Dancer Systems

[scottmurphy]

John Paley,
Thanks for your input there, it has given some information that I did not have.

In my experience, the operators don't like to see the dancer move while winding/unwinding. But it has to. here's why.

That is exactly what they ahve said to me, it should not move, they have an idea that they can rest their coffee cups on the dancer....not possible I guess.

In the old analog systems, this can be done very smoothly and dancer movement is minimal. In the digital systems of today, the diameter range is traversed in steps. The better the diameter resolution, the smaller the diameter step, which means less dancer movement. But it still has to move to work.

I think this is the key to g to understand the old system. There is a great selection of bias resistors, trim pots, adjustable resistors etc.

In an unwinder, first you need an accurate calculation of the starting diameter. This is done differently in different systems. But it is basically line speed ref / spindle RPM

How is this normally done, there is a Spindle RPM signal installed? This will be another key issue, as the rolls are changing in diameter depending on thickness. I was going to start with an average, and work from there.

The type of transmitter you use depends on your system.

I am using the LWH series at this site http://www.novotechnik.com/linear.html to give me dancer position.

At the moment, I am only going to do the UW end, so the RW will have the old analogue system still.

Will I be able to get the dancer to hold at 0 speed using this method?

 

[john paley]

The winders that i have experience with all did "flying splices". that is, the spindles changed automatically. to accomplish this, the operator "doctored" the starting end of the material on the incoming, or new, or full, roll with duoble sided tape. Then the new roll was positioned above the expiring web with a pasting roller an knife beneath the expiring web. The new roll accelerated to synchronize it's surface speed to line speed. Then the paster shoves the expiring web onto the rotating new roll until the double sided tape sticks them together. The knife fires and cuts the expiring web away, and the new roll switches from synchronize mode to winding mode.

Most winders use a "lay-on" Tach or encoder to measure the diameter of the new roll. This can be done before or during synchronization.

If done before synchronization, at calculate time, the new roll is rotated at a known RPM and the surface speed is measured by the lay-on device to calculate diameter. The diameter value is used by the drive to synchronize the roll speed during synchronization and as a starting value in winding mode. The lay-on device is engaged for calculation but not synchronization.

If done during synchronization, the lay-on device is engaged and the lay-on value actually replaces the motor tach in the speed loop to set the spindle rpm until the roll surface speed matches line speed. But the motor tach is still used to calculate the starting diameter for winding mode.

I've also seen distance sensors (ultrasonic, laser, etc.)used for diameter measurement. They sense the roll size.

You could also let the operator enter the value in some fashion, but this leaves room for errors.

 

[scottmurphy]

This system also has flying splice, I think I have that under control though, the biggest issue, will be the speed synchronisation.

Most winders use a "lay-on" Tach or encoder to measure the diameter of the new roll. This can be done before or during synchronization.

I have a tacho that lowers down to the new reel, either by the operator, or automatically when splice is required.

If done before synchronization, at calculate time, the new roll is rotated at a known RPM and the surface speed is measured by the lay-on device to calculate diameter. The diameter value is used by the drive to synchronize the roll speed during synchronization and as a starting value in winding mode. The lay-on device is engaged for calculation but not synchronization.

I didn't think of this method, this could be more useful for obtaining the initial diameter, and then, in turn, doing the decreasing diameter calculation.

If done during synchronization, the lay-on device is engaged and the lay-on value actually replaces the motor tach in the speed loop to set the spindle rpm until the roll surface speed matches line speed

This was the way I was planning on doing the speed sync, but I am going to have a look at the other method now. This will give me the facility to have a more accurate method of knowing the roll diameter and length, if all the operator has to do is enter the thickness.

You could also let the operator enter the value in some fashion, but this leaves room for errors.

This always happens when an operator is not competent in their machine, all you can do is train them, and let them deal with the consequences, not ideal, but that is the reality of it.

 

[scottmurphy]

Hi All,
I have finally as the chance to swap over to the new control system, replacing the old control boards and drives with PLC and modern drives. Have had mixed results commissioning the system though, have made some notes / comments / findings below.


Current / Previous System

1. Motor Enabled
2. Motor is delivered maximum current & ( effectively max speed )
3. If Dancer moves, then step 4, no dancer movement, then maintain max speed.
4. As dancer approaches mid position, current & voltage are reduced ( speed reduced ) and dependant on reel size, or possibly rate of change of dancer, a 'holding reference' is delivered to motor to hold the dancer & reel
5. When web begins moving, current & voltage are reduced, maintaining the dancer in the mid position.

