T-Pain Is On A Boat: [Electric Boats] Motor Controller Modes

Tuesday, January 8, 2013

[Electric Boats] Motor Controller Modes

Max-min current control sounds good. But would require a second
voltage comparator that can handle the spikes and noise. I just
copied the 'fixed off period' mode from another chip. (Allegro
A3938/A3932 IIRC. But that was intended for <1HP and couldn't take
the power/switching spikes from a larger motor.)

As it is, I managed to make the IR2133 into a single chip BLDC motor
controller except for the quad XOR gate to reverse direction.

I believe you but I'm not sure why external diodes would be better...
to share the heat loading?

...or if they were Schottky diodes with a lower forward voltage drop.
Say, I rather like that idea. It might improve regenerative braking a
little?

(I'd more expect this sort of subject would have come up on a motor
controller list like osmc@yahoogroups.com or ?? You just never know
where a conversation will lead!)

Craig

>Thanks Craig, I'll have a look at your controller design.
>
>It's an interesting idea to use a self-oscillating current-mode drive
>for the motor. I hadn't thought of doing the power control this way,
>although the technique is widely used in dc-dc buck regulators. I might
>have used a hysteretic current mode control though, turning the current
>flowing in a stator winding off when it has reached the desired
>set-point, then turning it back on again when it has fallen by a certain
>amount. The inherent source-drain diodes in the MOSFETs provide a
>current path when all four of the MOSFETs are turned off, although I
>have found huge improvements in reliability at full power in class-D
>amplifiers if you use a suitable diode bypass route to prevent the
>internal MOSFET diodes from conducting.
>As you say, the switching frequency will be variable as the di/dt will
>be a function of the motor speed (when the current is increasing
>anyway).
>
>Regards,
>Chris.
>
>In message , Craig Carmichael
>craig=saers=com> writes
> >Um, er...
>>
>>I wasn't thinking of the PWM frequency, which is of course a good
>>point. But I'll surprise you and say it's around 300 Hz. Now I have
>>to explain this...
>>
>>My controllers actually use a dual system. First there's "direct
>>torque control", or as I call it "Current Ramp Modulation", "CRM".
>>(Torque being directly proportional to current.) When you apply
>>voltage to the motor coils, as inductors, the current starts at zero
>>and rises. If the motor is stopped with no back EMF, it rises very
>>quickly, but much more slowly with the motor at full speed. The CRM
>>turns off the coils when the current has ramped up to the maximum
>>level (or alternatively to the control setting below maximum). It
>>turns it off for a fixed period like 50uS, then turns it back on to
>>begin again. So the modulation will be fast at low motor speeds with
>>high torque, and slow at high motor speeds.
>>
>>This minimizes switching losses. It variably reduces switching
>>frequency to no more than necessary for conditions. The problem with
>>it as the main control is that without external feedback, unless
>>torque load rises and falls with RPM (as with, eg, a boat prop), the
>>motor will keep speeding up until it hits max or slowing until it
>>stops for small control adjustments.
>>
>>So in an irregular load like an EV car, PWM (external in present
>>version V2) modulates the CRM (set to max). But since the CRM is
>>controlling the current, the PWM doesn't worry about it, and can
>>switch at any low frequency above where the on-off pulses would
>>become noticeable.
>>
> >So it's actually my controllers that keep the switching frequencies low.
> >http://www.TurquoiseEnergy.com/hybridize/MotorControllerManual/MotorCont
> >rollerManual.html
>>
>>(BTW If the circuit is of interest - There's some component value
>>changes to the schematic C10 is double+ for longer 'fixed off time',
>>and on the layout a wire touches pin 10 of the header strip instead
>>of going around it.)
>>
>>Craig
>>
>
>