PID problem or not?

daved

Member
Join Date
May 2002
Posts
19
I'm trying to control air pressure to brakes on a unwinding roll of paper. I have a dancer with a 0-10 v signal going to the plc. Also a pressure transducer measuring the brake pressure. The dancer should stay around mid point.
From mid to full back the air to the brakes is at a set minimum
from mid to full forward the air to the brakes increases linear to max.
I scaled the dancer pot input and the air to brakes input.
I then subtracted the air to brake from the dancer pos. After scaling
these should be linear. The result would be my error.
process variable = error
control variable = analog out to brake proportional valve.Now the dancer is very erratic because the brakes are overreacting.
Either I am going about this the wrong way or I don't have the PID tuned correctly. The processor is an Allen Bradley slc503 and the proportional valve going to the brakes is a Herion.
Anyone have any ideas or experience with this application?
 
I am not clear on the post are you using the PID in the SLC or did you write your own? If you are using the PID in the SLC in my experience it can be difficult for brake applications because when your error is zero you output is zero. (if some one else in the forum can tell me if I am doing some thing wrong I would appreciate it.) My limited attempts with using unwind brake and a SCL PID have been unsuccessful, (I also was under the gun). I ended up writing my own loop to give an output at zero error and it worked.

I used the dancer to make tension more air pressure on the dancer the more web tension.
The dancer is feedback device. Keep the dancer in the middle any deviation is your error. Take that error and scale to your brake.
I know this is not very clear, post again and I may be able to clarify some things
 
Hi,

From mid to full back the air to the brakes is at a set minimum

If I am reading this correctly, you have lost all control on your back side because your system lets go completely. It seems to me that your brake should only be set to minimum at just before full back.

from mid to full forward the air to the brakes increases linear to max.

This sounds about right except now it should be from just before full back to full forward.

There have been several threads regarding PID loop tuning and, as I recall, there are some pretty techniques for tuning listed in them. Here is one.
 
hello daved,as im used to dedicated systems to do what you want to do,i can see youve got a problem there.sounds like youre refitting
a stork contiweb unwinder,or could it be a martin unwind.anyway what jimtech was saying about not having a real zero ,but a small offset ,
just enough to let the pads rest on the disc,could be a help with the
the system,normally the running brakes when the machine is at speed need very little tension,only when there are speedups/slowdowns is when they have to get involved.if you really want to see how little you need to control the brakes,try this.take out the e/p transducer
and put a good hand ajusted air regulator in line with the brake,run the machine up and ajust the dancer zero with the regulator,every so often youll have to tweek it a little bit to restore the zero posicion,but youll be able to see the brake performance and what your pid really has to do.
 
I can't visualize your setup.

Can you provide a picture of some kind showing the physical layout and relative positions of Forward, Back, and "Just Right" on the Dancer?

This sounds like a fun-problem to me... of course, I'm not "up-against-the-wall" trying to make it work.
 
Seems like a complicated system you have there.

In order to keep the dancer in the centre, the brake would need to be on at all the time. By allowing the brakes to come fully off you are creating an acceleration of the paper roll by the action of the dancer dropping. The PID as to compensate for this and over reacts thus causing the dancer to go from a slow sensual dance into head banging.

I would scale everthing to between 0 and 4000. I can then say that 5 volts mid position is 2000

The set point I assume is mid point so should be 2000

The PV to the first PID is the dancer pot. Simce linear mid postion would be 5 Volts scaled to 0 to 4000 gives 2000

You now have an output to the ammount of braking required

Feed this to a second PID as the set point

The PV this time is the pressure transducer.

The output from this second PID is then used to control the proportional valve for the brake.

Since riding the brake is not a good idea, you could switch the output off to the proportional valve, wait until the pressure transducer goes to zero and then switch back on. Set up the PID's to work first then create the pulse train.
Remember to disable the second PID from changing during the period you have the output turned off.

Martin
 
Hello !
Sounds like you have a bad system there !!
If you have a unwinder you should really use a motor not a brake.
You should take in consideration that the braking power is propotional to the speed and the diameter of the reel.
If you would have a material that can withstand a lot of pull it would be no problem you only adjust the brake so that the reel doesn't run on it's own when the pull is realesed, but if you have paper that is not possible.
If it possible change the system to frequency controlled. Dancer gives speed setpoint or use tension control.
 
If you have a unwinder you should really use a motor not a brake.You should take in consideration that the braking power is propotional to the speed and the diameter of the reel.
If you would have a material that can withstand a lot of pull it would be no problem you only adjust the brake so that the reel doesn't run on it's own when the pull is realesed, but if you have paper that is not possible

I wonder how the 10 unwinds at my plant work then. We have Martins (all air) and Enkels, alot of electronics and there is a motor but not on the unwind shafts, the only item attached to those is an air brake. Those are on the gravure presses, the flexo's have their own system but they are basically the same..air is used to provide tension with a dancer bar being used.

