HUB-CHAR_

We were about to specify hub motors for drive work and had no basis for choosing between six candidates other than published curves. So we bought all six, built a fixture, and measured them.

The question was narrow and answerable: how far off are the published numbers, and are they off consistently enough that we can apply a discount and move on.

A hysteresis brake against a 250 mm torque arm on a single-point load cell, with the motor under our own field-oriented control rather than the vendor controller, so the current limit is a number we set instead of one we infer.

The fixture was the mistake. A torque arm is cheap and simple and its first bending mode sat at 44 Hz, which is inside the band where a drive motor is interesting. An in-line rotary torque transducer would have cost more than all six motors together, which is exactly why we did not buy one, and exactly why the investigation ended.

Six motors logged over 148 runs. Kv came in within 4.1% of published across all six — vendors get Kv right. Peak torque came in 22% below published on average, with the worst at 11.8 N·m against a claimed 18. Run-to-run spread on a single motor was ±2.1%, so the measurement itself was stable at steady state.

We shelved it on 30 June for two reasons and we want both of them in writing. First, the torque arm resonates at 44 Hz, so nothing dynamic we logged is trustworthy, and rebuilding the dyno properly is a bigger project than the question deserved. Second, we bought one unit per part number, which means every gap above could be sample variation rather than an optimistic datasheet. The honest summary is that we learned to discount published peak torque by about a fifth and nothing more. Two of these motors went into BDR-26-04 anyway, chosen on the measured numbers.