How to choose a motor test bench: six decisions that matter

1. Define the purpose before the specification

An R&D bench and an end-of-line (EOL) bench are different machines. R&D benches are built for measurement accuracy, operating-point coverage and complete data capture, and a single run may last tens of hours. EOL benches are built for cycle time, repeatability and consistent pass/fail judgement, typically 30 to 90 seconds per unit. Running production on an R&D bench wastes takt time; running development on an EOL bench usually leaves you short of measurement channels.

Write down three things at the start of the selection: the power and speed range of the motors under test, the report items you are contractually required to produce, and the machine types you expect to add in the next three years. The third one determines how much headroom the bench needs.

2. Speed and torque are not independent

A dynamometer's rated torque generally applies only below base speed; above it the machine enters the constant-power region and available torque falls as speed rises. What matters at selection time is whether the dynamometer's capability curve envelopes the full torque-speed characteristic of the motor under test — not whether two peak numbers look large enough.

For high-speed machines — compressor motors, flywheel storage motors or aerospace motors above roughly 20,000 RPM — shaft critical speeds, coupling balancing grade and burst-containment guarding become the dominant engineering problems. LANDI has delivered dynamometer benches up to 33,000 RPM, and on those systems the mechanical design is usually harder than the electrical design.

3. Accuracy is a chain, not a sensor

Efficiency uncertainty comes from the torque transducer, the speed measurement, the power analyser, temperature compensation and sampling synchronisation together. A ±0.1% FS torque transducer installed with poor shaft alignment, or sampled out of phase with the electrical power measurement, can easily give an efficiency error several times larger than the sensor rating suggests.

The practical requirement is to ask the supplier for a system-level efficiency uncertainty budget rather than a list of individual sensor accuracies, and to verify repeatability at acceptance using a reference motor with known efficiency.

4. Cooling decides whether endurance tests finish

Temperature-rise tests, endurance runs and efficiency MAPs all require stable, controlled coolant inlet temperature and flow. Oil-cooled machines add viscosity-temperature behaviour and filtration requirements. When the cooling system is undersized, long runs abort on over-temperature and the data set is unusable.

5. Software outlives the hardware

Mechanical parts can be replaced; the test software determines the next decade of daily use — whether you can define your own sequences, whether reports are generated automatically, and whether results can be archived centrally. LANDI's LDEV-X platform generates test programs from natural-language descriptions, logs up to 1000 channels at 1 kHz, and connects to the LD-LMS laboratory management system for central data handling.

6. Put the acceptance criteria in the contract

Specify in the contract: calibration certificate requirements, the repeatability test method and its acceptance threshold, the duration of the continuous-run demonstration, the software function list and the training scope. Adding these after delivery is expensive.