The original stir plate used a shaded-pole motor, to which was attached a steel bar with two magnets. Above the magnets was a loop of aluminum - the resulting eddy currents created a load with force proportional to the square of motor speed -- in effect a simple governor.
The system was very clever, but unfortunately it was also bulky, inefficient, and thanks to rust accumulated over the years, aesthetically ugly. I needed the space in the housing where the power rheostat was, and I needed to control the motor with an Arduino.
As luck would have it, my Mom's old VCR was sitting in my workshop, waiting to be taken to e-waste collection day. There's an incredible amount of precision in these machines we used to routinely buy for a couple hundred dollars and then throw away -- including not one but two pancake-style brushless DC motors -- one for the capstan and one for the head drum.
By this point in the production of VCRs, these motors were coming as units with their own controller/driver IC on the board. Reading the hall-effect sensors and driving the motor coils was all handled on-chip. So the only connections required were power in, two kinds of tachometer outputs, and a torque signal input. My (mom's) drum motor was a two-phase "Sankyo" motor, as pictured. Although the datasheet for the exact chip wasn't available, by poking around with some similar datasheets I was able to reverse-engineer the pinout and drive signal format and control the motor's speed with the Arduino.
With that done, it was a relatively simple matter to remove the head drum and fabricate a new rotor to fit the motor's brass hub, using rare-earth magnets for smaller size and improved appearance over the stir plate's cube-shaped ceramic blocks. However, I left the eight interconnection points for the heads intact as a reminder of the motor's original purpose.