BLDC Motor Drive waveforms

Harry, I know you know more about this than I do, but here's what I know:

One winding or one terminal? I suppose if it's a 'Y' wound motor then the answers are the same, but if it's delta wound...

I would expect that the current into a terminal, if plotted against time, would go (use mono-spaced font):

time: 0 1 2 3 4 5 6... i : zero A zero -A repeat

with the "A" and "-A" meaning the same waveform, but -A the negative of A. Presumably "A" should be convex down.

The half-bridge on any given terminal should be open for 1/3 of the time, in chunks that are 1/6 of a pole time, as I show above.

Certainly the (few) DC brushless (i.e. trapezoidal back-EMF) drive projects that I've been involved in have all done it this way, and that's what you see in the literature.

In fact, you kinda hafta do it that way if it's sensorless: you need that open-circuit interval to sense the voltage and hence the back-EMF.

If it's someone else's software, then by definition it's not done nearly as well as you would have done, and it's probably buggy as hell. That's just a universal rule. So there's your problem.

Look for the guy who actually wrote the code, and I bet he's only got five fingers on his left hand*. That would account for why he got five segments right and one wrong -- just tape a hot dog onto the edge of his left hand next to his pinky, and he'll do better next time.

Or suggest he take off at least one shoe.

• Right hand if he's** a lefty.

Not quite clear how this was measured, but maybe the flat portion is where that half bridge output is open circuit, and therefore zero current. Just the shape, alone, tells me the blue trace is current, the yellow seems to be voltage.

One thing you could try is to have all three half bridges always conducting, either high or low, rather than have two conducting and one O.C.

Jon

It doesn't look to me like the problem is that the half bridges are or are not conducting -- it looks to me like the problem is that the sequencing isn't right.

Tim, you are the man! The sequence is fudged up. It should be at any driver output; 2/6 PWM, 1/6 float, 2/6 low to sink current, 1/6 open, then repeat. Mine has 1/6 low, 2/6 PWM, 1/6 float, 2/6 low. That is 3/6 low and only 1/6 float. What a POS! Thanks Tim, Harry

Here is a pic of current in one of the three phases from my crack software weenies.

This trapezoidal waveform has ripple current on the positive and negative peaks causing torque chatter. The SW boys say that is normal. (Sorry for the current offset in the pic. poor tech) What say anybody?

Thanks Harry

Are these the same guys that just had to tape a hot dog to their left hands so they'd be able to count off all six stages of BLDC drive?

I have no direct experience, but the current has to depend on the driving voltage and the back EMF of the motor. Is it really trapezoidal, or are you driving a motor with a sine-shaped back-EMF?

Can you hook your scope up to all three phases of the motor (with the driver unhooked) and give the shaft a spin? See what the waveforms look like?