OkiePC
Lifetime Supporting Member
I have worked on a bunch of winders, unwinders, re-winders, slitters, etc.
Many of them had issues like you describe. I found that their control strategy was oversimplified.
Bring tension into a PID, send torque command out. Adjust the gains, walk away. There was also usually a bit of PID mode control logic, but not quite enough...
For high speed lightweight stuff like spiral wrapping on a tire machine, this was plenty good enough. The range of the radius was small, and the rotor to load inertia was favorable for striaght tension control.
Others, were much higher ranging ratios on intermittent feeders, and continuous lines as well, but they all had an oversimplified single PID strategy in the PLC.
I improved most of them by doing what Peter said. I actually got the formula from this site to calculate the diameter from a known length of a spiral (i think it was Steve Bailey who gave me that nugget, thanks again whoever it was...) Then I had a known thickness (+- tolerances) so I could properly control tension by controlling velocity.
If you do your math right, and the tension is wrong then you must be going at slightly the wrong speed, right? At least that's my Lawn Mower Man perspective on it.
The torque was monitored, sometimes filtered, and depending on the speed of the line, was compensated for by adjusting the calculated radius. Actually is was a float called "Diameter Factor" and I used common sense to limit it's range and influence on the final speed command. Ideally it would 1.0, and I would let the PID move it between fairly wide limits at first, and those limits became my application specific tune-ables.
Some machines required a wide range from like .85 to 1.15, but most without severe stretch or shrinkage going on would never get outside of 0.95 to 1.05.
I applied the corrections at total roll length intervals, so the faster the winder surface speed, the more often its torque limits and diameter calc. would get updated. Basically that summing action re-centered the PID.
So if I was way off on my specification for thickness because of normal variations, no big deal the ratio detector summed back into the diameter calculation took care of that, and I could allow a wider range of compensation to start with partial rolls and quickly catch up to an accurate diameter calculation.
Some machines were slow moving and continuously fed, with jittery load cells, and we used analog filtering to help smooth that out. Other machines needed to react more immediately, so precise math was a must with high speed winders so you could limit the drive torque more accurately, and follow line speed through rapid accel and decel moves.
The diameter and the weight was used to alter the torque limit input of stuff like steel belt mother-rolls which could weigh over 4000 lbs.
Some machines had analog torque limit inputs through hardwiring, and others were messaged on the fly.
Then, if the film/rubber/steel belt material wads up or breaks, the torque limiting can keep you from from overreacting and running away. If you see a sudden dip or jump in tension that doesn't calm down right away, then fault and stop the process.
At the core of all this is a PI used more or less for "ratio detection" or "diameter calculation error", nothing more. You don't want to give the PI full control of the drive, you just want to limit it to minor corrections in your radius math. It will have the tension signal as an input, and the output was in corporated into the calculated diameter.
We had one machine that was an unwinder feeding big rolls of steel belt material with a large drive programmed in torque control to wind the web.
It was the same deal. The OEM used torque control, and then we were shredding webs and had run away motion.
What they did for this intermittent feed cycle from a winder, was use messaging to switch modes of the drives involved.
The beauty of using diameter in the formula is the ability to deal with line stops and starts. For a continuous process with few interruptions, it might be overkill.
I didn't have any fancy schmancy hardware, just good drives with analog I/O signals and a load cell, dancer, or loop sensor analog input. The PLC read the velocity of the line and totalized length (more simple algebra). Those drives were super accurate on their digital display and it was pretty good data even via analog wires. The drives themselves had encoder feedback which is what made calibrating so easy. If the display said 857 RPM and you got 4.57 volts, it would be ded nutzz.
JMHO
Paul
Many of them had issues like you describe. I found that their control strategy was oversimplified.
Bring tension into a PID, send torque command out. Adjust the gains, walk away. There was also usually a bit of PID mode control logic, but not quite enough...
For high speed lightweight stuff like spiral wrapping on a tire machine, this was plenty good enough. The range of the radius was small, and the rotor to load inertia was favorable for striaght tension control.
Others, were much higher ranging ratios on intermittent feeders, and continuous lines as well, but they all had an oversimplified single PID strategy in the PLC.
I improved most of them by doing what Peter said. I actually got the formula from this site to calculate the diameter from a known length of a spiral (i think it was Steve Bailey who gave me that nugget, thanks again whoever it was...) Then I had a known thickness (+- tolerances) so I could properly control tension by controlling velocity.
If you do your math right, and the tension is wrong then you must be going at slightly the wrong speed, right? At least that's my Lawn Mower Man perspective on it.
The torque was monitored, sometimes filtered, and depending on the speed of the line, was compensated for by adjusting the calculated radius. Actually is was a float called "Diameter Factor" and I used common sense to limit it's range and influence on the final speed command. Ideally it would 1.0, and I would let the PID move it between fairly wide limits at first, and those limits became my application specific tune-ables.
Some machines required a wide range from like .85 to 1.15, but most without severe stretch or shrinkage going on would never get outside of 0.95 to 1.05.
I applied the corrections at total roll length intervals, so the faster the winder surface speed, the more often its torque limits and diameter calc. would get updated. Basically that summing action re-centered the PID.
So if I was way off on my specification for thickness because of normal variations, no big deal the ratio detector summed back into the diameter calculation took care of that, and I could allow a wider range of compensation to start with partial rolls and quickly catch up to an accurate diameter calculation.
Some machines were slow moving and continuously fed, with jittery load cells, and we used analog filtering to help smooth that out. Other machines needed to react more immediately, so precise math was a must with high speed winders so you could limit the drive torque more accurately, and follow line speed through rapid accel and decel moves.
The diameter and the weight was used to alter the torque limit input of stuff like steel belt mother-rolls which could weigh over 4000 lbs.
Some machines had analog torque limit inputs through hardwiring, and others were messaged on the fly.
Then, if the film/rubber/steel belt material wads up or breaks, the torque limiting can keep you from from overreacting and running away. If you see a sudden dip or jump in tension that doesn't calm down right away, then fault and stop the process.
At the core of all this is a PI used more or less for "ratio detection" or "diameter calculation error", nothing more. You don't want to give the PI full control of the drive, you just want to limit it to minor corrections in your radius math. It will have the tension signal as an input, and the output was in corporated into the calculated diameter.
We had one machine that was an unwinder feeding big rolls of steel belt material with a large drive programmed in torque control to wind the web.
It was the same deal. The OEM used torque control, and then we were shredding webs and had run away motion.
What they did for this intermittent feed cycle from a winder, was use messaging to switch modes of the drives involved.
The beauty of using diameter in the formula is the ability to deal with line stops and starts. For a continuous process with few interruptions, it might be overkill.
I didn't have any fancy schmancy hardware, just good drives with analog I/O signals and a load cell, dancer, or loop sensor analog input. The PLC read the velocity of the line and totalized length (more simple algebra). Those drives were super accurate on their digital display and it was pretty good data even via analog wires. The drives themselves had encoder feedback which is what made calibrating so easy. If the display said 857 RPM and you got 4.57 volts, it would be ded nutzz.
JMHO
Paul
Last edited:
