Bipolar PID question, Automation Direct plc

[rlawson]

I am doing a wire spooling application using an Automation Direct D-260 cpu, F2-8AD4DA-2 16 bit 8 ch in / 12 bit 4 ch out 0-10 vdc board.

The system will consist of a payoff unit using a 10hp ac flux vector drive, a dancer, and a takeup unit using a 2hp ac flux vector drive. What I am tinkering around with is this:

Using the speed pot for an overall master line speed reference signal (0-10 vdc) controlling the takeup speed. What I want to do is use the pid as a trim. For example if the takeup is running at 2000 fpm at 5 vdc reference, naturally the payoff needs to run at the same 2000 fpm. But assuming that 2000 fpm is 5 vdc reference on the payoff as well. The dancer (0-10 vdc) is the input to the pid algorithm (PV). I would like for the 5 volts dc going to the payoff to be made up of approx 90% speed pot ref and 10% pid. Specifically bipolar pid where if the SP is in the middle ie.. 32768=SP (16bit). I can cap the limit of the pid to around 10000 counts or 1.53 volts positive and -10000 counts negative. That way I would have from the 5 volts of reference going to the VFD a maximum of 1.53 of those volts would be from the pid, and the other 3.47 volts would be from the speed pot. What I am running into is by adding the maximum 10000 counts to the maximum value of the speed pot of 65535 (10v) that puts me over the limit. Conversely on the low end of the scale when the negative pid limit of -10000 is added to the speed pot reference I actually go below 0 and the counts roll over. I am currently using a CMPR instruction and watching if the converted Real value goes above or below the thresholds and doing an "ignore" values bit. Is there an easier way to do what I am wanting to do?

 

[rdrast]

First question, is, how are you compensating for diameter? Every bobbin / reel we run around here has at least a 2:1 build up, and some go up to 4 or 5:1. 10% dancer trim on a fixed line speed reference won't deal with that.

 

[rlawson]

First question, is, how are you compensating for diameter? Every bobbin / reel we run around here has at least a 2:1 build up, and some go up to 4 or 5:1. 10% dancer trim on a fixed line speed reference won't deal with that.

On my payoff unit I have a empty reel barrel diameter of 20 inches. On a fully loaded reel, the diameter is 25 inches. What I am thinking of doing is taking the average diameter between empty and full and scaling the speed pot reference to coincide with that, then add the negative or positive 10% or so PID to the scaled speed pot reference to get my final speed ref. Any suggestions? Thanks

 

[rlawson]

Anyone else have any ideas on this??

 

[Rich1955]

Well one thing I allways wanted to try was find the speed from full roll and empty roll, then scalling it through an ultrasonic sensor looking at the roll to try in maintain FPM.

 

[rlawson]

Well one thing I allways wanted to try was find the speed from full roll and empty roll, then scalling it through an ultrasonic sensor looking at the roll to try in maintain FPM.

Yeah, I thought about doing that, but it is not really an option that I can use. Here are the details of my system:

Material being spooled: 0.023 to 0.045 diameter nickel welding wire.

Payoff - 20 inch diameter empty barrel, max speed 2027 FPM
25 inch diameter full barrel, max speed 2534 FPM
1750 RPM AC motor, 4.52:1 gear reduction.
Takeup will use 2 different spools for wire (2 and 10lb)
1750 RPM AC motor, to be run at a max freq of 137Hz (4000 RPM), 2:1 gear reduction.
2# spool- 2.73 inch diameter empty, max speed 1429 FPM
3.875 inch diameter full, max speed 2029 FPM
10# spool- 3.75 inch diameter empty, max speed 1963 FPM
7.875 inch diameter full, max speed 4123 FPM
Dancer 0-10VDC input to PID loop.

Since this is a constant rpm application, the speed will increase as the effective diameter of the spool increases during spooling. The maximum speed will be limited to around 2000 FPM. Any suggestions are welcomed.

 

[Peter Nachtwey]

Yeah, I thought about doing that, but it is not really an option that I can use.

Why not? Can your real time estimate be off by more than your assumed average radius? I know I would do it. I would use the upstream speed to calculate feed forwards for the down stream spool. The more accurate this estimate for the feed forward is the less work the PID must do.

 

[rlawson]

Why not? Can your real time estimate be off by more than your assumed average radius? I know I would do it. I would use the upstream speed to calculate feed forwards for the down stream spool. The more accurate this estimate for the feed forward is the less work the PID must do.

I am using a speed potentiometer scaled at 1:1 for the speed reference on the takeup (2000 FPM). Going to the payoff speed reference is the speedpot value x (.757 on a 2 lb spool, and .667 on a 10 lb spool). I took the middle point of the takeup and payoff spools and calculated what the motor rpm should be. Any deviation from the scale should be trimmed by the pid, which by my calculation on a 2 lb spool should have a range from -1.93 to 2.44 volts (running at max speed), and a 10 lb spool should have -2.8 to 3.50 volts of pid to add to the speed pot reference for the final drive. I've thought about using a US sensor, but haven't made up my mind yet. I am exploring all options. If I can get the control system to function ok without it, I'll be ok. I have quite a bit of accumulation length for the dancer, an air cylinder with a 4 foot stroke and 7 sheeves. That helps a lot. Thanks for your input.

 

[JRW]

Have you made a decision on the drive manufacturer?

All of this could be done in the drive.

 

[Rich1955]

Yeah, I thought about doing that, but it is not really an option that I can use. Here are the details of my system:

I agree with Peter. You could use this for the gross change getting your speed close but not exact and then using an encoder on the wire to find exact FPM and compare that to desired to control thee miner change to offset the gross.

 

[rlawson]

Yeah, I thought about doing that, but it is not really an option that I can use. Here are the details of my system:

I agree with Peter. You could use this for the gross change getting your speed close but not exact and then using an encoder on the wire to find exact FPM and compare that to desired to control thee miner change to offset the gross.

I am using a SEW Eurodrive for the payoff (flux vector), and a Yaskwawa flux vector drive for the takeup. I do have an encoder on the drive sheeve of the dancer, and I am monitoring the FPM. Decisions, decisions..