My control method is as follows;

1. Motor Enabled
2. Enable PID control loop;
PV = Dancer Position,
SP = Mid Position,
CV = Motor Speed Reference,
3. When Dancer is within a 'deadband' level, 30-70%
and the dancer is stationary for a delay time, then disable
PID loop, and set Motor Speed reference to a 'holding value' that is just enough to stop the Dancer Pressure lowering the Dancer, but not too much so that the Pressure is overcome and begins rising. The reel size is also a factor, as this affects the weight.

4. Dancer SP is 'corrected' to current dancer position, this is to limit the response of the PID loop when enabled again.
5. When web begins moving, PID loop is enabled, motor speed reference is calculated based on line speed, motor rated RPM & gear ratio.
6. Percentages of the speed reference and PID CV are then added together to give the total motor speed reference.

As always, the theory and calculations come together, and these are used as a basis for implementing the control system. But again, as always, commissioning seems to deny the theory and math, and not quite perform as expected. There is the possibility that my calculations are incorrect also,

I thought that I was on the correct track with this method, but I seem to be suffering inconsistency when the web is moving. What is happening, is that I seem to lose control of the dancer, and once it starts to drop, the loop starts to respond, by decreasing
the PID CV, thus reducing the speed ref, but as there is still a line speed reference, this is the dominant factor.

Should I ignore the line speed ref, and rely totally on PID CV? This will give me more of a 'torque control method' where the speed reference is adjusted, depending on where the dancer position. ie. dancer dropping, negative reference, dancer rising,
positive reference.

Or could it be a tuning issue?

I think that my method for holding tension is ok, this seems to be the most consistent, even the 'dancer SP' auto correction that I am doing is reasonable, providing the reel diameter calc is correct.

This press has very good documentation, with a lot of circuit diagrams etc, but I am no electronics expert, so to decipher them is an uphill battle.

My belief is, that I should be able to get this to work, with the PLC control method, butthe problems that i am seeing, I think they are related to response, inertia etc.

Where can I find formulas for calculating the torque requirements for a given reel size, the torque OP of the motor / gearbox, the amount of inertia generated by the reel, the amount of torque required to control the reel.

I have the option of possibly changing the gear ratio from 5.66:1, to 8:1. I have not pushed this issue yet, as the previous system worked with the current setup. The benefit of the gearbox
change, will give me better torque control, but with DC motors, I should be able to get maximum torque at zero speed. I need to know how much torque I need, and I can deliver with the current
setup.

I am going to have to take a break from this, so that I can clear my head and approach it with a fresh attitude, as it seems that I make progress, and then go backwards.

Any comments, suggestions on my methods etc will be greatly appreciated.

 

[OkiePC]

i have worked with a bunch of different variable diameter winder applications, from the crudest and simplest straight PID control to the most complex over engineered ratio detectors written by Reliance.

The over engineered approach works very well for months at a time, but when a device acts up or fails, it is a bear to troubleshoot.

The straight PID approach will work if you can tolerate a little upset in the dancer position each time hte line speed changes. IF the accel and decel rates are moderate, it will work very well. Simply use the dancer as the PID PV and the PID CV is sent to the drive as it's speed reference. I don't like this method, but it is quick and easy to work with.

What I like to do is "model" the winder in the PLC. Use this formula to get the OD of the variable diameter winder/letoff:

OD=SQRT(L*t/(pi/4)+ID^2)

Where:
L is the length of the product IN THE WINDER/LETOFF STAND. You have to accumulate this length (or subtract from a loaded known total length for a letoff stand)

t is the thickness. (HINT: If the thickness is variable, use your PID to tune this parameter...)

ID is the core diameter.

When you have the OD, you can calculate the correct center-drive speed reference to match the line speed.

Even with the most perfect measurements, there will likely be some variability.

I use a PID like you, but only allow it to trim the final calculation or to trim the unknown or widest ranging value in the formula above. Just make sure you rescale and limit it's output. if , for example, you let it "tune" your thickness parameter. You should be able to "fix" the upper and lower limits of what the thickness can be, and make sure your tuned Thickness used in the calc stays clamped within this range.

If your thickness is a constant within reason, just use the PID to tune a TRIM ratio. I like to rescale my PID CV to a values like 0.95 to 1.05. (1.0 +/- the amount on influence you want it to have).

It is important to FREEZE (Manual Mode) the PID when the line stops, or more generally, any time the output of the PID cannot affect the PV.

