Built-in Coaxial Telecentric Lens Selection Guide 0.3X–2.0X

This article is also available in 简体中文.

Short answer. For flat, reflective parts that must be measured with even contrast across the whole field, the cleanest solution is a telecentric lens with a built-in coaxial (epi-illumination) port: the beam is injected inside the barrel and is therefore collinear with the imaging axis by construction, instead of being aligned by hand at every mount. In the POMEAS coaxial telecentric line the usable range is 0.3X to 2.0X, at 65 mm, 110 mm and 200 mm working distance. Two assumptions common in RFQs do not match the datasheets: there is no 0.2X model with built-in coaxial illumination, and no model is specified for 12 MP. What follows is the real model matrix, the geometry behind it, and the formulas to close the camera/lens match.

1. How built-in coaxial illumination works inside a telecentric lens

A telecentric lens with a coaxial port contains a beam splitter on the optical axis inside the barrel. Light enters through a side port, is deflected 90°, and travels toward the object along the same axis the lens images along; the return beam passes the same splitter and reaches the sensor, at the cost of one throughput loss. The result is an on-axis bright-field geometry defined by the lens mechanics rather than by a bracket on a rail, so it cannot drift with handling.

2. Why coaxial light plus object-side telecentricity is the combination that matters

A polished, flat surface behaves like a mirror: it returns light into the lens only when the illumination is normal to it — retro-reflection. Coaxial geometry creates exactly that condition, so flat regions perpendicular to the axis appear bright. A scratch, dent, laser mark or coating defect tilts or scatters the surface locally; those rays leave the aperture and the defect appears dark. Hence it is the standard choice for scratch, character, 2D-code and surface-defect inspection on mirror-like metal, glass and wafers.

Telecentricity then governs uniformity across the field. In a conventional lens the chief ray angle changes from centre to corner, so the effective incidence angle does too, and a mirror-like part shows a brightness gradient even under perfect coaxial light. An object-side telecentric lens keeps the chief rays parallel to the mechanical axis over the whole FOV. The official figures here are object-side telecentricity below 0.02° for the 1.2X and 1.5X models and below 0.016° for LTCM2-65C-S, which is what lets one exposure threshold work across the part.

3. Built-in coaxial vs external coaxial illuminator on a standard telecentric lens

The alternative is a plain telecentric lens with a separate coaxial light unit clamped in front of it. Both give on-axis illumination; they differ in where alignment error, volume and cost sit.

Two consequences follow. A volume-built station that must stay field-serviceable is better served by the built-in option, which removes an alignment that would need re-verification after every lens swap. Very high irradiance over a large field is better served by an external unit, whose aperture is not bounded by the barrel.

4. The POMEAS coaxial telecentric model matrix (0.3X–2.0X, WD 65–200 mm)

Coverage is model by model — a coaxial port is not a universal feature and must not be assumed for a part number. Figures are the official coaxial-variant values.

Two structural facts stand out: the lowest magnification with a coaxial port is 0.3X (next step up 0.5X), and the largest sensor in this family is 2/3″, with some models published down to 1/4″. A 1.1″ camera (Φ18.4 mm) falls outside the family, so you move to a 1.1″ design — POMEAS builds a 1.0X 1.1″ bi-telecentric lens at 110 mm WD with a Φ18.40 mm object field — and treat illumination separately. Whether it can be supplied with a built-in coaxial port is to be confirmed with POMEAS engineering.

Sensor-format coverage is per model — PMS-LTC1-200C spans 1/4″ 3.2×2.4 mm to 1/1.8″ 7.1×5.4 mm, PMS-LTC15-110C 1/4″ 2.1×1.6 mm to 2/3″ 5.9×4.4 mm, LTCM2-110C 1/4″ 1.6×1.2 mm to 2/3″ 4.4×3.3 mm.

5. The working-distance trade-off: 1.0X at three different working distances

The three 1.0X built-in-coaxial models show the WD trade-off with nothing else changing.

Longer WD buys clearance in front of the part and a deeper depth of field, and pays for both with a coarser resolution and a larger F#: pushing the lens away to make room for tooling costs resolution roughly in proportion to the aperture number, and no camera upgrade recovers it.

