SH

Custom BLDC motor and three‑phase drive

It passed every static check, then stalled under load. The firmware assumed four pole pairs. The rotor has three.

When
January to March 2026
Team
Four of us. I worked on every part.
Drive
TMS320F28377D, open-loop SPWM
Rev 3 on the bench: the lilac 3D-printed stator wired to the inverter board, which is driven by the DSP board.
Rev 3 on the bench.

How the motor came together

Rev 1Rev 2Rev 3wouldn’t armshortedranArduinoPWM moduleTMS320F28377DC2000 DSPPWMESC30 Anot armedMOSFET invertercurrent01234arms at 3 Aisolated JTAGdebuggerSPWM, phase AFirmware assumes43pole pairsRotor has3 pole pairsloadlocked

  1. Rev 1: an Arduino and a 30 A ESC. The ESC wouldn’t arm below 3 A.
  2. Rev 2: a PWM module. It shorted and blew the MOSFETs.
  3. Rev 3: a TMS320F28377D running open-loop SPWM in C, debugged over isolated JTAG. It ran.
  4. Under load it fell out of step and stalled. The firmware assumed four pole pairs. The rotor has three.
  5. One constant, four to three. The field and the rotor lock.

Build

A 9-pole stator, hand-wound at 100 turns per coil and connected in delta. Six neodymium magnets make three pole pairs. The housing is 3D-printed PETG.

Drive

A TI TMS320F28377D dual-core DSP runs open-loop SPWM, written in Embedded C and debugged over isolated JTAG. It drives a three-phase MOSFET inverter with gate drive and current and thermal limiting.

Three revisions

  1. An Arduino and a 30 A ESC. The ESC wouldn’t arm below 3 A.
  2. A PWM module. It shorted and blew the MOSFETs.
  3. The DSP build. It ran.

The stall

Under load the motor stalled and oscillated. The cause was a pole-pair mismatch: the firmware assumed 4 and the motor has 3. The bug passed every static check.

Measured

Startup torque, rotor inertia, and a stall limit of about 3 A, set by the windings’ thermals.

Next

Hall sensors, then closed loop, SVPWM, and field-oriented control.