Announcement

Collapse
No announcement yet.

Controlling a speed ramp using an encoder.

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts


  • Controlling a speed ramp using an encoder.

    I have a 3 ton roll of paper that I draw from to make small ~1000 foot rolls from. The paper is started onto a bobbin. No problem there.
    Then is accelerated to a fairly high rate of speed through most of the process to get it done efficiently.
    Towards the end of the process the motor pulling the paper off the source roll has to slow down, decelerating to almost a stop. hopefully just a couple of feet before completion where it crawls that last couple of feet to a stop.

    I have some vague ideas on how to do this in a rudimentary fashion, likely using a ton of trial and error coding.

    If anyone has done an encoder monitored speed control with a Productivity (P2) I'd love to hear what your general method was.

    I'm reading the encoder with an HSI if that's included in your technique.


  • #2
    Are you pulling the paper with a big AC motor? If so, controlled by a VFD? or are you pulling with a servo motor?

    Do you have a brake unit to slow down the paper roll and remove much of its inertia during decel?



    When you say "3 ton roll of paper", I am assuming you mean the roll weighs 3 tons, and is most likely quite wide (sort of a "master roll"), and is most likely anywhere from 3,000 - 10,000ft roll length when new?
    Last edited by MikeN; 06-10-2019, 11:27 AM.

    Comment



    • #3
      Hi Mike.

      Yes on all counts. It's a VFD. Paper is about 5ft wide.
      The big roll is not driven, the paper is pulled off of it by the bobbin spool.
      There is a drag brake that slows the big roll whenever the bobbin spool lets up.

      The old system had a "speed pot" on the operator panel that was the only speed setting. The VFD gets only digital go/stop. So once GO the drive ramps to the knob speed setting and once STOPPED the big roll slows at the drag rate, coasted by the VFD. Things are digitally finished off with a creeping jog command.

      The system now has an analog out from the PLC. I'm supplying the speed command to eliminate failing mechanical pots and to possibly allow automatic speed settings for, say, different product or spool lengths. I could or could not involve the dynamic control of the speed. I can see that being a problem in that if the VFD is called on to do much braking without an external braking resistor it will likely trip on DC bus over voltage. Furthermore, I have visions of a 100 feet of paper overrunning the bobbin. I should probably stick with a strictly RUN/COAST-to-a-stop scheme.

      The question becomes how do I reasonably calculate when to leave steady-state and coast to a stop? I assume there could be short rolls where I even need to transition to coast before reaching steady-state speed.

      Head scratching..

      Comment



      • #4
        It definitely is safer to coast to stop with a brake that is just a drag. Otherwise you may want a large braking resistor for the VFD to soak up all the voltage. However you may be able to do it this way:

        As long as all you do is the same paper, you should be able to easily determine how many rotations (and thus encoder counts) of where you are winding it up is equal to 1000'. You could set up your PLC program so that when you push "run" it will start the motor running and then when the encoder counts get to a specific number close to what correlates to 1000' then it starts to decel. You dont necessarily need to stop the motor and then start a creep cycle, but rather decelerate the motor slowly by manipulating the analog speed signal in the PLC. Plan out your decel slope so that you are almost at "stall speed/0 speed" right when you are about to hit that 1000' mark. So you would be under run conditions this whole time, never in coast to stop mode until you hit that end mark.

        So for example lets say you start and run at 10ft a second at full speed. You run that way until you hit 900', which you will know by the encoder position reaching that count. At this point you change the analog signal to run at 9ft per second. When you reach the count for 920', change the signal to 8ft per second. When you reach 940' change to 7ft/s. When you reach 950', change to 6ft/s. At 960', change to 5ft/s. At 970', change to 4ft/s. At 980', change to 3ft/s. At 990', change to 2ft/s. At 995', change to 1ft/s. At 999', cut speed and go to coast. At 1000', engage drag brake.
        Make minor adjustments to that as necessary to not overshoot your mark or change what speed you are going for the cycle. If that decel rate is too aggressive for the inertia of the roll (might be with a full master roll), then you can start decelerating at 850' instead and do it a little less aggressively.

        Since you really have no way of knowing the number of feet a short roll will be at the end of a master, the only thing I can think of is a manual or jog button button then the operator will press that starts ramping down speed when you hold it, and just runs at a creep value until the operator turns the cycle off at the proper time.