6. Formulas and a worked example: magnification, FOV, depth of field and resolution

The four relations you need for this family are:

β = W_sensor / FOV  →  FOV = W_sensor / β   |   DOF ∝ 1 / β²   |   R ∝ λ · F# (diffraction limit)

Worked example — FOV. Take 2/3″ and derive its active area from the official FOV of the 1.2X model (7.3 × 5.5 mm): 1.2 × 7.3 = 8.76 mm wide, 1.2 × 5.5 = 6.6 mm high (derived by formula). Applying FOV = W_sensor / β:

• PMS-LTCM03-110-C at 0.3X → 8.76 / 0.3 = 29.2 × 22.0 mm. Cross-check: the non-coaxial 0.3X 110 mm model publishes FOV(2/3″) = 29.3 × 22.0 mm — the two share the same object field.
• PMS-LTCM1-110C at 1.0X → 8.76 × 6.6 mm.
• LTCM2-110C at 2.0X → 8.76 / 2 = 4.38 × 3.3 mm, against the official 4.4 × 3.3 mm; the 0.5% gap is rounding in the published table.

Worked example — depth of field. DOF ∝ 1/β², so the 1.2X and 1.5X models of one family relate by (1.2/1.5)² = 0.64. From the official 1.2X value of 0.67 mm the predicted 1.5X depth of field is 0.67 × 0.64 = 0.429 mm (derived by formula), against the official 0.44 mm — about 2.5% apart. Use the rule for a first sizing pass only; quote the datasheet value in the final spec.

Worked example — resolution vs aperture number. For a diffraction-limited system R ∝ F#, so two 1.0X models scale their resolutions by the ratio of their F# values. From PMS-LTC1-110C (F# 19.94, R = 13.42 μm) to PMS-LTC1-200C (F# 33.18): predicted R = 13.42 × 33.18 / 19.94 = 22.33 μm (derived by formula), against the official 22.37 μm — about 0.2% apart. Use it to sanity-check a long-working-distance lens quoted with an implausibly fine resolution.

7. Pixel matching: the rule to use instead of a "12 MP" claim

The published camera-resolution statements for this range are limited: the non-coaxial 0.2X model LTCM02-110 is marked max 3 MP, and the 2.0X coaxial LTCM2-110C up to 5 MP. Per-model 12 MP support is not published, so a claim that "this lens supports 12 MP" is unverified. Screen the pair with the sampling interval instead:

p_obj = p_pixel / β   →   the camera's sampling-limited object-side feature is 2 · p_obj; the lens is not the bottleneck only when R ≤ 2 · p_obj.

With a 2/3″ camera whose pixel pitch is 3.45 μm (taken from your own camera datasheet — not a POMEAS figure), the check runs as follows:

• 0.3X: p_obj = 3.45 / 0.3 = 11.5 μm, so 2·p_obj = 23.0 μm. Lens R = 11.18 μm ≤ 23.0 μm → the lens has roughly 2× margin; the camera is the limit.
• 1.0X: p_obj = 3.45 μm, 2·p_obj = 6.90 μm. Lens R = 7.46 μm > 6.90 μm → the lens is now limiting, and the smallest object-side feature the system delivers is about 7.5 μm rather than the 6.9 μm the camera alone could sample.

In short: low-magnification coaxial models carry the most headroom for high-pixel-count cameras; from roughly 1.0X upward the lens resolution, not the sensor, sets the floor. Run this two-line calculation before accepting a megapixel number from either side. The published resolution figures do not state whether they are object-side or image-side — confirm the convention with POMEAS engineering before using them in a tolerance budget.

8. Step-by-step selection procedure

1. Confirm a bright-field case. Coaxial illumination suits flat, specular or semi-specular parts: mirror metal, glass, wafers, coated flats. On a matte surface it gives you nothing — use diffuse or dome light.
2. Fix the object field from the tolerance, not the image. Required FOV = part envelope + fixture positioning uncertainty, on both axes.
3. Pick the magnification from the intended sensor: β = W_sensor / FOV, kept inside 0.3X–2.0X. Below 0.3X there is no coaxial model — change the sensor or the station geometry.
4. Check the sensor format against that model's own published list, not the 2/3″ family maximum.
5. Check depth of field against part flatness plus stage Z repeatability. Read it from the table; use 1/β² scaling only to interpolate.
6. Run the pixel-matching test from section 7 before locking the camera.
7. Choose working distance last, once tooling clearance is known — and remember the resolution penalty of section 5.
8. Decide built-in vs external coaxial using section 3: built-in for alignment stability and a compact stack; external for very high irradiance.

