The control of the home-made motor is pretty much like the control of any brushless DC motor (BLDC). The main hardware components on the electronics side are a NUCLEO-F401RE development board from ST, a motor control shield by Infineon and a basic comparator circuit I put together to interface to the Hall sensors.

STM32 Nucleo-64 development board

I selected this board because I could easily get hold of BLDC control code for it. It features an STM32F401RE MCU with 512 Kbytes of Flash memory and an 84 MHz CPU. The MCU contains all the peripherals I need. In particular, a powerful set of timers to send pulse width modulation (PWM) signals to the coils of the motor and measure its speed, plus external general-purpose input/output (GPIO) pins I use to read Hall sensor signals and have them produce interrupts.

IFX007T motor shield

This PCB features three IFX007T half-bridges. A half-bridge is half of what you need to make an H-bridge, which is what is typically used to control motors. In essence, a half-bridge is made of two MOSFET transistors connected in series. In my case, I connect the higher one to the supply voltage (9V most of the time) and the lower one to ground. The output of the half-bridge is the mid-point connection between the MOSFETs, so by controlling them with the STM32 CPU I can connect the output either to ground or to 9V.

During operation, I can drive a variable amount of current through a coil by sending a PWM signal to the higher transistor of the corresponding H-bridge and connecting the lower transistor (of the other half-bridge in the H-bridge) to ground. The duty cycle of the PWM signal sets the average current through the coil. The sense of the current depends on which half-bridge is used for, respectively, PWM and ground, so can be easily changed by the MCU. See the firmware description for more details.

Interfacing with the Hall sensors

I use the AH49E Linear Hall-effect IC from Diodes Incorporated for sensing the instantaneous position of the motor. This is needed by the control firmware. The chips are conveniently available in a TO-92S package which allowed me to solder their pins directly to flying wires and keep full freedom in their physical placement. There are three Hall sensors in the motor, interleaved at regular intervals with the three windings in the stator. This means an angular separation of 120° between Hall sensors and 60° between any Hall sensor and its closest coil either side.

A close-up view of the placement of the Hall sensors in the stator

The orientation of the Hall sensors is chosen to produce a negative-going signal whenever the north pole of a permanent magnet in the rotor goes past them. The output of the Hall sensor sits around half-way between ground and the positive supply when not exposed to any field (the Earth field is much weaker than the one produced by the magnets). In order to get a proper 0V or 3.3V signal I can send to the MCU, I condition the signals coming from the sensors using a simple LM339 comparator with a bit of hysteresis.

Picture of the comparator circuit (Todo: post full schematics)

This short video shows the signals I get in the oscilloscope as I turn the rotor manually, making each of the Hall sensors see alternate north and south poles:

Testing Hall sensor output (you can see the signal stretch in time as the rotor slows down)

Connection tables

They say a picture is sometimes worth more than a thousand words. Well, not in this case! As you can see in the picture below, it is quite hard to discern the connections in the unholy mess of wires:

The Nucleo STM32 board connected to the IFX007T motor shield

Here’s a connection table to make sense of the wiring. Please refer to the Nucleo board user manual and the motor shield user manual as appropriate, In particular, connectors in the motor shield do not have proper designators, so you really need to look at the user manual to make sense of the table below.

Pin on Nucleo board Pin on IFX007T shield
CN11 (any pin) GND
CN10 pin 23 (PA8) INU
CN10 pin 21 (PA9) INV
CN10 pin 33 (PA10) INW
CN10 pin 15 (PA7) INHU
CN7 pin 34 (PB0) INHV
CN10 pin 24 (PB1) INHW

In addition to these Nucleo-shield connections there are also three wires to connect the Hall sensors (actually the outputs of the sensor conditioning board) to the Nucleo board:

Pin on Nucleo board Sensor signal
CN10 pin 3 (PB8) U
CN10 pin 5 (PB9) V
CN10 pin 17 (PB6) W

Bear in mind the following naming convention for the Hall sensors: their name is the same as that of the coil they are facing in the stator. For example, sensor U is the one surrounded by coils V and W, and so on.

Bill of materials

I live in France. The following is a list of items I purchased and convenient places for me to buy them from. If you want to reproduce and build upon this work, hopefully you can easily figure out where best to buy from in your location.

Item Provider
Nucleo board Amazon
Motor shield Farnell
Hall sensors Farnell
Comparator Farnell

You will also need resistors, capacitors and jumper wires (female-female and some female-male for easily probing with the oscilloscope). I trust if you have read this far and are interested in building this, you will be able to find these commodity items easily!