Need Help!- Sinusoidal pwm signal for brushless ac servo

[stev]

If I use EMC2 and Mesa Anything I/O card, can I get sinusoidal pwm signal for a H-bridge amplifier that will drive brusless AC servo?

[PCW_MESA]

Originally Posted by stev If I use EMC2 and Mesa Anything I/O card, can I get sinusoidal pwm signal for a H-bridge amplifier that will drive brusless AC servo?

What hardware do you wish to drive?

Driving a AC synchronous servo motor requires at least a 3 phase bridge, some way to commutate the motor (either hall and quadature encoder signals or better still an absolute encoder)
Vector drive requires current feedback so you need 2 A-D channels as well.

It is possible to do modeled voltage control without the current feedback and do pretty well (plus have more inherent damping than the torque control vector mode and a quieter drive) but this requres a lot of per motor tuning work.

That said, EMC has some of the pieces of 3 phase drive already available as HAL components and the HostMot2 firmware has a 3 phase bridge PWM output module (designed to directly drive a 3 phase IGBT bridge through isolators). Whats missing is Hostmot2 driver support for the 3 phase PWM firmware module, and someone brave enough to try it out...

[stev]

Thank you for your reply!

It will be a 3 phase linear induction motor in a cnc machine and some linear encoder, preferably an absolute. Also the motor will not have hall effect sensors, so it it possible that driver figure out the phases?
Also the amplifier will probably be Delta Tau Geo Direct PWM Drive

Edit: 3 phase AC synchronous linear motor. Not induction, like I mentioned before...

Thanks,
Steven

[stev]

If I have EMC2/Mesa anything i/o card -> pwm signal to servo amp, which is driving the sychronous AC linear servo, then I have linear encoder -> position signal to Mesa/emc2. What is current feedback and what is it for?

[PCW_MESA]

Originally Posted by stev If I have EMC2/Mesa anything i/o card -> pwm signal to servo amp, which is driving the sychronous AC linear servo, then I have linear encoder -> position signal to Mesa/emc2. What is current feedback and what is it for?

Looked at the DeltaTau direct PWM amp. Seems like it would be OK, its appears to have simple 3 phase PWM drive and what looks like a 2 channel SPI ADC for current feedback (24 signals total for 2 axis). So 2 axis could be driven by one 5I20 connector. Hardware wise it needs a cable adapter for the 36 pin cable and some signal inversions so power up state of the Amp. is disabled.

EDIT: Looks like interface signals are differential (good idea) so the 36 pin cable adapter card would need differential line drivers and receivers as well.

Current feedback is used to regulate the motor current to get torque mode operation, that is there would be a PI loop setup in HAL that measures the three phase currents (only 2 channels are needed since the third phase is just the difference of any 2 phase currents) and controls the applied motor voltage via the PWM signals

Software wise it needs the HostMot2 driver to support the 3 phase PWM firmware and the SPI firmware, plus the HAL file setup to implement the vector drive. Not trivial but not too scary either...

[stev]

That's good to hear! So with one 5I20 card I can drive 6 axis and with 5I22 card 8axis? Does Mesa have these 36pin cable adapter cards and differential line drivers and receivers for sale also or where i should look for these? If motors don't have hall effect sensor will it be a problem?

[jmelson]

Originally Posted by stev If I use EMC2 and Mesa Anything I/O card, can I get sinusoidal pwm signal for a H-bridge amplifier that will drive brusless AC servo?

I think there's a problem with your plan (but I am not absolutely positive). The problem
is providing the PWM drive to the switch transistors makes this a voltage-mode amplifier.
The motors provide a back-EMF from the moving magnets that varies in a sinusoidal
pattern. The drive increases the voltage above the back-EMF to force current through the windings.
So, I'm not sure that providing sinusoidal voltage gets you really good sinusoidal currents in all 3 windings all the time.

I've been using "6 step" drive on brushless motors with good results. There is a SLIGHT disturbance when the commutation switches to the next winding, but with almost every motor
I have tried, it is quite small. I only drive two terminals of the motor at any time, the 3rd terminal
is left to float. It sure looks like this commutation bump is less than the usual perturbations when a motor is running a real machine. I drive this with a SINGLE PWM signal.

Jon

[PCW_MESA]

Originally Posted by stev That's good to hear! So with one 5I20 card I can drive 6 axis and with 5I22 card 8axis? Does Mesa have these 36pin cable adapter cards and differential line drivers and receivers for sale also or where i should look for these? If motors don't have hall effect sensor will it be a problem?

No, we would have to make them. If there is enough interest we wouldn't mind doing it as its a very simple daughtercard.

Lack of Hall sensors does not matter if you have absolute encoders (or if you dont mind magnetic alignment at startup = thump start)

[PCW_MESA]

Originally Posted by jmelson I think there's a problem with your plan (but I am not absolutely positive). The problem is providing the PWM drive to the switch transistors makes this a voltage-mode amplifier. The motors provide a back-EMF from the moving magnets that varies in a sinusoidal pattern. The drive increases the voltage above the back-EMF to force current through the windings. So, I'm not sure that providing sinusoidal voltage gets you really good sinusoidal currents in all 3 windings all the time. I've been using "6 step" drive on brushless motors with good results. There is a SLIGHT disturbance when the commutation switches to the next winding, but with almost every motor I have tried, it is quite small. I only drive two terminals of the motor at any time, the 3rd terminal is left to float. It sure looks like this commutation bump is less than the usual perturbations when a motor is running a real machine. I drive this with a SINGLE PWM signal. Jon

We've done voltage mode and it works quite well. A little tougher with IGBTs because they are slow enough that the deadtime causes significant distortion
unless compensated for (and its hard to compensate properly without knowing at least the current direction).

The Delta Tau Amps (well drivers really) do have provisions for current feedback so a full vector drive (torque mode) could be implemented in EMC2, though to be done right we may have to phase lock the PWM to EMC2s servo thread.

[epineh]

Originally Posted by PCW_MESA Lack of Hall sensors does not matter if you have absolute encoders (or if you dont mind magnetic alignment at startup = thump start)

Thank you ! I have wondered about this for a while and I wondered if it could be done without hall sensors, the only problem would be at startup until the rotor locks to the magnetic position...

Cheers.

Russell.