For the measurement-accuracy side — how telecentricity and distortion tolerances turn into a repeatability budget — the telecentric measurement accuracy notes go further than this guide. If you are still deciding whether a telecentric lens is the right tool at all, start from what a telecentric lens is and when to use one and from the counter-case in when a telecentric lens is the wrong choice.

9. Where built-in coaxial illumination stops working

Three limits apply. A beam splitter costs throughput, so built-in coaxial is poor when the light budget binds. A matte surface gives no specular return and therefore no coaxial contrast. And coaxial light says nothing about feature height: vertical walls and deep recesses need a second channel.

Check the barrel length, port geometry and FOV table for your magnification: 0.3X PMS-LTCM03-110-C, 1.0X PMS-LTCM1-110C, 2.0X LTCM2-110C, the long-WD 200 mm PMS-LTC1-200C, the short-WD 110 mm PMS-LTC1-110C, and the full telecentric lens catalogue.

Send the part drawing, the surface finish, and your camera's sensor format and pixel pitch — contact our optical engineers with the station sketch and we will come back with a specific model, not a range.

Frequently Asked Questions

Can I get a 0.2X telecentric lens with built-in coaxial illumination?

No. In the POMEAS telecentric range the 0.2X 110 mm model is a non-coaxial design, and its official specification states a maximum camera resolution of 3 MP. The lowest magnification available with coaxial illumination is 0.3X, the PMS-LTCM03-110-C at 110 mm working distance on a 2/3″ sensor, with 11.18 μm resolution and 4.44 mm depth of field. If your field of view really forces you below 0.3X, the answer is not a different lens — it is a larger sensor, a taller station, or splitting the part into two inspection positions.

Does the built-in coaxial port work with a 12 megapixel camera?

POMEAS does not publish a per-model megapixel rating for this family, so a flat "supports 12 MP" claim cannot be made for any of them. The only two published statements are a 3 MP maximum on the non-coaxial 0.2X model and up to 5 MP on the LTCM2-110C. Instead of chasing a megapixel number, compute the camera's object-side sampling interval as pixel pitch divided by magnification, double it, and compare that with the lens resolution figure. If the lens resolution is finer than the doubled sampling interval, the camera is your limit; if it is coarser, the lens is.

Built-in coaxial or an external coaxial light in front of a normal telecentric lens — which should I pick?

Built-in coaxial wins on alignment stability and on volume. Because the beam splitter sits inside the barrel, the illumination axis and the imaging axis are the same line by construction and stay that way after a lens swap or a service visit; there is also no light housing or bracket occupying space in front of the part. An external unit wins if you need very high irradiance over a large field, because its aperture is not bounded by the lens barrel, and if you want to replace a failed light without taking the lens out of production. For a compact station built in volume, built-in is usually the better default.

My part is a polished metal cover. Why is the image bright in the middle and dark at the edges?

That is the classic symptom of a non-telecentric imaging path under coaxial light. With a conventional lens the chief ray angle changes from centre to corner, so the incidence angle of the illumination changes across the field as well; a mirror-like surface only returns light to the lens near normal incidence, so the corners lose signal. An object-side telecentric lens keeps the chief rays parallel to the axis across the whole field. In this family the published object-side telecentricity is below 0.02° for the 1.2X and 1.5X models and below 0.016° for the LTCM2-65C-S, which is what allows one exposure setting to work across the entire frame.

How do I choose between the 110 mm and the 200 mm working distance built-in coaxial models?

Decide on mechanical clearance first, then accept the optical cost. The three 1.0X built-in coaxial models show the trade clearly: at 65 mm working distance the resolution is 10 μm at F# 14.29; at 110 mm it is 13.42 μm at F# 19.94 with 1.6 mm depth of field; at 200 mm it is 22.37 μm at F# 33.18 with 2.67 mm depth of field. Longer working distance buys clearance and depth of field and pays for them in resolution, which no camera upgrade can recover. Do not fit a 200 mm model "for safety" if a 110 mm model clears the tooling.

How much depth of field can I expect, and how accurate is the scaling rule?

Read the depth of field directly from the model table rather than estimating it. Within one family depth of field scales as one over magnification squared, which is good enough for a first sizing pass: from the 1.2X figure of 0.67 mm the rule predicts 0.429 mm at 1.5X, and the official PMS-LTCM15-110C value is 0.44 mm — about 2.5% apart. Use the rule to interpolate between the magnifications you are considering, then write the datasheet value of the chosen model into the final specification. Also remember depth of field is an imaging tolerance, not a positioning allowance; the fixture's Z repeatability consumes part of it.