From a command to real currents.
Change a setting and recalculate the existing C++ model. Trace FOC through the six-switch bridge, explore heating, or build a motor-shaft torque–RPM map.
Synthetic SIL, not a calibrated motor. These isolated experiments do not control the live joint. Switching replay is inspection of a completed calculation.
One signal chain, one selected instant
The loop produces three phase voltages using six switches. 3PWM/6PWM describe driver inputs, not three versus six motor phases.
Command, true current and observed current
Inside one switching period
Where smoothing happens: winding inductance shapes physical current. The sensor has an analog low-pass filter; this implementation then adds capture noise and ADC quantization. A DAC may set a comparator threshold; it does not smooth the motor power current. Lower analog bandwidth does not remove noise added after that filter.
Heat accumulates over physical seconds
Prescribed current for 600 s, not an inverter-controlled heating run. The same winding/case thermal equations and R(T) are recalculated. FET heat is a conduction proxy; protection is bypassed in this diagnostic.
Compare with the separately labeled preheated derating experiment →
Torque–RPM and qualified efficiency
28 independent, cold-start averaged-inverter runs: 2 / 5 / 10 / 15 A at 0–3600 motor RPM. No gearbox. Last 20 ms of each 120 ms run supply time-weighted means.
Motor-shaft torque
DC-terminal → motor-shaft efficiency
Gaps mean unqualified, not zero efficiency. A point needs positive input/output power, current tracking, no fault, low stored-energy rate and 0 < η ≤ 1. Voltage saturation, copper/inverter/friction losses and numerical residual are retained in CSV. Omitted iron, switching-energy and auxiliary losses make this unsuitable as a product rating or continuous-torque map.
Inspect each calculated operating point
| RPM | Iq* [A] | Shaft [N·m] | DC [W] | η [%] | Saturation [%] | Fault |
|---|