Servo driver selection

Gatesy wrote: This was the only bit I found confusing about the PICO's. Do you have to do this to all AC servo drives, and does it have any effect on performance?

"The PWM frequency is determined by the motion controller, but should be between 25 and 100 Khz. The duty cycle can be reduced to zero, but must not reach 100%. Off time of at least 500 nS must be provided every cycle to recharge the "bootstrap" capacitors in the high-side transistor's gate drivers."

OK, first, (replying to an earlier message) there are no steps in the Pico Systems PWM controller/PWM servo amp system. Encoder position is read by the PC, LinuxCNC's PID computes a new velocity and sends it out to the PWM generator, and the servo amp provides current on the motor coils as selected by the Hall sensors.

Jon

Gatesy wrote: So you supply the drive with DC, and it then chops the DC into various proportions which becomes AC, so if you were to run the drive at 100% duty cycle, would you just end up moving the DC voltage line to a higher value, i.e. at 0% and 100% you have DC, and anything in between you have AC?

if the motor is turning, it is all AC, frequency determind by motor speed. The PWM pulses from the drive are integrated by the motor's inductance into a pretty continuous current flow.

So all the position control is done through LinuxCNC? As one closed loop circuit?

Yes, the position control feed through the PWM controller, and is managed by LinuxCNC.

Is this approaching more the setup the likes of Fanuc and Heidenhain use?

Fanuc sends 6 PWM signals to their servo amp, each drives one transistor. On the Pico PWM amp, there is one PWM and one direction signal, the current is sent to coils as directed by the Hall signals. I can't say how Heidenhain does it.

Jon

jmelson wrote: Fanuc sends 6 PWM signals to their servo amp, each drives one transistor. On the Pico PWM amp, there is one PWM and one direction signal, the current is sent to coils as directed by the Hall signals.

This probably needs a more detailed explanation, but at some point every AC servo drive is just 6 transistors driven by a PWM signal. Where is gets complicated is who/where the PWM signal is defined.

At one extreme HAL decides how fast the motor should move, and a HAL component computes the individual 6-transistor PWM duty cycle. At the other extreme a step-dir pulse stream sends position commands and the drive handles all the volts/ohms/henrys details.

Ok, so reading through the recent replies

So on Fanuc it sends 6 PWM signals to the 6 transistors (1 signal for each transistor) whereas on the PICO system it still has 6 transistors but 1 PWM signal is sent that then gets fed to the 6 transistors?

If the fanuc drives have 6 transistors and the PICO is a 6 step drive, is this the same thing? So if using a PICO drive on a Fanuc motors will it perform the same as if using a Fanuc drive?

Using the PICO system and having LinuxCNC compute the moves, to me this seems much more like a proper closed loop control system, rather than having other parts of the system computing things as well as LinuxCNC

Is there any advantage in sending the individual PWM signals to each transistor over just using 1 PWM signal, or was this just the way it was done at the time? Is it because technology has progressed that you can send 1 signal?

I'm guessing that like with everything the technology available now as compared to 30 years ago means you can do the same thing with far less components, and using the computer to do more of the processing?

Gatesy wrote: Ok, so reading through the recent replies

So on Fanuc it sends 6 PWM signals to the 6 transistors (1 signal for each transistor) whereas on the PICO system it still has 6 transistors but 1 PWM signal is sent that then gets fed to the 6 transistors?

If the fanuc drives have 6 transistors and the PICO is a 6 step drive, is this the same thing? So if using a PICO drive on a Fanuc motors will it perform the same as if using a Fanuc drive?

No, not the same. With the original version of the red cap motors, Fanuc apparently used a scheme where they have 16 steps in a synthetic sine-wave drive, as commanded by the 4 commutation bits from the encoder. The later systems with serial encoders generated a true sine wave using angle information from the encoder.

Using the PICO system and having LinuxCNC compute the moves, to me this seems much more like a proper closed loop control system, rather than having other parts of the system computing things as well as LinuxCNC

Is there any advantage in sending the individual PWM signals to each transistor over just using 1 PWM signal, or was this just the way it was done at the time? Is it because technology has progressed that you can send 1 signal?

Well, the Fanuc scheme allows them to have current in all 3 windings of the motor at all times. The six step scheme only has current in two windings at any time, and there is a current discontinuity every time the commutation advances to the next coil.

I'm guessing that like with everything the technology available now as compared to 30 years ago means you can do the same thing with far less components, and using the computer to do more of the processing?

In general, yes, but the Pico Systems servo amp is decidedly "old school".

Jon

Thanks for the further info.

Sorry but I'm more of a mechanical person than electronic, so I'm almost keeping up!!

So the output powers 2 windings, then as the motor rotates it moves round to the next winding, and so on. Is this like powering a 3 phase motor with single phase power where you rotate the single phase round the windings to simulate 3 phase?

Does the MESA 8I20 work the same way as it says this provides 3 phase power?