Velocity versus Torque Mode

Hi all,
I've been programming AB processors for quite a while but I haven't had much opportunity to work with motion controls.

I have an application coming up where I will be using AB K6K drives. If CIP drives are the better way to go, I am open to that also.

Corrugated media is fed from one system (via Vacuum Pull roll), slit in half and feeds into my system to a wind up the slit rolls.

From reading some of the posts, is it safe to say that after the media feeds into the center of the winder hub and pinches the media, maybe make one or two relatively slow revolutions so the media is secure. Then set the winder in torque mode using the SSV until the media winds from a 4" hub and grows to 27". After than, set the winder back to velocity mode so I can cut the media Just prior to the winder hub? I need to cut the media at the peak of the flute which is what the encoder is for. I think I need to do some type of regression calculation for the length of media so I know where I need tostop the winder at a certain location and cut the media?

Any help on this would be much appreciated!

I'm personally not a big fan of torque mode center winders unless you are extremely careful with the mechanical design. The problem you can have, especially at core, is that the drivetrain torque losses are greater than the torque required for the tension you are trying to run. In addition these losses tend to be somewhat variable so accounting for them can be tricky. Finally, I'm pretty sure the guaranteed torque accuracy for the whole K6K family is something like 5% of rated. At core, this is probably the better piece of the torque command. If the intent is to wind relatively high tension at core (requiring, say, 25% of motor rated torque) then torque mode operation can work.

I am more of a fan of velocity controlled center winders. The velocity loop in the drive or controller can compensate for system losses. In addition, velocity control can provide some damping for oscillations triggered by product web to product roll spring/mass interaction. The tricky part with velocity center winding is you need a reasonably accurate diameter value for it to work well.

This is all based on the assumption that this winder has either dancer or load cell feedback for control. If not, torque control is about all you will have to work with.

Finally, is the web cut-off automatic or manual? In terms of tracking the flute tips you might have better luck with a sensor close to the winding spindle detecting the flute tips and tracking the shorter distance to the core.

Keith

I prefer torque mode but only in certain special applications.

No matter what anybody says the motor turns using torque generated by current. There is always some form of torque control.

Keith is right but the drive is doing the torque control for him so he doesn't need to get into that nastiness. Keith must use fairly good drives.

My experience is that the torque loop in a lot of drives is not very good and we can control the motion better by just putting the drive in stupid mode where it just converts our +/- 10 volts to current.

My main concern with torque/current mode is safety. Usually a brake is needed to stop motion in case of power fail or fault.

Originally posted by Peter Nachtwey:

My experience is that the torque loop in a lot of drives is not very good and we can control the motion better by just putting the drive in stupid mode where it just converts our +/- 10 volts to current.

Click to expand...

Peter is correct. But keep in mind that when he states "we" and "our" he is talking about the output of a high bandwidth highly configurable motion control platform, not a plc.

Originally posted by Peter Nachtwey:

My main concern with torque/current mode is safety. Usually a brake is needed to stop motion in case of power fail or fault.

Click to expand...

This is correct, although I'm not as concerned about power failure as I am about web break. In the case of power failure the spindle can't go any faster than it was going when power was lost. Since it is presumably still connected to the web it will likely stop along with everything else. However, if the web were to break with a near full roll under tension the motor will accelerate up to the speed capped by the bus voltage limit. This could result in a very high surface speed. If you use true torque mode (as opposed to some velocity control/limited torque mode) you will want to be very conscientious about web break detection and response.

Keith

Keith,

You mentioned that when there is a web break, the spindle will accelerate up to the bus voltage.

I understand that the torque feedback is trying to satisfy the command. However, is there a way to shutdown the spindle by comparing some error or some other means? Can velocity error be observed when running the motor in torque mode?

Thanks!
Kevin

Originally posted by kpElec:

Can velocity error be observed when running the motor in torque mode?

Click to expand...

You won't get velocity error from the drive when operating in torque mode since velocity isn't a controlled parameter at that point. you will, however, get back actual axis velocity. You can calculate what the axis velocity should be based on web speed and product roll diameter and make sure the velocity feedback is within an acceptable deviation of the calculated value. If you detect a deviation, stop the axis.

I don't think and of the Kinetix 6K variants support SLAT mode (Speed Limited Adjustable Torque). If one does, that is really the functionality you are looking for.

Keith

I don't understand the limitations
We can do a dual PID low select. One for PID torque and one for PID for velocity. The amplifier/drive must be in stupid voltage to current mode. Should the web break the motor will only spin up to the speed selected by the speed control PID. The speed and acceleration are monitored at all times.

The trick is that the controller must not be in torque limit mode when trying to accelerate up to speed. Usually that is part of the user program to enter limit modes when the desired state is reached.

We do the equivalent all the time for presses and similar applications that require squeezing but not squishing.

Originally posted by Peter Nachtwey:

I don't understand the limitations

Click to expand...

Talk to your pals in Milwaukee. You have a deeper in with them than most of us do.

I'm not sure why they went the direction they did with the command interface to the K6K line. What we are talking about would be relatively easy with a PowerFlex 750 series drive. It is a built-in function with the new K5700. Where is went with the K6K I'm not sure.

Keith

Thank you both for your insight and great information!

Keith, I think your reference was to me in why we went the way we did?

I'm actually working on a design that has been used for years where there are two rotary hubs (spindles) that extend to each other to form one hub with a very little gap between them.
The rotary hubs are homed at the beginning of each cycle so that the slots in the hubs are aligned and facing the entry of the media to be inserted into them.

The media is slit before going into the hubs. After the hubs are done spinning up the media, they retract and some stationary stripper plates are in place to strip the media off the hubs onto a conveyor.

