Telecentric Lenses: Easy to Mount, Hard to Align

Spent the weekend messing with telecentric lens selection, hit quite a few pits.

The cause was the client's production line. Doing dimensional measurement of metal parts, workpieces come along the conveyor belt, and there's always some wobble in the Z direction. They used ordinary industrial lenses before, where magnification follows object distance — get the workpiece a bit closer and the image gets a size bigger, and the measured dimensions drift along. The software side calibrated for two days, then a batch change meant recalibrating, and the engineers were very resentful.

Telecentric lenses are made for exactly this. In foreign terms, Telecentric Lens — chief rays parallel to the optical axis, entrance pupil at infinity. By analogy, an ordinary lens is like a crowd walking toward a door from all directions; a telecentric lens requires everyone to walk straight facing the door frame. The direct result is that as an object moves back and forth within the depth of field, the image size doesn't change. This line was driven by precision inspection demand in the early 21st century, used most in gear measurement-type scenarios.

I couldn't take the lens home to test, so I went with the team to the client site, and also got a demo video and spec sheet from Linghou to look at. This series has two types: internal coaxial telecentric and liquid lens. Internal coaxial, as I understand it, saves the external coaxial light source — light shines directly out along the lens axis, so when measuring reflective metal parts you don't have to arrange lighting around the workpiece. The liquid lens type focuses via electronic control, so switching product models doesn't require manual turning.

A few points from what I saw on site.

Installation isn't hard — standard mount, just screw it on. The hard part is adjustment. Working distance is fixed; too far or too close to the workpiece won't do, so you have to get the mechanical structure aligned first. Once coaxial light shines on it, scratches and edges on reflective parts come out very clean; switch to matte parts and the effect isn't as obvious, and you still need side light. Depth of field is indeed wide, but it's not infinite — beyond that range it still blurs.

On calibration, my original expectation was that it'd save effort, but it's not that optimistic in practice. Constant magnification means one calibration lasts a while — that's true, much better than recalibrating every two days with ordinary lenses. But low distortion doesn't equal zero distortion; precision measurement still requires a distortion correction first, to subtract that residual error. Telecentric lenses are expensive because they can push down the calibration cycle.

Price has to be calculated. For the same field of view, telecentric costs quite a bit more than ordinary industrial lenses; the specific multiple depends on aperture and magnification, and I won't guess without accurate numbers. Also, large field of view and high magnification are basically contradictory — wanting to see wide and see fine means multiple sets or a larger sensor, pushing cost up further.

So the conclusion is: it depends. For measurement scenarios with high per-piece value, tight tolerances, and unstable workpiece positioning, it's worth it — like precision hardware, connectors, gears. Conversely, if you're just doing presence detection, rough positioning, or code reading, an ordinary lens is enough, and telecentric is a waste. Defect inspection depends on type — surface scratches with coaxial light are good, but internal or large-area color-difference types, no matter how good the optics, still need algorithms to back it up.

For next steps, I plan to have the team take sample pieces of two typical workpieces and run a comparison test between telecentric and ordinary lenses, recording both calibration hours and misjudgment rates, then decide whether to switch this production line entirely.