Every ordinary lens is a perspective device. Objects closer to the camera appear larger, objects near the frame edge are slightly compressed, and a part that moves 2 mm toward the lens will measure differently than the same part sitting 2 mm further away. For presence and classification work none of that matters. For dimensional measurement it is the difference between a gauge and a guess.
What telecentricity buys you
A telecentric lens accepts only light travelling parallel to the optical axis. The practical consequences are threefold:
Constant magnification. Parts at different heights within the depth of field image at exactly the same scale. A 10 mm feature measures 10 mm whether it sits on the fixture surface or 3 mm above it - which is what makes it possible to measure parts on a conveyor without a precision Z stage.
No perspective error. The sides of a hole or a boss are not visible, because the lens cannot see anything but the parallel ray bundle. You measure the true silhouette rather than a silhouette plus a sliver of the wall.
Low distortion across the whole field. Straight lines stay straight to the frame corners, so one calibration constant applies everywhere instead of a per-region correction table.
The costs, stated plainly
A telecentric lens is larger, heavier and several times the price of an equivalent fixed focal lens. Its field of view can never exceed its front element diameter, so measuring a 300 mm part means a front element over 300 mm across. Working distance is fixed by design, with no focus ring to turn. And most true telecentric designs have a modest depth of field, typically a few millimetres, which has to cover the part's height variation.
There are cheaper middle grounds. An object-space telecentric that uses a large front element to stay telecentric only at the object plane is common in measuring applications and costs less than a fully telecentric design. Adding a telecentric attachment to a standard lens is a third option, with its own magnification and aperture limitations.
Sizing a telecentric lens to a sensor
Magnification equals sensor width divided by field of view. If you have a 2/3 inch sensor 8.4 mm wide and need to inspect a 40 mm field, you need 0.21x magnification. Now check the two things vendors rarely put on the same page: the sensor size the lens supports, and the minimum pixel size it resolves. A 12 MP lens such as the MVL-KF3528M-12MP is specified for a 2/3 inch sensor, but a 20 MP 1 inch sensor pushes the edge-of-field resolution down, and a lens designed for 2/3 inch will not cover the larger image circle at all.
For a 5 MP measurement station, the MV-CS050-10GM's 3.45 um pixels are a forgiving pairing - roughly 2.45 um line pairs per pixel is well within the capability of a good 12 MP class telecentric. For a 20 MP sensor such as the MV-CS200-10GM, expect to pay more for the lens than the camera.
Do you actually need one?
Ask whether the measurement is dimensional and whether the tolerance is under about 50 um. If both are yes, telecentric is usually the only way to get a stable reading. If you are checking presence, counting features, verifying colour or reading a label, a conventional lens with a stable fixture will do the job at a fraction of the cost, and the money is better spent on illumination.
One more consideration that decides many projects: telecentric lenses need on-axis illumination to work as intended. A ring light on a telecentric lens wastes most of the aperture and reintroduces oblique rays. Budget for a coaxial or telecentric illuminator at the same time as the lens.