It was my understanding that servos are needed to run in torque mode for winding up the media and switching to velocity mode so the media can be cut on the flute peak of the media for the next operation.

One other thing, is there a better and more user friendly software out there for calculating the servo size of an application? Something better than motion analyzer? I've asked Werner Electric to size our servo because of new demands and they are coming up with two sizes larger than what we had before. The biggest change is that the decel of the servo requires a 2000 mm/sec2 ramp.

Thanks,
Kevin

I always have been under the assumption that 'max allowed velocity' or 'speed limit' parameter is always a part of the torque mode control. This is how Yaskawa servo drives operate. So this seems to not necessarily be the case with other brands?

Originally posted by kpElec:

Keith, I think your reference was to me in why we went the way we did?

Click to expand...

I was actually referring to Rockwell and it was a little bit anecdotal. I don't think the guys at Rockwell really envisioned users using their servos the way you (or I) usually do. So they didn't include that tool set when they developed the upper level control structure. Multi-axis coordinated motion of a single piece of tooling? Yes. Tools for web handling? Not so much.

Originally posted by kpElec:

I've asked Werner Electric to size our servo because of new demands and they are coming up with two sizes larger than what we had before.

Click to expand...

The guys at Werner are a bit of a slave to the tried and true servo axioms. One of those axioms is that the motor to load inertia ratio shouldn't be greater than 10:1 in order to achieve stable, high bandwidth control. You are getting bigger motors because they will have higher inertia rotors and will get you closer to the 10:1 inertia ratio. It's not just Werner/Rockwell. Many of the servo suppliers will default to this unless they really take the time to understand what you are trying to do. In your case you are using the motor more like a no-slip/no wear clutch. You aren't as concerned with high bandwidth control. As long as the motor has the required amount of torque to satisfy your winding requirement you will likely be golden. There are several more items that go into the whole stability thing, by the way. Not the least of which is mechanical system compliance.

Originally posted by kpElec:

It was my understanding that servos are needed to run in torque mode for winding up the media and switching to velocity mode so the media can be cut on the flute peak of the media for the next operation.

Click to expand...

I don't know enough about your system to be able to say this definitively, but I'm not convinced you need to wind in torque mode. It depends on whether you have tension or position feedback before the spindle or not. If you have wound this material successfully in torque mode on previous similar machines that is a pretty good indicator that you will be able to on this machine. How are you determining where the flute peak is so you can cut there? How do you get the flute tip where you need it to cut? Since this is a web process the material needs to come from somewhere and that somewhere will also need to move to get where you want, unless you are backing up to get to the flute tip.

Originally posted by LadderLogic:

I always have been under the assumption that 'max allowed velocity' or 'speed limit' parameter is always a part of the torque mode control.

Click to expand...

This may be true but one would expect that the max speed would be set to the value needed to get to web speed at core, which, if sized right, is very near motor max speed. In any case, if you break the web on a full roll and still let the axis go max speed you will end up with a relatively large inertia spinning at a much higher surface speed than you want it to be.


Keith

kamenges said:

This may be true but one would expect that the max speed would be set to the value needed to get to web speed at core, which, if sized right, is very near motor max speed. In any case, if you break the web on a full roll and still let the axis go max speed you will end up with a relatively large inertia spinning at a much higher surface speed than you want it to be.


Keith

Click to expand...

Just for my own curiosity, couldn't the max speed be dynamic?

I'm not familiar with spool applications, but I have worked with variable speed extrusion applications. In my mind I am thinking the two could be similar.

From what I've seen in our applications we control by speed with max torque. So we set the speed to what the material coming out is, but limit the drive to a maximum torque that we want to keep tension on the material.

So if our material breaks, a runaway doesn't occur cause it only goes to the speed which it thought it should be moving at.

It could be but the K6K doesn't make it particularly easy to do that.

I think you could perform the equivalent of a SLAT by setting the axis in velocity mode and continuously writing the torque limit parameter with an SSV. Set the velocity setpoint 10-20% higher than what you expect is needed, which would bury the axis in torque limit. To the best of my knowledge the axis torque limit is not a parameter you can map in the cyclic data transfer so it makes that a little harder to get to.

Keith

KpElec,
I sent you over a PM. I live in MN and have a good amount of web handling equipment experience. Let me know if you would like some help on this.

I don't know enough about your system to be able to say this definitively, but I'm not convinced you need to wind in torque mode. It depends on whether you have tension or position feedback before the spindle or not. If you have wound this material successfully in torque mode on previous similar machines that is a pretty good indicator that you will be able to on this machine. How are you determining where the flute peak is so you can cut there? How do you get the flute tip where you need it to cut? Since this is a web process the material needs to come from somewhere and that somewhere will also need to move to get where you want, unless you are backing up to get to the flute tip.

Click to expand...

Keith, for my application, there is no tension feedback nor dancer per say. The current process winds the media to a specific number of turns and at the end of the run, the winder reverses direction in velocity mode to a low speed. An encoder counts the flutes as the encoder has a flute wheel that matches the flutes on the corrugated media being counted.
I want to eliminate the reverse motion and calculate the length of the media to be wound. Then decel the winder to a speed where the encoder can count the flutes to a point while still moving forward. Then stop the winder motion and cut the media. Thank you.

KpElec,
I sent you over a PM. I live in MN and have a good amount of web handling equipment experience. Let me know if you would like some help on this.

Click to expand...

blperkins, Thanks for the offer. When you say you sent over a PM, what is that acronym? Does that stand for project manager or a different acronym that I'm not thinking of?