I can not state specifics but when I get to work I will try to offer more details on how ours work but I think Steve may have offer part of it, the rest would involve the air pressure settings involved and how your system is designed. Does it also auto splice?

I will see what I can do to offer some pictures too.
 
Daved,

The biggest challenge to tuning a system like yours is the wide range over which it has to operate. With a 60 inch wide web at 2.5 PLI, you need 3600 inch-lbs torque when the roll diameter is 48 inches and 600 inch-lbs at a 4 inch core. That's a 6:1 ratio for just one roll of paper. Then include the fact that the brake was also designed to handle different web widths and tension settings. Then consider the difference in inertia between a full roll and the core. You begin to see how difficult it can be to find a single set of tuning values that cover the full range of operation. You need a high P gain to get good response when the diameter is large, but that same gain can cause instability as the diameter decreases.

From your description, it sounds like you're not using any bias in your PID. Bias is the only term in the PID equations unaffected by error. I would suggest adding a bias term that starts high and decreases with the roll diameter.

If you're using a disk brake with multiple pairs of pads, you might consider adding valves to control each pair of pads. At the start of the roll, use all of the pads. As the diameter decreases, you shut off the air to more and more of the pads.

Don't overlook the possibility of interaction between the brake system and the motor/drive that's pulling the paper.
 
Does the roll or "web stock" vary for this application?

Do you have 2 sides to this rollstand?

Do the operators enter setups for a web width/length change?

Do you have hmi on this machine?

Calculate the roll diameter for your loop. This will give you true full roll to core measurement, should smooth out the tension for you also.
 
I like Martin's answer, Steve has a possible improvement.

Steve is right about the effect of a particular pressure will change depending on the radius ( intertia ) of the roll. Adding a roll diameter input to the control system can provide information for the bias of Martins second PID. This roll radius must be squared and scaled before use. Using the radius-squared-scaled as a bias would be a great benefit if the radius of the roll changes rapidly. If the roll radius changes slowly then Martin's two PID technique will have time to adjust. It uses the inputs that daved has now.
 
OK I think I have my problem isolated. I am using a Herion proportional valve to control the brakes to the unwinding roll. What I found is that the valve is not very responsive to the input voltage when small voltage changes are made.The valve is always lagging the voltage by seconds. This seems to be causing the overcorrecting.
Any Thoughts
 
Do you have the manufacturer's data sheets for the valve? Is the behavior you're experiencing abnormal? If so, then you're probably seeing excessive stick/slip, or 'stiction'.

Can you clean the valve? Don't just run out and take it apart. Find out the manufacturer's recommended procedure for cleaning and lubricating it. This is a case where a 'bull in a china shop' approach can make things worse.

Are you supplying the pilot side of the valve with instrument air or simply teeing off the same supply that feeds the brakes? How clean is your shop air?

You say the valve is lagging the voltage by 'seconds'. Do you mean that when you apply a step change to command voltage, 'seconds' later there is a step change in the output? Or when you apply a step change to the command voltage the output takes 'seconds' to ramp to a stable value? If the latter, there may be a 'slew rate' adjustment on the valve that could speed up the response.
 
unwinder

daved if youre doing an electrical refit,have they also done mechancal
changes on the unit.have they changed the rollers in the festoon.this is very important,if they are not totally balanced you will never be able to adjust the brakes.as far as putting diferent sets of pads to scale the baraking,dont,youre complicating the thing even more.ive
commisionedlots of unwinders,meg,stork, martin.some more complicated than others,but all with the same principal.what type of system do you have to give you the tension,air clutch or magnetic particle,maybe these are giving you osscilacions the problem of web widths and paper type shouldnt be a problem,divide everything by 2 when its half web.problems only occur on less than 45 grammes
during splicing because of low web tension causes the dancer to go down quickly thus the web speed is limited.
 
The splicer is a MEG DME. No festoon just about 24" of dancer. I think I need some type of loop to control the pressure to the proportional valve with a transducer feedback for a quicker response time from the valve. Ideas?
 

Similar Topics

Dear all . I made a program for boiler tank using TIA V15 , the program consists of 2 PID blocks in OB30 . but the PID didn’t work automatically...
Replies
3
Views
2,628
Dear all . I made a program for boiler tank using TIA V15 , the program consists of 2 PID blocks in OB30 . but the PID didn’t work...
Replies
1
Views
1,114
Hello Friends; I am working on a GE Funac LM 90-30 software. I face the problem to set the CV value when PID in auto mode. The parameters are set...
Replies
13
Views
2,984
I had a problem on a hydraulic servo controlled cylinder, which I believe was caused by the PID loop and the cylinder reaching a hard stop at one...
Replies
10
Views
2,423
We have a machine we're trying to commission. It's a web converting machine. It takes webs of foil, wicking paper & a label and creates a chemical...
Replies
1
Views
1,306
Back
Top Bottom