand, a task focuses on developing a specific system or module to meet particular requirements.

When we treat the design process as a series of isolated tasks, we will lose the benefits of continuous learning. One key advantage of a process-driven approach is how it allows us to develop our skills in predicting the performance of our designs.

To achieve this, we cannot rely solely on measuring the parameters directly related to customer requirements, while overlooking the optical performance of optical subsystems. In many cases, the optical system is just one component of a larger machine. If the machine performs correctly even once, we often draw the conclusion that all parts are functioning well. However, in my experience, that assumption is frequently incorrect. Complex systems can mask underperforming components. Just because a machine works doesn’t mean all its parts are performing as designed.

There is a key difference between optical system design and what is often confused with it: lens design. Lens design starts with specific requirements like size, imaging type, wavelength range, field angles, image size, and optical quality across these factors. In contrast, system design involves determining what those parameters should be. Once these parameters are clear, a ray-tracing specialist using software like Zemax/OpticStudio or Code V can deliver excellent results. Without this foundation, everything becomes guesswork or just a best effort.

For instance, how do we know how much spherical aberration, field curvature, or distortion is acceptable to meet customer needs? Are there coherence requirements, either spatial or longitudinal?

Let’s dive deeper into the mathematics. How do we describe the optical performance of an imaging system? Is it appropriate to use Zernike polynomials? Many argue that for circular pupils, the Zernike set is ideal. For telescopes, projectors, or camera lenses, that’s absolutely correct. However, if performance requirements aren’t directly tied to wavefront variance, the Zernike set may not be the best fit—even if it’s often preferred or widely liked. Determining the right approach is the optical system designer’s responsibility.