What Is an Optical Comparator?

An optical comparator, also called a profile projector or shadowgraph, is a non-contact metrology instrument that shines light past a part, magnifies its silhouette through a lens, and projects that shadow onto a glass screen where an operator measures the outline against a scale or an overlay chart.

The comparator is one of the oldest tools still bolted to a modern inspection bench, and it survives for a plain reason: for small 2D parts it is fast, it never touches the work, and anyone can read a shadow. This guide walks how the light path works, what magnification buys you, how to measure a part against an overlay step by step, and the honest line between the jobs a comparator owns and the jobs that belong on a coordinate measuring machine.

What is an optical comparator?

An optical comparator is a projection instrument that turns a physical part into a magnified shadow you can measure. Light passes the workpiece, runs through a telecentric lens, bounces off one or more mirrors, and lands on a translucent screen as a sharp, enlarged silhouette. Because the part blocks the light rather than being probed by it, the only thing making contact is the beam. That is what the words profile projector and shadowgraph describe: you are inspecting an outline, not feeling a surface.

It belongs to the same family as the other bench tools a plant reaches for in a dimensional check. Where a caliper or a micrometer reads one dimension at a time and a go/no-go gauge only tells you pass or fail, a comparator shows the whole profile at once and lets you measure any point on it. It sits alongside the rest of the gauge types used in manufacturing as the answer for small, detailed 2D features: threads, gear teeth, stamped contours, molded radii, the tip of a cutting tool.

How does an optical comparator work?

It works by backlighting the part and enlarging the shadow. A lamp and condenser throw a steady beam through the workpiece, which sits on a stage that moves in X and Y on precision slides. A telecentric projection lens keeps the magnification constant even if the part shifts slightly toward or away from the lens, so the edges stay true instead of ballooning. Mirrors fold that long optical path into a cabinet, and the image lands on a ground-glass screen, usually with a cross-hair reticle and angular grid etched into it.

You read a dimension one of two ways. To measure a length, you line the screen cross-hair up with one edge of the shadow, zero the digital readout on the stage, then drive the stage until the cross-hair meets the other edge. The travel the readout shows is the real dimension, because the stage moves the part, not the magnified image. To check a shape, you compare the shadow against a reference laid over the screen: a scale, an angular protractor built into the reticle, or an overlay chart. The magnification only makes the shadow easier to see; the measurement itself comes from the stage travel or the overlay, so accuracy does not depend on the operator judging distance by eye.

What can an optical comparator measure, and at what magnification?

A comparator measures anything you can cast as a clean 2D outline: outside dimensions, angles, radii, thread pitch and form, gear tooth profiles, stamped and molded contours, the geometry of a drill point or insert. Magnification lenses typically range from 10x to 100x, and some benches go higher; a 20x lens turns a 0.5 mm feature into a 10 mm shadow on the screen, which is why comparators dominate small-part work where a caliper simply cannot resolve the detail. This 2D profile focus is what separates a comparator from the broader task of dimensional inspection which may also demand depths and 3D relationships.

The limits matter as much as the reach. A comparator sees a projected outline, so it cannot measure a blind bore, a feature hidden behind another feature, or true position of a hole relative to datums on different planes. It reads best on parts small enough to fit the stage and thin enough that the silhouette is a real edge rather than a fat, ambiguous shadow. Push it past those limits and the measurement gets soft.

How do you measure a part against an overlay chart?

The overlay chart is the comparator move that a CMM cannot copy quickly: you lay a transparent drawing of the part, with tolerance bands drawn as two outlines, over the screen and check whether the shadow stays inside the band. It is a go/no-go call on a whole profile at once. Here is the routine an inspector follows.