So now you have a calculated line speed and it is trimmed up to 5% for variations by a PID with the dancer position as its input.

Using this method, I have a machine that runs from 0-80 meters per minute with 0.5 second accel and decel times. It is a steel belt cutter/splicer/winder so there are many frequent and quick stops and starts between a few minutes of high speed operation.

In my machines, I set the drive accel and decel rates as fast as possible, and manage all ramping in the PLC. A slow responding drive can cause PID overshoot and oscillations in the dancer.

My "dancer" is a free hanging loop of stock being monitored by a 13" tall analog light curtain. The loop position doesn't change at all during the winding operation, and stops and starts only disturb it about 1 inch or so. I am very satisfied with it.

Using this modelling approach, with accurate numbers, you will be able to overdamp the PID and set the gains quite low for a very stable operation.

Hope this helps...

 

[kamenges]

Based on scottmurphy's last post it almosts sounds like speed reference and torque reference are being used interchangably. Be careful with that thought.

For example, the comment is made that when the drive is first enabled the PID is used to bring the dancer inside a deadband. Then the PID is disabled and a fixed 'speed reference' is sent to the drive to maintain dancer position. If that were a 'torque reference' it might have a chance to do do what is desired. Sending a speed reference will cause the motor to run at a fixed speed, moving the dancer. That is, unless, you have disabled the integrator on the velocity loop. If that is the case you need to tell us as that will influence the action of the system to some degree.

Assuming you are using a 'standard' implementation (PID provides velocity trim to the baseline speed derived from line speed and reel diameter) I would leave the PID loop enabled any time the drive is enabled. If designed right a dancer system should be statically stable. There should be no reason to disable the PID at zero speed. Re-enabling the PID when you start is just another case to worry about. Now, if you use OkiePC's method of trimming the model with the PID, then you need to keep it disabled if you are not running.

If you are using the PID output as a speed trim you may want to consider adding the PID loop output to the line speed reference and scale that sum based on roll diameter. This will keep the PID response constant over the range of diameters. You are trying to control dancer position, which responds to tangential roll speed.

Based on how you describe your problem I think your biggest issue is your baseline speed is not right. Look through your roll diameter and spindle speed command calculations very closely. Don't just look to make sure the equation forms are correct; they probably are. Look more closely at conversion factors (bits/1000 motor RPM and stuff like that).

As an aside, I have also gone the auto-correcting model route like OkiePC. I did it with a load cell system, however. In my case I used the PID to trim the linespeed-base velocity. However, I used the integral sum of the PID to trim my diameter calculator. I needed to take a differential on the integral sum to make sure that I didn't overcompensate in the model. That ended up working pretty well, especially at speed change. Accel and decel are where the model-based systems really show their advantages.

Keith

 

[rsdoran]

I have dealt with numerous types of winders and I am not fully understanding exactly how this one works. I THINK you are using an Unwinder that is using a DC drive to unwind paper from the roll at a rate that will basically match line speed, with minor difference to create X amount of tension.

What I do not understand is why you are attempting to "control" the dancer position, the dancer feedback should be regulating the UW speed to provide the necessary tension etc which will maintain the dancer roll in the proper position. Maybe I misunderstood your process.

I am late in providing this material and my apologies but hopefully this info can help;
http://www.geindustrial.com/products/brochures/DB2.pdf

http://www.apicsllc.com/apics/Ie_ias01/Ie_ias01.htm

http://www.bcconverting.com/content/8techtips/unwindapplications.pdf

I have more things on a CD, will attempt to post it when I locate the information.

 

[kamenges]

Originally posted by rsdoran:
What I do not understand is why you are attempting to "control" the dancer position, the dancer feedback should be regulating the UW speed to provide the necessary tension etc which will maintain the dancer roll in the proper position. Maybe I misunderstood your process.

Ron, I think it's just a terminology thing. He says 'control dancer position', you say 'regulate unwind speed'. In this case that is the same thing. You effectively put the dancer where you want by changing the unwind speed.

Keith

 

[rsdoran]

Keith that may be the case but this statement:

3. When Dancer is within a 'deadband' level, 30-70%
and the dancer is stationary for a delay time, then disable
PID loop, and set Motor Speed reference to a 'holding value' that is just enough to stop the Dancer Pressure lowering the Dancer, but not too much so that the Pressure is overcome and begins rising. The reel size is also a factor, as this affects the weight.

