BLDC hardware emulation
helge wrote 04/30/2018 at 16:25 • 0 pointsSomething that bothers me for quite a while now: why is there no small / cheap hardware emulator for BLDC motors? I've seen contraptions where two motors are coupled and one is actively driven whereas the other just has a resistive load hooked up to it or a rectifier at best.
There's really nothing wrong with connecting BLDC driver chips back-to-back with inductors in between. Here it's being done to benchmark a large 3 phase inverter but the principle is pretty much the same.
Allegro makes an array of sensorless BLDC chips with sinusoidal and trapezoidal output modes and different options to set torque and nominal rotation speed.
https://www.allegromicro.com/de-DE/Products/Motor-Driver-And-Interface-ICs/BLDC-Drivers.aspx
All that's missing is a way to scale the sensed average current to e.g. a VCO control voltage and fudge around with loop stability to match the host BLDC controller.
At the end of the day I assume noone ever bothered to look into it because... I mean... why would you?
Unless you need to emulate a BLDC motor for arbitrary controller chips of a host device (DLP beamer :D) to be hacked...
ps. candidates for sensorless BLDC controllers:
Texas Instruments DRV10983
Allegro A4942, A4964 to name a few
ON Semi LB11683H
Owo JY02A ... just joking
Discussions
Become a Hackaday.io Member
Create an account to leave a comment. Already have an account? Log In.
You can find maybe a solution here
https://avl-set.com/en/products/lv-power-hil/
Are you sure? yes | no
Interesting. Also thinking of GaN-based active power filters as a starting point.
Are you sure? yes | no
Assume a BLDC driver chip is connected to a motor through separated filter inductors. It would act as a voltage source and drive currents. There shouldn't even be a requirement for fast dI/dt so as long as an inverter with sufficient control loop bandwidth connected to the other side of the filter can establish a time dependent current-voltage relationship this "load" could be made indistinguishable from a physical motor.
Then there are hardware in the loop - modules like this one:
https://www.opal-rt.com/simulator-platform-op4510/
which has power amplifiers to do just that. But it's totally overkill to trick a simple BLDC controller (and $20k price tag is ridiculous). For me it feels like a three phase bridge with small filter inductors should be all it takes to circulate reactive power and make the target BLDC chip happy. Maybe it's necessary to add DCDC converters to extend the emulator supply rails beyond the target rails.
Seems like I'm missing a bit of theoretical background there. Still it feels much like simple and stupid controller chips (that don't rely on sinusoidal back-EMF) should just take a blocky current or voltage waveform and roll with it.
And so the war continues.
Are you sure? yes | no
I'm calling it Low-Power-hardware-In-The-Loop from now on :D
Are you sure? yes | no
Presumably it isn't done because to be useful it would need variable inductance, and that is a harder than just calculating the motor behavior by hand.
If you're building a controller, you can probably model the motor as inductors and resistors and get a decent result. There isn't really much variance in motor behavior so you can just test a real one.
That said, it does sound interesting! Maybe it could use tuned tank circuits to create various levels of inductance, like in an LLC resonant converter. Seems like that and an active circuit to set the impedance and you're there.
The firmware to calculate the frequencies for the LLC circuit would be really useful for building DC-DC converters, so it might really be worth writing.
You'd need probably a dozen inductors, though.
Are you sure? yes | no