Announcement

Collapse
No announcement yet.

Pneumatic brake control (tension maintainer)

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


  • Pneumatic brake control (tension maintainer)

    I am a plc self learner who is trying to control and maintain a certain set rope tension. The tension feedback is from a load cell is series with the rope. What will keep the desired tension "stable", is the pneumatic i/p valve (load cell > i/p valve).

    The process only runs for one minute at a time and can fluctuate alot, especially at the start (higher tensions). What is the best and simplest way to maintain tension with "rapid response". I HAVE A BRX PLC with analog inputs. I have tried simulating using pid control, but appears to not "correct/adjust" quick enough.

    Ex:
    200lbs is my set point. Initial start can be a high as 350lbs which stinks. I want to keep it in a set range +-10lbs. Fluctuations per 2 seconds are approximately +- 70lbs currently.

    Does the CLAMP function work for this? Any thoughts or suggestions.

    My load cell input is 0-10vdc (feedback) and i/p valve output (tension regulator) is 4-20mA.

    Happy Memorial Weekend! USA!


  • #2
    In my opinion, tight tension control at high material speed without some material accumulation is not possible (bear in mind I'm not primarily engaged in high speed material handling / converting).

    I once worked for a company where the mechanical engineers had a tractor drive that needed to be slaved to the main material drive. ~60fpm material speed and the “accumulation” was about 1/4” lateral deflection in the material path. Now, figure out the difference maximum and minimum path length, and translate that into time at 60fpm. That's how long you have to synch the slave and master drives. My boss asked if I needed the lever measuring deflection needed to be a different length, as if measurement could alter the fundamental relationship between time, material speed, and material accumulation.

    Tension in a system with no accumulation can vary widely almost instantaneously, so if it can be done quickly enough with controls as opposed to mechanical design, I believe it's going to be dedicated fixed-purpose electronic controller vs. load-cell → signal conditioner → PLC AI → PLC logic → PLC AO → IP → cylinder. WITH accumulation, you could use a counterweight, or better a counterbalance cylinder. The pneumatic counterbalance is better due to low inertia, and a frictionless actuator like a Belofram or graphite-piston cylinder would be better than a generic cylinder. Care should be taken to keep the moving assembly as low-inertia as possible, to keep the tension as constant as feasible while the accumulation is increasing or decreasing. If you want to program the tension from the PLC, you can still use an AO to drive the IP and set the tension. If the cylinder swept volume is too large, you may need a pilot-operated volume booster to supply and relieve enough air to maintain constant tension while in motion.

    But.....you said you're stuck with the mechanical design as is, so not sure how else to help....





    Comment



    • #3
      Originally posted by ControlsGuy View Post
      In my opinion, tight tension control at high material speed without some material accumulation is not possible (bear in mind I'm not primarily engaged in high speed material handling / converting).

      I once worked for a company where the mechanical engineers had a tractor drive that needed to be slaved to the main material drive. ~60fpm material speed and the “accumulation” was about 1/4” lateral deflection in the material path. Now, figure out the difference maximum and minimum path length, and translate that into time at 60fpm. That's how long you have to synch the slave and master drives. My boss asked if I needed the lever measuring deflection needed to be a different length, as if measurement could alter the fundamental relationship between time, material speed, and material accumulation.

      Tension in a system with no accumulation can vary widely almost instantaneously, so if it can be done quickly enough with controls as opposed to mechanical design, I believe it's going to be dedicated fixed-purpose electronic controller vs. load-cell → signal conditioner → PLC AI → PLC logic → PLC AO → IP → cylinder. WITH accumulation, you could use a counterweight, or better a counterbalance cylinder. The pneumatic counterbalance is better due to low inertia, and a frictionless actuator like a Belofram or graphite-piston cylinder would be better than a generic cylinder. Care should be taken to keep the moving assembly as low-inertia as possible, to keep the tension as constant as feasible while the accumulation is increasing or decreasing. If you want to program the tension from the PLC, you can still use an AO to drive the IP and set the tension. If the cylinder swept volume is too large, you may need a pilot-operated volume booster to supply and relieve enough air to maintain constant tension while in motion.

      But.....you said you're stuck with the mechanical design as is, so not sure how else to help....



      Based on what I currently have, what would be the best way to program? All I have is the load cell > i/p valve? Something to set a range similar to comparative contacts?

      Comment



      • #4
        Originally posted by e2dad View Post

        Based on what I currently have, what would be the best way to program? All I have is the load cell > i/p valve? Something to set a range similar to comparative contacts?
        A simple loop like out=sqrt(err)*gain might work. In the actual math you need to use the absolute value of the error, and then multiply the result by -1 if error was negative. Crank the gain up to the edge of oscillation, then back it off. It’s simple, but can be quite effective.