And your comment:

For example, the comment is made that when the drive is first enabled the PID is used to bring the dancer inside a deadband. Then the PID is disabled and a fixed 'speed reference' is sent to the drive to maintain dancer position. If that were a 'torque reference' it might have a chance to do do what is desired. Sending a speed reference will cause the motor to run at a fixed speed, moving the dancer. That is, unless, you have disabled the integrator on the velocity loop. If that is the case you need to tell us as that will influence the action of the system to some degree.

This has me slightly confused I guess.

All my experience with winders and unwinders is primarily the paper industry but have worked with some sheet metal applications.

On the gravure presses the unwinds were primarily festoon type that did not have a driven roll, the paper was just pulled from the festoon by the line drive (or in some cases a take up motor)and a brake applied to the reel to incorporate tension. In this situation dancer feedback was used to determine braking applied.

On one of the flexo presses we had turret unwind and rewind, these did drive the reels to feed the paper. On these machines the line drive was Master and the dancer feedback was used in conjunction with line speed to set wind/unwind speed to apply specified tension.

I have rebuilt 2 rewinds but I used WebPak 3000 drives from Reliance, most of the calculations etc can be handled right in the drive.

I am not as good or do these things as many of you do but it appears, to me, scott is trying to use speed for control. I always thought line speed was a primary reference that develops the torque/tension control which in turn will develop the unwind reel speed based on changing reel sizes etc.

I am not attempting to start a debate, just trying to understand what is being said and attempted. I will have to take some time to absorb some of this.

 

[kamenges]

Ron-

Part of the problem is that dancer based winding has been done so many different ways that it is a little tough to talk in absolutes. So we will make a couple of assumptions. The biggest assumption is that scottmurphy is attempting to implement a fully speed based (as opposed to torque based) dancer controller. This seems to be that method that is most in favor these days.

This may seem like a minor semantic point meant to confuse the issue, but it is important. You are NOT controlling tension with a dancer based control system. You are controlling dancer position. The tension in the web is a result of the amount of force the dancer puts in the web. Assuming you keep the dancer floating the dancer controller is not specifically concerned with tension.

Ultimately the drive/motor needs to generate torque to develop the required tension. However, this isn't a direct result of what the dancer position controller is commanding. It is a result of the drive's internal velocity loop. From the standpoint of what scottmurphy is concerned with, everything revolves around speed command.

Because of this we can take the line speed reference, scale it based on roll diameter, and send it out directly to the drive. This will produce the major part of the speed command. Added to this will be a speed offset based on dancer position. Both of those values are speed values.

Assuming the drive in question has a properly tuned proportional/integral speed control loop, the drive will internally develop the torque required to suspend the dancer. Another way to look at this is, if we suspend the dancer, enable the drive at zero speed and then let go of the dancer, the drive will put out the torque required to keep the motor shaft at zero speed.

Like I said, this might be a fine point, but it helps to think about the fact that the commands to the drive aren't specifically concerned about torque/tension. The commands are concerned with velocity.

Keith

 

[SimonGoldsworthy]

Any comments, suggestions on my methods etc will be greatly appreciated.

Until you can control the unwind roll with the material taped down and get the roll surface speed to follow line speed (with the diameter temporarily fixed at the correct diameter and the dancer correction disabled) I would not attempt to run material through your machine. Get your unwind speed control right first, then enable the dancer and check it trims correctly, then run limited material then finally enable the diameter calculation.

What signals are you sending to the drive ? Speed demand/torque demand/torque limit ?

 

[scottmurphy]

Hi Guys,

Thanks for your replies, I apologise for not providing all the information.
The System is as follows, a center driven UW & RW, a pulling motor, and a main drive that controls the print cylinders.

I have a transducer mounted to the dancer to give me the position 0-100%, and a surface driven tacho that gives me the line speed in mm/min ( I calculate this based on the value of the tacho IP, and is pretty accurate ).
At the moment, I am only working on the UW, the RW, pulling motor & main drive are driven by the old system. The RW has the same control system as what the UW has, the pulling motor is used to match the web speed to the cylinder speed before lowering the print cylinders, and the main drive is controlled by a motorised pot.
So when the pulling motor is enabled, the web starts moving, and I get a line speed reference from the tacho.

For example, the comment is made that when the drive is first enabled the PID is used to bring the dancer inside a deadband. Then the PID is disabled and a fixed 'speed reference' is sent to the drive to maintain dancer position. If that were a 'torque reference' it might have a chance to do do what is desired. Sending a speed reference will cause the motor to run at a fixed speed, moving the dancer.

