Telecentric Lens vs. Standard FA Lens: Distortion, Accuracy, and Cost Compared

A standard FA lens images a part in perspective, so its magnification changes with object distance and its measured distortion reaches 1.0–2.4%. A telecentric lens keeps the chief rays parallel to the optical axis, holding distortion at 0.02–0.1% and magnification effectively constant no matter where the part sits. That single optical difference decides whether a machine vision station can measure dimensions or only detect presence.

Most lens comparisons stop at “telecentric is more accurate, and more expensive.” That is not enough to specify a system with. This article puts the two families side by side using published distortion and depth-of-field figures from current POMEAS datasheets, works through what those numbers mean in microns on a real field of view, and then shows where the extra cost of a telecentric lens stops being worth paying.

The short answer

Choose a telecentric lens when the output of the station is a number — a dimension, a position, an angle — and that number drives an accept/reject decision. Choose a standard FA lens when the output is a judgement — present or absent, readable or not — and when the field of view is large relative to any lens barrel you could physically mount. Everything below is the reasoning behind that split.

Three optical properties separate the two families

Distortion is the property people quote, but it is only one of three. All three come from the same root cause: where the aperture stop sits relative to the lens groups.

The first three rows are geometry, not tolerances. No amount of polishing makes a standard FA lens hold magnification constant with object distance, because the perspective is designed into where the chief rays converge. Software can correct the result of that geometry, but it cannot recover the depth information the lens never recorded.

Distortion, measured: what the datasheets actually say

These are published values from current POMEAS datasheets, not industry rules of thumb. Note that FA lens distortion is normally specified at a stated image height, because it grows toward the edge of the field.

The minimum sensor format, working distance, depth of field and resolution figures for each lens above are listed on its product page. As always, confirm against the current selection manual before designing a station — published values are typical for the stated configuration.

What a distortion figure costs you in microns

Percentages hide the size of the problem. Take a 2/3″ sensor (8.8 × 6.6 mm) at unit magnification, so the field of view is 8.8 × 6.6 mm, and measure a feature 5 mm from the optical axis.

That is a 10–80× difference in a quantity that lands directly in your tolerance budget. If the part tolerance is ±20 µm, an uncorrected 1% FA lens consumes the entire allowance before sensor noise, illumination or mechanical repeatability are considered. This is why measurement-grade stations start with the optics rather than with a more elaborate calibration routine.

Magnification stability when the part moves in Z

The second property is harder to see on a datasheet and easier to feel on a production line. A telecentric lens specifies telecentricity — how close the chief rays come to being truly parallel — typically as a maximum angle. The bi-telecentric PMS-10VP110C111T, for example, is specified at object-side telecentricity <0.1°.

On a well-fixtured line the part does not move half a millimetre — but parts vary in thickness, fixtures wear, and conveyor systems repeat to tens of microns at best. With an FA lens every one of those millimetres of Z variation reappears as a size error; with a telecentric lens it is suppressed to roughly the ray-angle term. This is the property that removes the need to recalibrate per part.

Depth of field: read the number, not the marketing

Telecentric lenses are often described as having unusually deep depth of field. The datasheets are more specific, and the values differ a great deal with magnification.

The pattern matters more than any single figure: at 0.2× the depth of field is measured in millimetres, at 2× it drops below a third of a millimetre. Depth of field trades against resolution and against numerical aperture — stopping down buys axial tolerance and costs the fine-detail response you bought the lens for. Estimate it against the real height variation of your parts, and remember that a telecentric lens does not have unlimited depth of field simply because magnification is stable.

Field of view, size, and the practical ceiling

To hold chief rays parallel across a field, the front group must be at least as large as the field itself. A 40 mm field of view implies a front element of roughly 40 mm; a 200 mm field implies optics most stations cannot mount or afford. Standard FA lenses carry no such constraint, which is why they remain the only practical choice for wide-field tasks such as pallet inspection, robot guidance and large-panel presence checks.