1. Confirm the magnification. The overlay is drawn for one magnification, say 10x. Fit the matching lens and verify it against the calibration standard before you trust a single reading.
2. Set the datum. Mount the part on the stage and align its reference edge to the screen cross-hair, so the shadow and the overlay share the same origin. A shifted datum turns a good part into a false reject.
3. Lay on the overlay. Clip the transparent chart to the screen. The chart shows the nominal profile flanked by an upper and lower tolerance line, the two outlines that define the pass band.
4. Compare the shadow to the band. Drive the stage so the shadow sweeps across the chart. Every point of the outline should ride between the two tolerance lines.
5. Judge the profile. If the shadow stays inside the band the whole way, the part passes. Where it crosses a line, you have found the out-of-tolerance feature, and its location on the chart tells you which one.
6. Measure the miss. For a feature that fails, switch to the cross-hair and stage travel to record the actual value, so the reject has a number, not just a verdict.
7. Log the result. Record the reading against the characteristic on the inspection sheet, the same discipline a first article inspection demands.

Where does an optical comparator still beat a CMM?

It wins wherever the part is small, thin, or soft, and the feature is a 2D profile. A rubber gasket or a thin stamping will deflect under even a light touch probe, quietly biasing a CMM reading, while a comparator measures the undisturbed shadow. A run of molded parts checked against an overlay is a pass/fail sweep an operator does in seconds, no programming. A cutting-tool grinder can drop an insert on the stage and read the point geometry against a chart faster than a CMM boots its routine.

None of that makes the comparator a CMM replacement. The moment you need a bore depth, true position across datums on different planes, or a full 3D surface, the coordinate machine is the right tool and the comparator is the wrong one. Choosing between them is a CMM inspection question of what geometry the characteristic actually lives in, not which tool is newer. Whatever you pick, the gauge itself has to be capable, which a measurement system analysis proves before you trust its numbers on a control chart or a report.

By the numbers: measurement standards behind the shadow

A comparator reading is only as trustworthy as the metrology discipline behind it. The NIST/SEMATECH Engineering Statistics Handbook lays out measurement process characterization, the practice of quantifying a measurement system so its error is known rather than assumed (NIST/SEMATECH, Measurement Process Characterization). The tolerances a comparator checks a profile against are defined by the geometric dimensioning and tolerancing rules in ASME Y14.5, the standard that turns a drawing note into a measurable pass band (ASME codes and standards). Both point at the same truth: the number on the screen means nothing until the instrument is calibrated and the tolerance is defined, which is why a comparator lives inside a formal calibration program and not on its own.

How do you keep an optical comparator accurate?

You calibrate it, and you control the two things that quietly ruin readings: magnification and datum. Verify the lens magnification against a certified stage micrometer on a set interval, because a lens that reads 10.2x instead of 10x throws every dimension off by two percent. Check the stage travel and the screen reticle against known standards. Keep the lamp intensity and focus consistent, since a fuzzy shadow is a fuzzy edge and a fuzzy edge is an argument, not a measurement.

The operator side matters just as much. Fixture the part the same way every time so the datum does not wander, keep the glass and lenses clean, and record readings against the specific characteristic so a drifting dimension shows up as a trend, not a surprise. A comparator that is calibrated but loaded carelessly still generates false rejects and missed defects, and both cost more than the check.

Where does the optical comparator fit in the quality system?

The comparator is a data source, and its value depends on where those readings go. On its own it produces a verdict at a bench. Wired into the quality system, those readings become the dimensional evidence a customer wants to see: the profile checks that back a first article inspection the layout results that feed a PPAP submission the point measurements that populate a control chart. A comparator is often exactly the gauge that generates the dimensional results element of a PPAP package, which is why keeping its calibration current is not busywork.

The weak link is almost never the instrument. It is what happens to the reading after the operator writes it down. When comparator results live on a paper inspection sheet, a drifting dimension hides until the month-end tally, and by then the scrap is already made. When the reading is captured at the bench and tied to the part, the tool, and the characteristic, a trend surfaces during the shift that produced it. That live capture is what Harmony puts at the point of inspection through station-level data capture turning a bench verdict into a signal the floor can act on before the next skid ships. CLS made exactly that move, from measurements found the next morning to measurements visible while the run is still going. No rip-and-replace.