The reason for this method, is so that the tension is taken up, and will hold with no dancer movement when there is a zero speed reference from the tacho.

That is, unless, you have disabled the integrator on the velocity loop. If that is the case you need to tell us as that will influence the action of the system to some degree.

Do you mean the integrator in the drive, or the PID in the PLC?

Based on how you describe your problem I think your biggest issue is your baseline speed is not right. Look through your roll diameter and spindle speed command calculations very closely. Don't just look to make sure the equation forms are correct; they probably are. Look more closely at conversion factors (bits/1000 motor RPM and stuff like that).

I sort of think my calcs are right also, when you say 'bits/1000 motor RPM' are you meaning the accuracy of the reference that I am giving to the drive?
I am using a 0-1000 range in PLC, so by my calcs, for a change of value by 1, the full resolution of the AO is 0-6000, so when multiplied by 6, I get a change of 0.01 volts. The drive sees this as a value 0-100% as well, the diagnostics of the drive, shows an accuracy down to 3/4 decimal places on the 0-10v signal.

Ron, I think it's just a terminology thing. He says 'control dancer position', you say 'regulate unwind speed'. In this case that is the same thing. You effectively put the dancer where you want by changing the unwind speed.

Keith

This is correct, I use the line speed reference, and use the dancer position as a trim on the line speed.
I take the line speed reference, calculate the motor RPM required based on gear ratio, reel size and motor nameplate RPM.
I then take a % of each, and add the 2 together to give total speed reference to the drive, with the line speed being the dominant factor, but fully adjustable.

The problems that I think I am having, is that the UW does not match the Line speed as accuratly on acc/dec

In my machines, I set the drive accel and decel rates as fast as possible, and manage all ramping in the PLC. A slow responding drive can cause PID overshoot and oscillations in the dancer. ......Using this modelling approach, with accurate numbers, you will be able to overdamp the PID and set the gains quite low for a very stable operation.

Modelling approach? How did you acheive this? I don't fully understand, I have also set the acc / dec rates to the max, and can control the acc / dec of the web by adjusting the speed of the pulling motor, ( this is still manual at this stage ).

I have not got any acc /dec control in the PLC as yet, I am trying to follow the tacho reference, whci is the determining factor for line speed.

OD=SQRT(L*t/(pi/4)+ID^2)

I am using the formulas from this thread http://www.plctalk.net/qanda/showthread.php?t=19333&highlight=Diameter+calculation
which seems to be pretty close at this stage, Okies formula is different, and gives a different result, I tried re-working the formula, and the structure is completely different?

I have put the old system in at this stage, production is needed, but hope to be back at it tomorrow. I will read up on Ron's links, and see if I can pick up anything else.

Thanks for your help, any other suggestions are most welcome also.

 

[scottmurphy]

Until you can control the unwind roll with the material taped down and get the roll surface speed to follow line speed (with the diameter temporarily fixed at the correct diameter and the dancer correction disabled) I would not attempt to run material through your machine. Get your unwind speed control right first, then enable the dancer and check it trims correctly, then run limited material then finally enable the diameter calculation.

What signals are you sending to the drive ? Speed demand/torque demand/torque limit ?

I will do this as soon as I get the chance. I am only sending speed demand to the drive.

The biggest assumption is that scottmurphy is attempting to implement a fully speed based (as opposed to torque based) dancer controller

This is correct, my understanding, is that most dancer systems typically use speed speed control, whereas load cell systems use torque control. I have another machine, with a servo running in torque control with load cell as feedback, that works reasonably well.

Assuming the drive in question has a properly tuned proportional/integral speed control loop, the drive will internally develop the torque required to suspend the dancer. Another way to look at this is, if we suspend the dancer, enable the drive at zero speed and then let go of the dancer, the drive will put out the torque required to keep the motor shaft at zero speed.

I think that I need to look at the speed loop in the drive, I have not touched this as yet, it is still at its default. This is where I wanted to calculate the torque, and work out if I changing the gearbox is going to be beneficial? I know it works with the old system, but the new method is exactly that, new.

 

[kamenges]

Simon makes some very good points here. That is a very reasonable order to proceed in.

Originally posted by scottmurphy:

The reason for this method, is so that the tension is taken up, and will hold with no dancer movement when there is a zero speed reference from the tacho.

