Controller Topology

[Alan B]

I was looking at some different configurations for BLDC motor controllers and wondering if a different configuration would be an improvement in a useful way on ebike applications. Here's one short paper that shows several topologies:

http://www.ijitee.org/attachments/File/v2i1/A0355112112.pdf

It looks to me like a motor controller made with a buck type current source feeding the commutating inverter (Buck-CSI config) would have some useful advantages, especially for a homebrew controller. The PWM is confined to the buck stage and it can have a much higher quality and lower loss coil (than the ones in the motor). It can be operated at a much higher frequency which makes the input capacitors much smaller, and the coil much smaller. There is a single place to measure phase current. Shoot through events are current limited on the commutation FETs, and they can be switched at the commutation rate which potentially simplifies their drive. Either Trapezoidal or Sine can be implemented, though Sine puts more requirements on commutation FET switching. Phase current throttle feedback becomes much simpler. dV/dT on motor windings goes away as motor sees only DC, and a lot of the noise and EMI go away, what is left is confined to a small volume inside the controller. This configuration would also drive a low inductance motor nicely.

 

[dnmun]

i am gonna read more but this jumped out at me in the middle of the first paragraph. a typo.

>>>
"Based on the combination of these three hall sensor signals, the exact sequence of computation can be determined."

spellcheck sees commutation as valid as computation

 

[Alan B]

Technical papers are not always written with a high level of language skill.

 

[Alan B]

One downside with this approach is the buck stage must handle the full battery and phase currents all the time, instead of spreading this heat across three devices. Of course the good news here is that cooling can be concentrated on a single semiconductor or pair (the FET and diode, or 2 FETs if synchronous buck regulation is used).

This technique also simplifies the software of the controller, separating the high speed PWM from the commutation. Not a big deal perhaps, but makes it easier for a DIY program to be developed and tested. Also the micro doesn't need six PWM outputs, so a simpler smaller micro can be used. If a diode buck regulator is used then only one PWM output is needed. The commutation will require six binary outputs but those can be simple output bits.

The bottom line is that a standard Arduino could do this type of a motor controller. It would take 7 FETs and their drivers, a Schottky diode, a coil, and two current transducers. If two FETs were paralleled in the buck regulator (8 FETs total) the performance would improve as the commutation FETs are working very hard, perhaps more than two bucking FETs would turn out to be a good balance.

Anyway, it is interesting to think about different topologies of motor controllers. Some have buck/boost converters in front of the inverter which would facilitate using low voltage batteries with higher turn count motors and not being limited in speed and not needing the high current and heavy copper.

 

[Alan B]

I was reviewing some threads here on the Colossus low inductance motor, and I see I was thinking about buck regulator type controllers in 2011. Not a new idea, even for me.

It would be a great way to handle low inductance motors.