Wind / Unwind Application, Torque vs. Speed Control

So what I don't see is a mechanical master - a roll that makes the web run at one speed. The closest thing that I see is if your winder is in speed control, it will set the web speed. However, without perfect knowledge of the winder diameter, the winder will not make the web run at constant speed.

So if you try to run your unwind diameter calculator, the calculation is flawed because you don't really know the web speed - only the web speed setpoint from fieldbus. That is not necessarily the web speed at every given moment.

So I think the winder needs speed control, but is there a diameter sensor there? Is there an intermediate roll that has an encoder for web speed feedback? Somehow the winder needs some feedback so it can half accurately run at web speed setpoint.

After that, I would suspect that the unwinds might be better off running in torque control with diameter sensors rather than diameter calc.

One thing to check during your next test. Can you track the actual webspeed at the winder to see how consistent it is.

At small diameters you will have your greatest challenge to control well - not surprising

There is some communication happening behind the scenes, but I'll share a few screen captures that I'm sending in email messages. The first test is focused on the initial ramp up. I extended ramp times for all drives from 30 sec. to 60 sec. You can see that it still loses control at the higher RPMs.

Note that I started with a half roll already on the winder, so the high RPMs come quicker.

New development... I'm been asked by management to develop a 'plan B' for if we are unsuccessful in getting the ABB drives to work. I need a little help with a question...

I have a CompactLogix 5380 PLC connected to the drives through Ethernet IP. If I needed to use the PLC to run the torque reference loop, is there a source for a developed wind/unwind application that I can plug in to RSLogix Studio? After all, that's a fairly universal application.

In the mean time, we're working with ABB to get the original plan working.

You could run the unwinder in torque limit mode only but this would require that the torque loop PID has a very high bandwidth to respond to the down stream winder. Wouldn't it be better to compute a target velocity for the upstream rolls based on what the down stream winder is doing? I know the calculations will never be perfect but they should be very close. Then I would adjust or trim the upstream rolls speed based on torque. The only tricky part is knowing how fast the upstream rolls should be turning because the diameters are always changing but this calculation can be done every millisecond along with changing the gains every millisecond as a function of the roll diameter.

It looks to me like the gains are too high because the torque loop isn't fast enough to respond to down stream changes fast enough.

It would be nice of have a legend for the plots so we know what each line is. I can figure out some.

This would be a good application for a common buss configuration.
first make sure that the drives can do torque control some can and some can't not just torque limit
Remember the torque in the unwinders must be in the opposite direction of the web travel you are trying to maintain tension on the web. the speed is controlled from the winder only.

GaryS said:

This would be a good application for a common buss configuration.
first make sure that the drives can do torque control some can and some can't not just torque limit
Remember the torque in the unwinders must be in the opposite direction of the web travel you are trying to maintain tension on the web. the speed is controlled from the winder only.

Click to expand...

This is a common-bus configuration. The winder is connected to the line, and the unwinders are connected to the winder through a common DC bus.

Torque control... that's the way the old ACS800s were configured. That's why I'm surprised that they had me switch to speed mode. I don't know for sure torque mode won't work. I just know several people couldn't get it to work, and a senior app engineer told the junior guy to tell me to switch to speed mode. Right now I'm trying to talk to the senior engineer.

The biggest tell-tale factor in which mode is being used is what happens when you hit 'run'. In torque mode, the unwinders start turning in the opposite direction as the line web, and in speed mode they 'pay out paper' to the line.

it look like you have the torque in the wrong direction an increase in motor torque the feed back torque drops
I assume brown is feedback torque green is motor torque
the unwinder torque must be in opposite direction from the web travel

It's not in torque mode, it's in speed mode since torque mode wasn't working with the PID.

Peter Nachtwey said:

It would be nice of have a legend for the plots so we know what each line is. I can figure out some.

Click to expand...

Here's your legend, Peter.

Update: ABB has assembled a team of experts to look at this this application. They will be giving me a list of options Monday afternoon.

I don't think I could ask for more. I'll let everyone know how it goes.

How does that crow taste
Where is DickDV ?

I would love to eat crow if they're serving it. All I want is for the project to be completed. The fact that ABB has decided to invest time into helping me is a good thing. If these folks can't make the firmware work, I doubt that anybody could.

I'm going to go with a loop/calculation rate issue in the paper unwind drive. If you look at the tension feedback, the tension drops like a rock then relatively quickly recovers to near the tension setpoint. The tension then "slowly" increases (relative toe the drop) up a speed dependent value then drops back down. My guess is the estimated unwind diameter is not calculating quickly enough. A load cell tension loop with something as stiff as paper will need to have relatively low gains to keep it out of oscillation. This means the zone will be relatively intolerant of diameter errors. It looks like the unwind is calculating a diameter, holding it for a while then calculating another diameter.

We program velocity based center driven winders here almost exclusively. We also program them in the plc almost exclusively. It can be a little difficult to get closed loop torque winders to stay stable because there is either little inherent damping in the system (low gear ratio) or very little torque feel at the motor (high gear ratio systems). The trick to velocity based center winders is roll diameter and speed stability. If you have a very stable base web speed and a relatively accurate diameter you can get very accurate tension control.

Open loop torque based tension control is a gimme. As has been suggested, you might want to do that as a test just to see what you can accomplish as a base.

Keith

winder speed control with the reel sensor to keep it a bit within limits, so if reel is empty run it fast and when reel is full slow down the winding.
unwinders when tension sensor is high, speed up unwinder. if tension is low stop unwinding, but never run backwards.
no need for communication etc.

How did you end up on this? Did the ABB guys get you taken care of?

Keith