A properly tuned PID will do this also. There should be no reason to disable the PID. A case could be made that you don't want any integral action with zero line speed. At the very least the proportional term should stay active. Even if you don't disable the integral term all that will happen is you MAY get a very slight waver about center. Once you pretension you shouldn't need to disable the PID.

Originally posted by scottmurphy:

Do you mean the integrator in the drive, or the PID in the PLC?

I mean the velocity loop integrator internal to the drive. It sounds to me like you need to send out some speed reference to the drive in order to maintain dancer position. A drive with decent velocity loop and an active integrator will eventually start running at that commanded speed. It won't stay still. If you can send a speed command other than zero to an enabled drive and the motor stays still that infers that either the velocity loop integrator is disabled or the motor speed feedback is not correct. with the integrator disabled the velocity loop proportional term will develop a torque reference proportional to velocity error.

Originally posted by scottmurphy:

I sort of think my calcs are right also, when you say 'bits/1000 motor RPM' are you meaning the accuracy of the reference that I am giving to the drive?

No. I stated that poorly. I mean look very carefully at all the conversion factors you have as you convert from linear material speed to spindle rotational speed. It comes back to something Simon said. In a system like this you need to be very confident that the spindle speed command based on line speed is as accurate as you can possibly get. It sounds to me like this portion of the total speed command is not right. This could be due to an incorrect product roll diameter calculation, an incorrect interpretation or calculation of actual web speed or an incorrect scaling between a plc velocity command and how the spindle drive interprets that same command, among other things.

Like Simon said, set the product roll diameter to what you know the diameter to be, disable the PID loop and diameter calculator so the diameter doesn't change, tape up the roll and run the line without pulling material through. The surface speed of the product roll should match the surface speed of your pull roll. If it doesn't, figure out why it doesn't. Until you know beyond a shadow of a doubt that this is the case you really can't move forward.

Originally posted by scottmurphy:

I then take a % of each, and add the 2 together to give total speed reference to the drive, with the line speed being the dominant factor, but fully adjustable.

The problems that I think I am having, is that the UW does not match the Line speed as accuratly on acc/dec

It's probably not just during accel/decel that they don't agree. Typically a system like this that displays accel/decel issues is relying way too heavily on in the integrator in the dancer position controller PID. It needs to rely so heavily on the integrator because the spindle speed based on linespeed is not correct. At steady speed the integrator comes up with a speed offset value that compensates for this error. The required offset changes too quickly during accel/decel for the PID loop to keep up with.

But the way, why do you only take a percentage of the speed command based on the line speed. This should be a solid calculated number. There should be no need to condition it with some percentage scheme.

Also, this was probably already stated but I missed it. What type of drive are you using? Is it DC, AC servo, AC vector, AC sensorless vector, AC volts/hertz, etc?

Keith

 

[scottmurphy]

I mean the velocity loop integrator internal to the drive. It sounds to me like you need to send out some speed reference to the drive in order to maintain dancer position. A drive with decent velocity loop and an active integrator will eventually start running at that commanded speed. It won't stay still. If you can send a speed command other than zero to an enabled drive and the motor stays still that infers that either the velocity loop integrator is disabled or the motor speed feedback is not correct. with the integrator disabled the velocity loop proportional term will develop a torque reference proportional to velocity error.

I just read up on the drive manual, and if not using an external feedback device, there are suggestions to change settings which I did not notice before.

A properly tuned PID will do this also. There should be no reason to disable the PID. A case could be made that you don't want any integral action with zero line speed. At the very least the proportional term should stay active. Even if you don't disable the integral term all that will happen is you MAY get a very slight waver about center. Once you pretension you shouldn't need to disable the PID.

This is another issue, I had this working, but changed to the holding method, with the parameters that I can change, I will go back to this method.

But the way, why do you only take a percentage of the speed command based on the line speed. This should be a solid calculated number. There should be no need to condition it with some percentage scheme

Should I only add a % of the trim value, and 100% of the Line speed ref? My thought was to establish the motor RPM required to drive at the line speed, based on the diameter, gear ratio etc. Then take 95% of this, 5% of the PID CV, and this will give the total ref to the drive.

What type of drive are you using?

I am using DC drives, on the existing system, the control boards are 30 years old, have been attempted to tune the firing etc, but the results were not optimum.

I am going to go back to the start, make sure my line speed ref is accurate, and that I can get the reel to rotate at the speed that I set. This is where I was wanting to calculate torque required / available, based on varying roll sizes, to see if changing the gear ratio would be beneficial.

Scott