Note the working distance row. A telecentric lens is designed around a specific working distance and its magnification is only guaranteed at that value; an FA lens focuses from 0.15 m to infinity on a 25 mm focal length, for instance. Flexible focusing is a real advantage — just not for measurement.

Cost: what the money buys, and when it pays back

A telecentric lens costs several times a comparable-focal-length FA lens of similar resolution. The premium is not margin — it is glass and tolerance. The front group is larger than the field, the design needs additional precision groups to hold distortion and telecentricity near 0.1%, and every element must be aligned to a tighter angular budget.

The decision is therefore an economic one, and it usually comes down to three questions:

• Is the output a measurement? If a human or a downstream process acts on the number, the lens premium is normally cheaper than the scrap it prevents.
• How much rework does an error cause? On a high-value assembly, avoiding a handful of false rejects per shift repays the lens quickly. On a low-value part sorted by presence, it may never repay it.
• How much calibration time does the FA path need? Correcting 1–2% distortion in software requires a quality calibration target, stable lighting and a repeatable procedure — all of which have their own running cost. Count that before declaring the FA lens cheaper.

Decision table

When a standard FA lens is the correct answer

It is worth stating plainly, because over-specifying optics is as expensive as under-specifying them. A standard FA lens wins when the decision is qualitative rather than quantitative, when the field of view exceeds any mountable barrel, when working distance must stay adjustable, and when the required accuracy is looser than a properly calibrated FA lens can already hold. A well-calibrated FA lens with good illumination is perfectly adequate for a very large share of vision tasks — it is simply not a measuring instrument.

Frequently asked questions

1. Is a telecentric lens always more accurate than an FA lens?

For dimensional measurement, yes — the geometry guarantees it, since distortion is one to two orders of magnitude lower and magnification does not drift with object distance. For detection tasks the accuracy is wasted and the cost is not recovered.

2. Can I correct FA lens distortion in software and get the same result?

Software correction reduces distortion error across the field, and for moderate fields it can be good enough. It cannot correct perspective error caused by the part sitting at a different height, because that is a change in magnification, not a fixed geometric warp. If Z variation is part of the process, software is not a substitute.

3. Why does the distortion figure change with sensor format?

Distortion is specified at a stated image height, and it grows toward the edge of the field. A lens quoted at −1.78% at y = 8.0 mm on a 1″ sensor is quoted at −2.4% at y = 11.0 mm on a 4/3″ sensor simply because the outer field is being used.

4. Does a telecentric lens have unlimited depth of field?

No. Magnification is stable with object distance, but sharpness is not. Published values range from about 10 mm at 0.2× to under 0.3 mm at 2×. Depth of field is a resolution trade-off, not a free benefit of telecentricity.

5. Can I put a telecentric lens on any camera?

Check that the lens is designed for your sensor format or larger, that the mount matches (C-mount in most cases), and that the lens resolution matches your pixel pitch. A 2/3″ lens on a 1.1″ sensor will screw on but will not cover the sensor correctly.

6. Which POMEAS products map to each side of this comparison?

For measurement, the telecentric lens family spans 0.2× to 2× in object-space and bi-telecentric designs, including the LTCM02-110 0.2× for wide fields and the PMS-10VP110C111T 1.0× bi-telecentric for 1.1″ sensors. For detection and general imaging, see the FA lens family.

To go deeper on the optics themselves, read what a telecentric lens is and how it works, then work through the step-by-step selection guide to convert a tolerance into a magnification, working distance and depth-of-field requirement.

Key takeaway: the choice is not between a good lens and a cheap lens. It is between an instrument that reports a stable number and an imager that reports a stable picture. Decide which one your station needs, and the distortion figures will tell you exactly what the difference is worth.

Further reading: the test methods behind this article are documented in POMEAS Technical Reference the measurement-relevant distortion metric and how to measure it (§§8.7） + #s41).