EVALSTDRIVE 101 and Nucleo u5a5, motor not starting with full moment without a push

[risqy]

Setup

• MCU: STM32U575, MCSDK v6.4.1-Full, FOC, hall-sensored, MCM_SPEED_MODE, no rev-up
• Sensing three-shunt (R3_1), external amplifiers (no internal OPAMP)
• Power: 24 V bus, RSHUNT 5 mΩ, gain 6.5 (→ ~50 A full scale)
• Motor: 57BLR70-24-02 BLDC, 2 pole pairs, 120° halls

Problem
The motor does not start and produces inconsistent/no torque at any operating point.

What I have already verified

1. Full HALL_PHASE_SHIFT sweep 0–360° (10° steps): no start at any value. If only the offset were wrong, a start should occur somewhere in the correct ~60° window — it never does.
2. Hall sensors healthy (checked live in Motor Pilot): clean 6-state sequence, ~60° symmetric, no invalid 000/111, 5 V levels OK.
3. Current offset calibration OK: PhaseA/B/COffset ≈ 36350/36361/36518 (consistent), offsetCalibStatus = true, SWerror = 0.
4. Key symptom: Motor Pilot reports ~100 W, while the bench PSU shows ~0 W actually drawn. Occasionally a Speed Feedback Error fault appears (M1_SS_MEAS_ERRORS_BEFORE_FAULTS = 3).
5. Shunt jumpers gU/gV/gW closed; some solder bridges on the board were removed.

Question
The ~100 W (firmware) vs ~0 W (real) mismatch suggests the measured phase currents don't match reality, so the current loop runs effectively open — which would explain why no HALL_PHASE_SHIFT produces torque. Before I keep chasing the hall offset:

• Does the "100 W displayed / 0 W real" pattern point to a current-sense/scaling path issue or to repeated fault shutdown?
• Any known cause for hall-sensored FOC failing to start at every phase shift while halls and offset calibration look healthy (e.g. hall-vs-field direction inversion)?

Thanks!

[GMA]

Hello @risqy,

Use the STO-PLL sensorless estimator as the main speed sensor and the HALL sensor as the auxiliary speed sensor. Verify that the electrical angles of the HALL sensor and STO-PLL are aligned.
An incorrect HALL Placement electrical angle causes a phase difference.
Incorrect motor-phase or HALL H1/H2/H3 connections cause rotation-sense estimation errors.