Telecentric Lens Numerical Aperture: Resolution, Depth and Light

Numerical aperture sets three things at once

Numerical aperture describes the cone of light the lens accepts. A wider cone carries finer spatial information, which improves resolution. It also means rays arrive over a wider range of angles, which shortens depth of field. And it collects more light, which improves signal on dim or fast-moving parts.

Because all three move together, NA cannot be tuned to fix one without affecting the others. A specification that demands fine resolution and large depth of field simultaneously is asking for a wide cone and a narrow one, and the optical design has to compromise somewhere.

Understanding which of the three constraints is actually binding is the core of telecentric lens selection. In most industrial applications it is depth of field, because parts have height variation and the machine cannot refocus per part.

The measured ranges published for the POMEAS telecentric models express these consequences directly: object-space resolution over 1.1 to 398 um and depth of field over 0.017 to 172 mm, both varying with the model's magnification and aperture.

Reading those two figures together, rather than one at a time, is what makes the NA trade visible.

Higher NA gives finer resolution and shallower depth

The relationship is a physical one rather than a design preference. Finer resolution requires a wider angular cone, and a wider cone means the focus falls off faster on either side of the best plane. Both effects scale together, so a lens with twice the resolution has roughly a quarter of the depth of field at the same wavelength.

For a part that is flat and presented at a controlled height, that trade is free: the application does not need depth, so it can spend it all on resolution. This is the case where a high NA model is clearly correct.

For a part with height variation, the trade is real. Either the depth is available and the resolution is coarser, or the resolution is fine and the part must be brought into focus some other way, typically with an autofocus axis that costs cycle time.

The decision is therefore about the part's geometry before it is about the lens. Stating the height tolerance first turns the NA choice into a calculation rather than a preference.

In practice the height tolerance plus the resolution requirement together select a narrow band of models, and the rest of the range drops out immediately.

Higher NA also means more collected light

A wider cone collects more photons from the same surface, which matters for short exposures, dim illumination and low-reflectance parts. In those conditions a higher NA lens can be the difference between a usable image and a noisy one, independently of its resolution advantage.

This is often the dominant benefit in practice. A dark, matte or deeply textured part returns little light, and increasing illumination is limited by heat, cost or the risk of washing out contrast. A lens that collects more of what is returned solves the problem optically.

The same property makes high NA models more sensitive to stray light arriving from outside the field, so baffling and coaxial illumination geometry matter more. A wider acceptance cone accepts unwanted light as readily as wanted light.

For applications using coaxial illumination on specular parts, the NA also sets how much of the illumination cone can be used, which affects contrast on edges and on shallow features.

Weighing light budget alongside resolution and depth avoids the common outcome where a correctly specified lens produces an image that is simply too dim to threshold reliably.

Low NA models exist for a reason

A low numerical aperture model is not an inferior version of a high NA one. It trades resolution for depth of field and for a longer working distance at the same magnification, and that combination is exactly what many production stations need.

A part with several millimetres of height variation presented on a conveyor needs depth far more than it needs micrometre resolution. A low NA model keeps the whole height range in focus and removes the need for an autofocus axis, which is both cheaper and faster.

Low NA also tolerates more tilt. A part that is not perfectly presented has its edge displacement reduced when the acceptance cone is narrow, so the measurement is less sensitive to fixturing error. That robustness is often worth more than the resolution given up.

The POMEAS telecentric range spans magnification from 0.061 to 12.5 X and working distance from 9.5 to 468 mm, so the low NA options are available across the range rather than only at the coarse end.

Choosing a low NA model deliberately, with the trade stated, is a stronger engineering decision than defaulting to the finest resolution available.

Connecting NA to the measurement tolerance

The measurement tolerance determines how much resolution is actually needed. If the tolerance is 20 micrometres, a lens resolving 2 micrometres provides no additional benefit and costs depth of field and light budget that could be spent elsewhere.

A useful rule is to specify resolution at roughly one third of the tolerance, which leaves margin for repeatability, fixturing and edge detection. Finer than that does not improve the accept-reject decision.

This rule prevents the most common over-specification in telecentric selection, where the finest available model is chosen and the resulting depth of field is too small for the part.

Where the tolerance is genuinely tight, the depth constraint usually forces a different solution: a flatter fixture, a shorter height range, or an autofocus axis rather than a lower NA lens.

Stating the tolerance and the height variation in the same document is what makes this trade visible before a model is ordered.

Practical selection sequence

Convert the measurement tolerance into a resolution requirement, typically one third of the tolerance. Then convert the part's height variation, including fixture error, into a depth of field requirement.

Find the models that satisfy both at the required field of view. Usually this leaves a small set, and the remaining choice turns on light budget and working distance.

If no model satisfies both, the constraints are in conflict and something else must change: the fixture, the field, or the measurement principle. Trying more models will not resolve a physical conflict.

Once a model is shortlisted, confirm telecentricity and distortion over the frame area actually used, and confirm the light budget with the intended illumination.

Following this sequence keeps the NA trade explicit at every step, which is what makes the final specification defensible to someone reviewing it later.

Frequently asked questions

Can I increase depth of field by stopping down a telecentric lens?

Only on models with an adjustable aperture, and at the cost of resolution and light. Fixed-aperture telecentric lenses have their depth of field set by design. Where adjustment exists, stopping down helps depth but diffraction and reduced light quickly reduce the benefit.

Why does my high resolution telecentric lens lose focus so easily?

Because high resolution and shallow depth of field come from the same wide acceptance cone. A model resolving a few micrometres may have a usable height range of tens of micrometres. Parts with height variation need a lower magnification or an autofocus axis, not a better lens.

Does a wider numerical aperture improve contrast?

It improves light collection, which can improve signal to noise, but it also accepts more stray light from outside the field. On specular parts the effect on contrast depends on the illumination geometry. More collected light does not automatically mean better contrast.

What is a reasonable ratio between measurement tolerance and lens resolution?

Roughly three to one, meaning the lens resolves three times finer than the tolerance. This leaves margin for repeatability, edge detection and fixturing. Finer resolution than that does not change the accept-reject outcome and costs depth of field and light budget.

Are low NA telecentric lenses suitable for measurement?

Yes, when the tolerance allows it and the part has height variation. Lower NA gives more depth and a longer working distance at the same magnification, plus better tolerance to part tilt. The resolution given up is often not needed for the application.

How does numerical aperture affect working distance?

Higher NA generally means a shorter working distance at the same magnification, because a wide cone must be formed close to the object. Where a long working distance is required for clearance, a lower NA model is usually the only option that also keeps the magnification.

Lighting and camera lens selection

• Machine Vision Lighting Selection Basics
• Industrial Camera Lens Selection