Width Readings Drift With Distance: Telecentric or Fixture?

A width reading that moves when an operator nudges the part is perspective error, the signature of a conventional lens: apparent size grows as the part moves closer and shrinks as it moves away. That finding does not sentence the standard lens. The decision turns on one number: how much of your width tolerance the largest seating error consumes. If it is a small fraction, fix the fixture. If the part cannot sit at one height, or the tolerance is tight, buy the telecentric lens. Either way the maker should prove the choice and leave two numbers in writing.

The Two Fixes Side by Side

Before the physics, the choice in one view. The fixture attacks the cause of the variation, the part’s height. The telecentric lens attacks the sensitivity to it.

	Standard lens plus fixture	Telecentric lens
How it controls the error	Constrains the part to a repeatable height and orientation	Keeps magnification nearly constant across a depth band, so seating height stops mattering
What it buys you	Keeps the existing optics and calibration; cheapest path when seating can be disciplined	Stable scale even when parts vary in thickness or cannot be seated identically
What it costs you	Fixture engineering, operator discipline, and a wear check in the maintenance plan	Larger front barrel, smaller field of view for a given lens size, fixed working distance, higher price
How it fails	Seating drifts with wear, chips or a rushed operator, and the error returns silently	The part leaves the telecentric range, or tilt and lighting problems appear that no lens cancels
Evidence to demand	Fixture repeatability data from several operators on real parts	Measured width drift across the full depth band you actually run, at your working distance

One caution before you pick a column: “telecentric” is a family, not a feature. Object-side, image-side and bi-telecentric designs solve different halves of the problem, and our telecentric lens selection guide walks that split in detail.

Where the Extra Width Comes From

The symptom has a name and an arithmetic, and once you can do the arithmetic on your own station, the decision mostly makes itself.

The Nudge You Can See

A standard imaging lens works like your eye: its field of view is a cone, so the same object fills more of the frame when it is near and less when it is far. For gauging, the consequence is awkward. The magnification of a conventional lens changes with object distance, so an identical part seated one millimeter taller images slightly larger, and the software reports a width that never existed in the metal. Nothing is out of calibration. The optics are simply reporting the geometry they were given, and the geometry includes the nudge.

Turn the Nudge Into Microns

Estimate it on paper before anyone touches the station. For a standard lens at working distance WD, a small seating change dZ changes the magnification by roughly dZ/WD in relative terms. Work the example that matters to you: at a 200 millimeter working distance, a 1 millimeter seating error moves relative size by about half a percent, so a 40 millimeter width reads about 0.2 millimeter wide. Run the same nudge through your own numbers and compare the result with your width tolerance. If the worst realistic nudge eats a tenth of the tolerance, the fixture conversation is about discipline, not optics. If it eats the whole tolerance, no fixture culture will save the station. One practical note: define the nudge band from evidence, not from memory. Three sources cover it, measured once and written down: the fixture drawing’s stated seating tolerance, the worst reset an operator can produce on purpose, and any part-to-part thickness variation the process allows. Take the largest of the three, add half again for surprises, and that band is the range every candidate has to survive. This is the same trade-off reasoning used when an aperture change buys depth at the cost of resolution, which we quantify in the pick-and-place aperture guide: every optical comfort has a price, and the price is always a number.

Three Ways Out, Not Two

The debate usually gets framed as fixture versus telecentric, but a third path exists and fails in a characteristic way. Software calibration can correct fixed geometric distortion, the stable field-dependent bending that every lens has, but it cannot correct magnification error caused by an unknown seating height, because the software never learns how tall the part was sitting. Only a measurement of height, or a constraint on height, or optics indifferent to height can close that loop. So the real menu has three lines: constrain the height with a fixture and keep the standard lens; add height sensing or a hard stop and recalibrate; or remove the sensitivity with a telecentric design. In an object-side telecentric lens the chief rays run parallel to the optical axis, so magnification stays essentially constant while the part moves through its telecentric range; the RP Photonics optics encyclopedia gives a compact definition of telecentric lenses and their object-side and image-side variants. That indifference to seating is exactly the property our telecentric lens family exists to provide when measured dimensions must not change with depth or thick parts need constant magnification. If your problem is actually a wide band of part heights that all need reading rather than one width measured precisely, that is a depth-of-field selection question instead, covered in the parcel code depth guide, and it leads to a different purchase.

What the Maker Must Demonstrate on Your Station

Whether the recommendation is a fixture or new optics, accept one protocol and refuse improvisations. The demonstration happens on your station, with your parts, your lighting and your software, and it ends with numbers against your tolerance.

1. Freeze everything except the candidate. Same camera, same light, same exposure, same edge tool, same calibration routine. When only the fixture or the lens changes, the comparison measures the fix and nothing else.
2. Measure a master part through the real nudge band. Take the widest seating error your operators actually produce, add margin, and step a known width artifact through that whole band. Record width at every step.
3. Test the fixture with hands, not wishes. Several operators load and unload across a shift. If the fixture only holds height when a careful person loads it, the study has to say so.
4. Test the telecentric candidate across its range, not at one height. The claim to verify is width drift across the depth band you run, alongside field of view, working distance and magnification stated with test conditions.
5. Write the acceptance line before the demo. Something like: width error across the full seating band stays inside a stated fraction of the width tolerance. The fraction is your call as the buyer; the maker’s job is to beat it on your bench.
6. File both reports. The winning report becomes the station’s baseline record, which is what your next drift investigation, retrofit or audit will be compared against.

Questions That Come Up on the Line

Can Calibration Software Absorb the Nudge?

Not the part that matters. Distortion calibration removes the lens’s own geometric bending, which is real and worth fixing, but it is a static correction. A nudge changes the scale of the whole image, and the software cannot know from the image alone which part of the change came from the part’s height. Unless the system measures height independently, software corrections stop where seating variation begins. A vendor who proposes to “calibrate out” perspective error without touching either the fixture or the optics has moved the problem, not solved it.

Does Telecentric Mean Sharp at Any Distance?

No, and this is the misunderstanding that burns budgets. The telecentric range and the depth of field are different specifications. Magnification staying constant does not mean the image stays in focus; the focus still has to be set for the working distance, and contrast still falls as the part moves away from the focused plane. When you qualify a telecentric candidate, judge focus performance and scale stability as the two separate lines they are, and hold both to written criteria.

Conclusion: Buy the Decision With a Number

The nudge that moves a width reading is information, not an emergency. Convert it once into microns against your tolerance, and the choice between a fixture and a telecentric lens stops being an argument between habits. If the arithmetic says the fixture absorbs it, write the fixture’s height discipline into the maintenance plan and move on. If the tolerance is already consumed, choose the telecentric design and demand the drift curve across your real depth band, with conditions, before the purchase order moves. Either exit is respectable; buying optics to soothe a feeling is not. When the winning candidate is an FA lens rather than a gauging lens, the same rule about documented per-batch evidence applies across the FA C-mount machine vision lens family. Make the maker demonstrate on your station, keep the report, and the next operator who nudges a part will nudge nothing that matters.

Written by Baker Zhao, Founder at Lensmind Optics