I’m a quality compliance manager at a precision drive manufacturer. Last year I reviewed more than 200 product verification files before they were published to our catalog. I rejected about 9% of first submissions, mostly for missing thermal or load-cycle data. That gives me a particular window into why motors fail in the field — and why so many failures have nothing to do with the motor.

The motor that wasn’t at fault

Not long ago, a customer reported that a maxon liftgate motor was tripping its thermal switch after a few minutes of work. They had already replaced one motor. Our lab tests said the first return was good. The second one passed as well. It took an on-site data logger to show what was actually happening: the hydraulic fluid was cold, the supply voltage dipped, and the drive was asking for more torque than the motor’s continuous rating allowed. The motor wasn’t defective. It was selected without a complete load profile.

This isn’t a rare exception. In one internal analysis of returned servo motors, 43% of the units we tested showed no measurable defect under datasheet conditions. That number made me uncomfortable because it means thousands of service hours are spent replacing parts that weren’t broken. If you replace a motor, the motor is gone. But if the root cause was a misaligned mechanical axis, an unsupported cable shield, or a VFD setup, the new motor will eventually fail too.

Spec sheets are conditions, not promises

The deeper problem is not component quality. It’s the way engineers weight datasheets against system context. A datasheet is a set of measurements taken under controlled conditions: nominal voltage, room temperature, free-air cooling, new bearings. Real applications are almost never those conditions. As soon as the environment differs, the curve shifts. The motor isn’t lying; neither is the datasheet. But neither one knows what your machine is doing.

There is also a quieter assumption: that two motors with the same voltage, speed, and power are equivalent. From outside, they look similar. What you don’t see is the consistency inside a product. Magnet grade variation, winding tension, varnish quality, thermal test limits, supplier change controls — these small differences show up as outliers. A legitimate motor manufacturer works hard to keep that distribution narrow. At maxon-motor, we don’t test just a first article and assume the rest will follow. The whole production line is set up to keep unit-to-unit variation under control.

I know that “quality” sounds vague if you read about it in a catalog. But in my work, it’s concrete. A motor can meet datasheet specs and still fail in your machine because the batch-to-batch consistency isn’t there. The best way to see it is to ask for capability data — Cpk, field return reports, thermal test results — before you select a supplier.

The costs of skipping diagnosis

When a motor is replaced without diagnosis, the real problem stays invisible. A packaging machine lost position repeatedly on a high-speed line. The customer spent about $22,000 in engineering time and downtime before anyone looked at the encoder cable running beside a VFD power cable. The servo motor was fine; the position feedback was corrupted by electrical noise. Replacing the motor didn’t help because the signal path was the issue.

Read enough servo motor news and you see the same story over and over: high torque density, low inertia, clever algorithms. But no machine is built from the sensor alone. I’ve yet to see a headline that says “loose ground wire ruins throughput.” That’s the reality in the field.

Automotive repair does this differently. A search for “volvo timing belt replacement” describes a planned maintenance event: you replace the belt because time and mileage tell you to. But a mechanic would not replace a timing belt to diagnose a misfire. They would check spark, fuel, compression, and alignment first. In machine troubleshooting, we often skip the diagnostics because the motor is the easiest component to swap. That substitution is expensive. A 1% failure increase on a 50,000-unit annual program is 500 extra service events. Even at a conservative $150 per visit, that’s $75,000 in hidden cost — before downtime and lost production.

What helps more than a different motor

I don’t want this to sound like an excuse for bad motors. But if you are the person responsible for a machine that keeps losing motors, the next step is not necessarily a new supplier. It’s a more complete application profile.

First, write down the duty cycle. Not just peak torque, but the sequence of torque steps, speed, time, ambient temperature, enclosure heat, and voltage variation. This single document eliminates most false accusations I see.

Second, include the drive. If you are searching “how VFD control motor speed,” that’s a sign you know the topic matters. VFDs produce voltage waveforms that can stress motor insulation, especially when long cables amplify the reflected-wave effect. A motor can be electrically healthy but incompatible with an older drive or a poorly grounded installation.

Third, test the motor as it will be installed. Use a current probe and a temperature sensor; run the machine under the worst realistic load. That data is more valuable than any catalogue claim.

The same principle applies to our mobility products. When we released the maxon air s motor e-mtb, the validation wasn’t just the motor on a dynamometer. It was the motor plus controller plus battery under hills, heat, and rider patterns. If we had published only a bare-motor datasheet, people would see over-temperature warnings in summer and assume the motor was weak. It isn’t. But every component has a boundary.

One caveat: I work mainly in precision drive systems, not household motors or high-inertia industrial drives. So the exact thresholds on your motor will differ. The way to find them is the same: measure, specify, and validate before you change parts.

When I sign off on a product, I try to ensure the datasheet is accurate enough that a customer can make a smart decision. That’s the job. But a datasheet cannot decide for you. It can only tell you where the motor is safe. Your real job is to decide where your system lives — and buy based on that, not on the promise of one magic part.