PID Guru Needed!

[TonyK]

Hi fellow PLC gurus..... Here's the situation - I have a piece of machinery that unwinds product from a spool to be processed. THe tension on the product is maintained by a dancing arm setup that is attached to a pivot point. The angle of this pivot is the PV of a small PI loop, which controls the speed and direction of the spool motor. ie. angle changes in the pivot indicate changes in tension. The spool motor then incresae or decrease speed and direction to compensate for this change. The problem is - as product is unwound from the spool, the weight of the spool changes. THis leads to unstable control at either completely full spools or almost fully unwound spools. A potential solution is to have more than one set of PI parameters and swap in as needed depending on the current weight of the spool. Someone must have encountered this type of situation before - any help???

[Peter Nachtwey]

Yes, why would you ever think you are the first to have this problem? You are on the right track. A good motion controller should be able to adjust the gains on-the-fly as a function of the radius ( intertia ). Easy, No problem.

[rammin48]

In most of the winding applications I have dealt with I used a tension sensor rather than an angle sensor of the tension arm. The reason is the as the roll gets bigger inertia plays a factor on the winding and it is barely linear. As you have found it is difficult to tune and if you have a product change where mass varies you may have to start over. I have used tension sensors or rollers that measure the tension of the product as the PV. This makes tuning and winding easier. If the winding motor is undersized like many you may have switch from speed as the CV to a torque or current mode to maintain the same tension without over winding. There are several things you can do. Set up a set of tuning parameters for different diameters based on the angle. Then you will have to tune each step individually and move the new set of parameters into the PID as the diameter increases. This will take alot of time to get it right but once done should not need to be messed with again. Good Luck

[Peter Nachtwey]

It isn't linear. The interia varies with the radius squared. If you tune at one spot then you can make use of the fact that the inertia changes with the radius squared to estimate the interia on a continuous basis. As stated above, this will take some effort but once done it should be required again.

[TonyK]

Thanks for your replies Peter and rammin48. Since I am having problems primarily at the very beginning or end of a spool (depending on where it is tuned - there is some play in the meachnical setup) it sounds like the plan of attack is to ascertain three sets of PI parameters and use them based on an ultrasonic distance measurement or the like. Thanks for your help.