Behind every smooth-running gear set is a piece of steel whose buyers almost never see directly: the material inside the blank. Casting shrinkage, forging bursts, laps, inclusions and quench cracks do not appear on a finished gear drawing, yet any of them can end a driven shaft, a rolling mill pinion or a wind-turbine stage months before the calculated service life. Magnetic particle and penetrant testing catch what sits at or near the surface, but defects buried a few millimetres below the skin are invisible to both, and they are precisely the ones that fracture under fatigue load. Ultrasonic testing, or UT, is the common industrial answer for looking into the volume of a steel gear blank or a finished gear without cutting it apart. This guide explains how pulse-echo UT actually works, what kinds of internal flaws it can and cannot find, which standards buyers and shops rely on, how to read a UT report so you can tell a genuine pass from a rubber stamp, and what to write into your purchase order so the testing you pay for is the testing you actually receive.
Pulse-echo ultrasonic testing sends short bursts of high-frequency sound, typically between one and ten megahertz, from a piezoelectric probe into the part through a thin layer of couplant such as glycerin, gel or oil that squeezes out the air gap between the transducer face and the steel. Sound travels through uniform steel at a roughly constant velocity of about five thousand nine hundred metres per second, until it meets a boundary of different acoustic impedance. At a boundary, part of the energy reflects straight back to the probe and part continues onward. On the instrument screen, the first vertical pulse is the initial surface echo, often called the main bang, and the echo from the opposite face of the part, the back wall echo, arrives later and marks the full sound path. Any internal reflector, a pore, a crack or a slag stringer, sits between the probe and the back wall and produces its own echo at a position on the horizontal axis that corresponds directly to depth, because depth equals sound velocity multiplied by half the measured travel time. The taller the echo on the screen, and the more it compares with a reference reflector of known size such as a flat-bottomed hole, the larger the flaw is judged to be. Higher probe frequencies give sharper depth resolution and smaller-flaw sensitivity, but they attenuate faster in coarse-grained cast or forged steel, so thick or grainy sections often drop to two or two-and-a-quarter megahertz, while fine finished areas can take five or more. Manually scanning the probe along the tooth face, bore, hub and web in a regular pattern, and comparing each echo against a calibrated distance-amplitude curve, is how a technician decides whether a response is a genuine planar defect or simply geometry such as a keyway, a fillet or a change in cross-section.
UT targets the volume of the gear, and the candidate defects depend on how the blank was made. In cast gear blanks, the dangerous internal features are shrinkage cavities, micro-shrinkage porosity, gas holes and non-metallic inclusions, clusters of slag or sand that break up the sound path and scatter back broad, fuzzy echoes. In forged blanks, UT looks for forging cracks, bursts, laps and folds, typically formed at ends or in heavy sections where deformation was uneven, and for stringers of non-metallic inclusion aligned with the grain flow. After heat treatment, UT is used again, because quenching can introduce hardening cracks that start deep in the web or at the tooth root and never break the surface until the gear already fails in service. The reason the non-destructive method is chosen by location is straightforward. Magnetic particle testing and liquid penetrant testing react to discontinuities at or immediately below the surface, so they are the correct tools for grinding cracks, quench cracks that opened to the skin, machining laps and surface forging folds. They are cheap, fast and sensitive on the surface, but a void or a crack buried forty millimetres inside a hub leaves no trace whatsoever. Radiography sees through volume too, but it is poor at catching planar, oriented defects such as a tight fatigue crack running parallel to the beam. UT is therefore the standard volumetric method for heavy gears: it reaches deep, is sensitive to planar reflectors, and can be performed on the shop floor without film, radiation safety zones or long processing times.
UT results are only as good as the standard they are judged against, and the first layer of control is the person doing the scanning. Industrial practice normally requires operators to hold a certification such as ISO 9712, or the equivalent national scheme, in ultrasonic testing at an appropriate level, with level two able to evaluate and report indications and level three responsible for procedure approval and interpretation. For forged steel gears, buyers commonly reference acceptance frameworks drawn from standards such as EN 10228-3 for steel forgings, ASTM A388 for heavy steel forgings, or ISO 4992-2 for steel castings and forgings used in gear units, depending on industry and end market. For cast blanks, the relevant family is often EN 12680 part three for castings used in machinery such as gears, together with the material clauses that sit inside the gear rating standards. Within each standard, the part is assigned a quality class or acceptance grade, from loose to strict, and indications are evaluated against calibrated reference curves such as the distance-amplitude curve, which compensates for the natural drop in echo height as a reflector sits deeper in the steel. An indication is described by its equivalent size, expressed as the diameter of a flat-bottomed hole that would give the same echo height, and by its indicated length when a crack-like planar reflector is swept along its long axis. Anything above the reference curve, or longer than the permitted indicated length, moves into the reject zone for that class; everything below it is acceptable for that service.
A UT report that you cannot interpret is, for purchasing purposes, barely better than no report at all. The first thing to check is scope: which faces, bores, hubs, webs and tooth regions were scanned, with which probe and at which frequency, and how much of the surface was actually covered. A report that says only that the ultrasonic test was passed, without listing the scan areas, is asking you to take the shop's word on coverage, and coverage is exactly where corners are cut. Next, any recorded indication should carry a depth position measured from the scanned surface and an axial or circumferential position around the gear, so you can tell whether the reflector sits in a stressed root region, a lightly loaded web or a non-critical rim area. The report should state the equivalent size and the indicated length of each indication, the acceptance class applied, and a clear conclusion of accept or reject against that class. Credibility comes from consistency: the operator's certification number and level, the instrument model and probe serial numbers, the calibration date, the reference block used, and the date and time of the scan should all be present, and the report should match the heat number or serial number on the gear itself. If a report shows no indications at all on a casting that visibly has riser cuts, or quotes a quality class that never appears in your drawing, treat it as a document to question rather than as proof of sound material.
Putting UT into your inquiry, instead of discovering it at final inspection, is how you keep the cost and delivery penalty predictable. State the standard and the acceptance class explicitly: reference EN 10228-3, ASTM A388 or whichever framework your end customer recognises, and name the grade you require rather than asking for UT as if one level fits all. State the timing of the test: most buyers ask for UT on the rough forged or cast blank before any expensive machining is committed, then again after heat treatment and ideally before finish grinding, so a quench crack is found before it becomes scrap at the last operation. State the coverage: a one-hundred-percent scan of the hub, web and tooth area, or an agreed sample plan for the production lot, and specify which surfaces must be scanned. State whether the report must include dimensional location, equivalent size and indicated length for every recordable indication. State whether the test must be witnessed by an independent third party such as a classification society, an owner's inspector or an accredited laboratory, and whether you want the raw echo printout attached to the report. Finally, ask how UT affects price and lead time: a full volumetric scan with third-party witness on a heavy gear blank adds both inspection hours and calendar time, and booking it in the inquiry lets the shop plan the scanning window, the coupling-surface preparation and the calibration blocks before the forging even arrives.
Ultrasonic testing does not replace dimensional inspection, hardness tests or the visual surface checks that go into a gear acceptance packet. It adds the one piece of information those methods cannot give you: a statement about the internal volume of the steel you are buying. When the method, the standard, the operator qualification, the coverage and the reported indications are all written down, cross-checked and tied to the serial number on the part, UT becomes contract-grade evidence rather than a vague promise of good material. If you are buying gears for rolling mills, mining drives, marine gears, energy equipment or any service where a burst hub or a hidden quench crack would stop a production line, specifying pulse-echo UT with a named acceptance class in the inquiry is a small cost against the downtime that one buried defect can cause.