        When you are done with a cycle, the operator would get the new setup going and press a reset button that resets the encoder position back to 0. So you always start at 0 at the beginning of a cycle, and run up to the number of counts necessary for 1000' each time. As long as you are not using an extremely high pulse count encoder you should be able to stay within a 32-bit count limit just fine. If you are using a 360 count encoder for example then at the end of the 1000' you should only have a few inches between pulse counts which is probably accurate enough for you? Or go up to around 1000ppr encoder if you want to have pulse counts that would get you an inch or less resolution at the end of a roll. A 32-bit integer has something like more than 2 billion count limit, so you should be good to stay within that for a roll run.
        Last edited by MikeN; 06-11-2019, 11:17 AM.

        Comment



        • #5
          Hi Mike!

          It definitely is safer to coast to stop with a brake that is just a drag. Otherwise you may want a large braking resistor for the VFD to soak up all the voltage. However you may be able to do it this way:
          Totally agree. Thinking about it; since I can't even physically brake the master roll with the motor since the only connection is the paper web, it would be a disaster if the take-up reel actually slowed faster than the master roll. Picture huge gobs of paper piling up and getting yanked in half. The entire thing must be slowed and stopped by the master roll slowing against the dragging pneumatic disk brake.

          As long as all you do is the same paper, you should be able to easily determine how many rotations (and thus encoder counts) of where you are winding it up is equal to 1000'. You could set up your PLC program so that when you push "run" it will start the motor running and then when the encoder counts get to a specific number close to what correlates to 1000' then it starts to decel. You dont necessarily need to stop the motor and then start a creep cycle, but rather decelerate the motor slowly by manipulating the analog speed signal in the PLC. Plan out your decel slope so that you are almost at "stall speed/0 speed" right when you are about to hit that 1000' mark. So you would be under run conditions this whole time, never in coast to stop mode until you hit that end mark.
          I think this will work but I think the first thing I'll do is learn how the master roll drags it all to a stop maybe it's just 20 feet or something. Luckily, they have some other machines doing it all day. I suspect it probably has a pretty fast slow down.

          So for example lets say you start and run at 10ft a second at full speed. You run that way until you hit 900', which you will know by the encoder position reaching that count. At this point you change the analog signal to run at 9ft per second. When you reach the count for 920', change the signal to 8ft per second. When you reach 940' change to 7ft/s. When you reach 950', change to 6ft/s. At 960', change to 5ft/s. At 970', change to 4ft/s. At 980', change to 3ft/s. At 990', change to 2ft/s. At 995', change to 1ft/s. At 999', cut speed and go to coast. At 1000', engage drag brake.
          Make minor adjustments to that as necessary to not overshoot your mark or change what speed you are going for the cycle. If that decel rate is too aggressive for the inertia of the roll (might be with a full master roll), then you can start decelerating at 850' instead and do it a little less aggressively.
          Yeah this is the hard part. I think I might need some form of equation to handle the fact that rolls are frequently different lengths since the customers order what they want. It may be that short ones never can actually accelerate up to full speed which would mean I'd switch to coast-down at some different number of counts because the roll won't have as much kenitic energy and would stop way early.

          And the difference between a full master and an empty one will be large since the kinetic energy is tied to the diameter.

          Since you really have no way of knowing the number of feet a short roll will be at the end of a master, the only thing I can think of is a manual or jog button button then the operator will press that starts ramping down speed when you hold it, and just runs at a creep value until the operator turns the cycle off at the proper time.
          Good point. I need to ask them what happens with a running-out master roll.



          When you are done with a cycle, the operator would get the new setup going and press a reset button that resets the encoder position back to 0. So you always start at 0 at the beginning of a cycle, and run up to the number of counts necessary for 1000' each time. As long as you are not using an extremely high pulse count encoder you should be able to stay within a 32-bit count limit just fine. If you are using a 360 count encoder for example then at the end of the 1000' you should only have a few inches between pulse counts which is probably accurate enough for you? Or go up to around 1000ppr encoder if you want to have pulse counts that would get you an inch or less resolution at the end of a roll. A 32-bit integer has something like more than 2 billion count limit, so you should be good to stay within that for a roll run.
          I think I have this covered. Once the operator hits the two-handed switch I zero the encoder count. They've already entered the length. The encoder wheel has a 12 inch circumference and the encoder breaks that 12 inches into 100 counts. So, I just multiply the number of feet desired by 100 and that's the target count. I'll have to come up with that equation that takes the feet, and the peak speed, and the ramp-up time, and spits out a coast-point encoder count some percentage of the target count.

          Thanks a bunch for your thoughts!

          Comment

          Working...
          X