Closing the loop with Path Pilot. Possibly dual (global) closed loop?

[SCzEngrgGroup]

I was wrong, mostly reading old litterateur, and I was so wrong on Servo motors. Have to update my knowledge about it.

Also, this closed loop steppers recover from lost steps.

Which is trivial to do under essentially static conditions like that. I could replicate that video using a $20 encoder and a $2 microprocessor. But it's near impossible to do the same on a CNC machine when it is actually running, unless it's running very slowly... And, the video shows a situation that rarely occurs - the motor is over-loaded, then the overload is removed. In the real world, when an overload occurs that causes those lost steps, it is probably not going away on its own, so the motor WON'T be able to recover like that. There is a world of difference between a benchtop test, and the real dynamic situation on a running CNC machine.

Regards,
Ray L.

[R.DesJardin]

I was wrong, mostly reading old litterateur, and I was so wrong on Servo motors. Have to update my knowledge about it.

Also, this closed loop steppers recover from lost steps.

Thanks for the video its good to see positive comments that you can actually learn something from.............................I understood the point of feedback stepper servos was that you could keep caught up if you lost steps or prevent you from getting behind might be a better way to look at it. Interested in how this works out. I'm always interested in changes or mods that we can do, sometimes just because we can.

[RCaffin]

Also, this closed loop steppers recover from lost steps.

Well, no, it does not recover in a real CNC while milling. I would go so far as to say it cannot recover.

You see, when a stepper loses a step, it means the motor does not have enough torque to drive the load. To be sure, you might be able to sense that with an encoder, but there is nothing you can do to increase the torque of the stepper motor to overcome that load (with ordinary stepper drivers). Most likely the motor will just stall and sit there.

To be sure, you could tell the driver to increase the winding current in the stepper, but even if this recovers the lost step, the next step will also fail unless you maintain the higher drive current. But most steppers are driven at their full current rating anyhow, so you would be headed deep into thermal overload.

This is where a servo drive wins: when the driver senses that the encoder signals are starting to lag the required path, the drivers does increase the drive voltage and hence the drive current. When the encoder signals catch up, the drive automatically reduces the drive power. That is how a servo drive works.

If you want to move something 1 mm/hr, use a stepper motor geared down to a ball screw. But if you want to move something heavy and quickly, use a servo drive.

Cheers
Roger

[pickled]

I think that any way to error proof a set-up is a big win in my book. I'm not going to argue servo merits vs. stepper merits because they each have their place in this world. What I gather from this discussion is that there may be a relatively inexpensive way to improve upon the design that is in use on the Tormach itself by incorporating feedback. Essentially the biggest benefit would be that lost steps could be detected and the process could be stopped at a threshold level of missed steps. I'm not perfect and have fat fingered things like 1,000 RPM instead of 10,000 RPM or set the safe level mid way in the stock (oops) in the past. If the logic loop could be made to make the machine tool more intelligent about the failure mode of lost steps it could definitely assist in a situation like that. For any of you that are familiar with the concept of FMEA the current design scores poorly in the detection of anything less than a catastrophic upset condition. Thankfully the design itself is quite robust and missed steps are pretty rare minus operator error- but I am definitely NOT a perfect operator !

[SCzEngrgGroup]

If all you want is to detect following error, and throw an E-Stop, that can be done very inexpensively - add an encoder to the ballscrew, and a microprocessor, or hard-wired logic, to monitor the step pulses from the controller, and the encoder position, and throw an E-Stop when the differ by some threshold value. There are a few boards out there to do exactly this. It does not require $$$ linear scales, or new drivers - only the small amount of logic required to detect the error. With PathPilot, this detection can be done in software, in PathPilot itself, with the only added hardware being the encoders themselves, which are about $20-30 each.

Regards,
Ray L.

[pickled]

You are correct there. I was thinking of trying just that with an Arduino Uno in the loop to monitor and trigger a stop condition. I was waiting until I had a large issue that would justify some downtime for a repair and attempt an upgrade at that point. I'm still on Series II internals here so an overall motion control upgrade will be coming at some point (next driver failure most likely) hence why I was thinking those drivers and steppers would be pretty sweet.

[Zetopan]

Well, no, it does not recover in a real CNC while milling. I would go so far as to say it cannot recover.

You see, when a stepper loses a step, it means the motor does not have enough torque to drive the load. To be sure, you might be able to sense that with an encoder, but there is nothing you can do to increase the torque of the stepper motor to overcome that load (with ordinary stepper drivers).

That is not correct for all circumstances. If, for example, steps start getting lost on one axis at a non-zero stepping rate (which has to be the case), then adaptively lowering the stepping rates for all axes would also increase the torque while potentially allowing for regaining the lost steps on the stalled axis. Note that the stalled axis would need additional steps to get back into sync with the other axes. I am not personally aware of any commercial products that do this but it is potentially possible. In any case there would also be a machining defect where the stall occurred since the other axes cannot instantaneously stop to let the stalled axis catch up.

[SCzEngrgGroup]

That is not correct for all circumstances. If, for example, steps start getting lost on one axis at a non-zero stepping rate (which has to be the case), then adaptively lowering the stepping rates for all axes would also increase the torque while potentially allowing for regaining the lost steps on the stalled axis. Note that the stalled axis would need additional steps to get back into sync with the other axes. I am not personally aware of any commercial products that do this but it is potentially possible. In any case there would also be a machining defect where the stall occurred since the other axes cannot instantaneously stop to let the stalled axis catch up.

The trajectory planner CANNOT work well if the drive is deciding by itself to change velocity and acceleration, rather than closely following the path it creates. If the machine does not follow the path the planner specifies, you'll run into all kinds of problems. Bottom line, the ability to "catch up", other than over VERY short distances and VERY short periods of time (like a servo), is of little to no value. Besides, doing what you describe would require all the drivers to be cooperating, and reacting in unison. How in the world is that going to happen? They have no means of "talking" to each other. So, you'd slow down one axis to recover from a following error, while the other axes will blindly continue on as commanded. I don't think that's what you really want...

And, once again, since it doesn't seem to be sinking in with some people, stepper motors run at full torque ALL the time. There is NO reserve torque, like in a servo motor, that can be applied to make a stepper motor "catch up" once it has lost even a single step. If it loses a single step, ANY recovery will require slowing down, likely by a lot, to reduce the torque demand. At that point, all the work done by the trajectory planner is out the window.

There's a reason big machines use servos. Steppers can work flawlessly, if designer properly, and operated within their limits. Any attempt to exceed those limits will only bring grief, scrapped parts, and broken tools, no matter how much hardware you throw at it.

Regards,
Ray L.