Telecentric versus Standard Lens: How to Choose

A standard lens forms a perspective image: bring the part a little closer and it appears a little larger. A part floating a millimetre forward within the depth of field therefore measures differently. The awkward part is that the amount of change depends on the distance and the magnification rather than being a constant, so there is no offset that cancels it — a different position each cycle means a different error each cycle. A telecentric lens makes the chief rays run parallel to the optical axis, so magnification stays fixed as the object distance changes. That entire dimension of error is removed at the source rather than corrected afterwards.

Three telecentric types, and what each one solves

What the market calls a telecentric lens is usually object side or double telecentric. Dimensional measurement almost always means double telecentric. Where the only goal is more even brightness toward the edge of the image — avoiding the threshold drift that vignetting causes — an image side telecentric may be sufficient, and there is no reason to pay for the object side specification. Writing that sentence into the requirement saves money that would otherwise disappear into an unused specification.

When a telecentric lens is required

• Parts with thickness or height steps. Measuring a standing cylinder, the depth of a flange or the step between two faces: a standard lens gives a different magnification at each height, so the dimension carries a systematic offset that tracks the height.
• Absolute dimensions that must be reported. A pass or fail verdict can tolerate a magnification error of a few percent; a reported measured value cannot, particularly where the record is retained.
• Parts that inevitably float within the depth of field. Parts on a conveyor or placed by a robot have limited positional repeatability. A telecentric lens removes that dimension of error entirely and leaves the rest to mechanics and algorithms.
• Features at the edge of the field. Standard lens distortion is worst at the edge, so dimensions measured near the border are the least accurate. Telecentric distortion is typically within a few parts per thousand, so one conversion applies across the whole field.

What it costs

The first cost is that field of view is limited by the front aperture. The field can never exceed the clear aperture at the front of the lens, so working across a 200 mm field means a front element larger than 200 mm — and that space has to come out of the machine envelope. The second cost is that working distance is fixed near a design value with very little adjustment range, which leaves almost no slack when a workstation is modified. The third is price: typically an order of magnitude above a standard lens for the same field, and double telecentric adds another step.

A fourth cost is easy to overlook: light throughput. The angular aperture of a telecentric lens is small, so it naturally collects less energy than a standard lens, and the same brightness requires a stronger light or a longer exposure. On a fast line that turns straight into a cycle time problem, and cycle time lost to exposure is not recovered by algorithms later. Lighting and the inspection time budget should therefore be worked out at selection time, rather than discovered when the hardware arrives.

Telecentric and depth of field

Telecentric lenses are often described as having a large depth of field, which is not quite right. The advantage is not a deeper field but constant magnification within whatever depth of field exists — and that depth is still limited by aperture, wavelength and magnification, often shallower than a standard lens of the same field. What it solves is dimensional consistency inside the sharp range, not the ability to image a taller object. Where the height range exceeds the depth of field, the correct answers are stepped scanning, two cameras looking from opposite sides, or a 3D method. Expecting the lens to stretch its depth of field is not one of them.