pwm vs step/dir

[will gilmore]

I'm starting on a mill retrofit. I'm hoping to reuse the servos. I've used Gecko steppers on a earlier lathe project and I'm comfortable with how the controller / driver / motor work with step pulses using steppers or servos.

For this new retro I've been considering EMC2 and that led me to Pico (Motion Control Interfaces) who offers PWM servo drives. Is this better / worse / older / newer technology?

Thanks.

--Will

[samco]

pwm input drives are similar to +/-10v drives.. The pwm is signal is proportional to velocity, current or voltage going to the servo. This depends on the drive. There are many upsides to using this style of drive system with emc. This is a true closed loop setup. Emc knows the actual position of the machine at all times.

-home once. As long as emc is still running - it knows where it is.
-estop without loosing position.
-move the machine around manually and it will know its position.
-following error is known. mine is .00017" at the moment http://www.electronicsam.com/images/...ing/tuning.png
-I can set the amount of following error before it trips an estop.
-home to encoder index.

sam

[will gilmore]

are there upsides to step/dir drives?

[samco]

easier?

sam

[will gilmore]

ok fair enough. Assuming I'm willing to spend the time learning and configuring, will PWM give better results?

[neilw20]

One of the things with PWM is PWM distortion.
You need to use a high speed opto in the MB region so that the on rise time and fall times are identical, for practical purposes.

I use accurate PWM at the VFD generated by step/dir signals.
My step signal drives a monostable, that gives a very accurate controlled pulse width. The faster you hit it the mono the higher the percentage PWM. When pulse rate meets the mono time you are at 100%.
No pulse = 0%. Needs some safety logic, like requiring a minimum pulse width, so noise can't start the VFD when stopped.

That way PWM distortion from the computer end is solved. You can have the best of both worlds this way.

A standard VFD with analogue input, and step/dir signal so it is all digital, where the noise matters.

[will gilmore]

Originally Posted by neilw20 One of the things with PWM is PWM distortion. You need to use a high speed opto in the MB region so that the on rise time and fall times are identical, for practical purposes. I use accurate PWM at the VFD generated by step/dir signals. My step signal drives a monostable, that gives a very accurate controlled pulse width. The faster you hit it the mono the higher the percentage PWM. When pulse rate meets the mono time you are at 100%. No pulse = 0%. Needs some safety logic, like requiring a minimum pulse width, so noise can't start the VFD when stopped. That way PWM distortion from the computer end is solved. You can have the best of both worlds this way. A standard VFD with analogue input, and step/dir signal so it is all digital, where the noise matters.

Is it safe to assume that professional PWM amp designers have thought of this? For example this Pico drive: Pico Systems PWM Servo Amplifier

[neilw20]

Looking at the specs, it says only 3mA is needed for the opto.
Without knowing what opto is used I can only lead with these comments...

In my experience, so far, suitable optos need to driven at 6mA to minimize PWM distortion and speced in the 10MBd range, like ACPL-M61L typically 6nS error. Well under 0.1%, and it's output is CMOS suitable to drive an integrating circuit, which must be charged AND discharged at the same drive current.

A PWM of between 25 and 100KHz need high speed devices well into the MHz range.

A 4N35 has a time difference 18 off and 9 on, of 9uS. At 100Khz that is 50%!!!
A 6N137 has a time difference, of 35nS. At 100Khz that is under 1%
Problem is caused by the difference in rise and fall time.

The output of the opto coupler needs to drive a very accurate pulse width and height to integrate the PWM signal. This needs to be independant of temperature changes, and that is another story.

Many devices are not up to the task if precision is required, and the output side needs using the PWM has to maintain accuracy.

At mid range, you won't notice much error but at each the low and high ends the error can be easily measured with a tacho.

And all that aside, the perfect PWM signal has to presented to the PICO systems unit. Read back my previous using the set/dir signal so that all the errors are only at the decoding point.

A soon as you take PWM from a PC or similar, errors are inevitable, without special attention to propagation delays all the way to the recovered DC signal.

[samco]

I think you are over thinking it..

This is what a motion guru says.

All this accurate PWM stuff is rather silly if theres feedback loop around it

And that is what emc2 can do..

sam

[SWPadnos]

Originally Posted by will gilmore are there upsides to step/dir drives?

What's easier with step/dir is that there is no need for feedback to the PC, since step motors only move in "steps". (ie, since each pulse theoretically corresponds to a specific amount of motion, all the computer has to do is output the right number of pulses for whatever distance it wants the motor to move)

A second perceived advantage is that they are generally somewhat less expensive, especially when you take the cost of the feedback components into account.

- Steve