        Comment



        • #5
          Originally posted by BobO View Post

          A simple loop like out=sqrt(err)*gain might work. In the actual math you need to use the absolute value of the error, and then multiply the result by -1 if error was negative. Crank the gain up to the edge of oscillation, then back it off. It’s simple, but can be quite effective.
          Since I am a noob , could you explain this in layman terms. I am pretty fluent in excel. However, I never have used domore's MATH function. Would I have to somehow scale my analog input (0-10V, load cell) and analog output (4-20mA, i/p valve) with scale block first? If possible, any example screenshots would be appreciated...in a bit of a jammy .

          Happy Memorial Day!

          Comment



          • #6
            It's not a programming issue, it's a physics issue, and you can't change those laws just like Scotty can't.

            Significantly reducing the accel on startup should give whatever tension control you have a better shot at maintaining desired tension during ramp-up, but I'll bet you'll still miss your +/-10 lb target.

            Comment



            • #7
              Originally posted by e2dad View Post

              Since I am a noob , could you explain this in layman terms. I am pretty fluent in excel. However, I never have used domore's MATH function. Would I have to somehow scale my analog input (0-10V, load cell) and analog output (4-20mA, i/p valve) with scale block first? If possible, any example screenshots would be appreciated...in a bit of a jammy .

              Happy Memorial Day!
              You can use the built-in analog scaling, but in this example I used boxes, since I was testing on the Sim's process simulator.

              Tension (RX0) is just the scaled WX0 input. Valve (RY0) is the 0-100% output prior to scaling to WX0. Gain (R1) probably has units, but I don't know what it is. Error is R0.

              I'm a PLC developer, not a PLC user, so this may be a total bust. I have used it before to follow a real time position and it worked well. Gain is the key. Too low and response is bad. Too high and it will oscillate.

              Success will be a gain that doesn't oscillate and meets your performance spec. If your spec ends up needing a gain that is high enough to oscillate, this approach will not work. Only was to know is to try.

              Click image for larger version

Name:	SimpleLoop.png
Views:	74
Size:	113.5 KB
ID:	122929

              Click image for larger version

Name:	SimpleLoop2.png
Views:	51
Size:	62.7 KB
ID:	122930

              Comment



              • #8
                Originally posted by BobO View Post

                You can use the built-in analog scaling, but in this example I used boxes, since I was testing on the Sim's process simulator.

                Tension (RX0) is just the scaled WX0 input. Valve (RY0) is the 0-100% output prior to scaling to WX0. Gain (R1) probably has units, but I don't know what it is. Error is R0.

                I'm a PLC developer, not a PLC user, so this may be a total bust. I have used it before to follow a real time position and it worked well. Gain is the key. Too low and response is bad. Too high and it will oscillate.

                Success will be a gain that doesn't oscillate and meets your performance spec. If your spec ends up needing a gain that is high enough to oscillate, this approach will not work. Only was to know is to try.

                Click image for larger version

Name:	SimpleLoop.png
Views:	74
Size:	113.5 KB
ID:	122929

                Click image for larger version

Name:	SimpleLoop2.png
Views:	51
Size:	62.7 KB
ID:	122930
                Why did you scale twice? Is it because one is input (0-10V) and output (4-20mA)? My set point is 900lbs, where would I insert that information? Ex: My range let's say is + - 50 lbs, and set point 900lbs

                Comment



                • #9
                  Originally posted by e2dad View Post

                  Why did you scale twice? Is it because one is input (0-10V) and output (4-20mA)? My set point is 900lbs, where would I insert that information? Ex: My range let's say is + - 50 lbs, and set point 900lbs
                  The first scale is for the input, the second is for output. Strictly speaking, neither is required, just convenient. ‘Target’ is the setpoint.

                  Comment



                  • #10
                    And to be clear, I have very big concerns that your I/O speed/latency will make it impossible to do what you want. I only propose this as a simple and fast way to create a control loop.

                    Comment



                    • #11
                      Originally posted by BobO View Post
                      And to be clear, I have very big concerns that your I/O speed/latency will make it impossible to do what you want. I only propose this as a simple and fast way to create a control loop.
                      I appreciate your feedback!

                      Comment



                      • #12
                        Originally posted by BobO View Post
                        And to be clear, I have very big concerns that your I/O speed/latency will make it impossible to do what you want. I only propose this as a simple and fast way to create a control loop.
                        I/O speed...is that relative to the plc's processing speed?

                        Comment



                        • #13
                          Originally posted by e2dad View Post

                          I/O speed...is that relative to the plc's processing speed?
                          That’s a part of it, but in this case likely not the critical factor. BRX onboard inputs and outputs are fast, probably 5ms round trip. Your load cell and associated signal conditioner, as well as the control valve, both introduce lag. If the full path time from input to output exceeds the spec, you are completely toast.

                          Since 10lbs was the requirement, and 35lbs/sec is the change you are trying to control, then a lag of 10/35, or 286ms is the absolute limit, assuming control output’s ability to effect the result is instant...and it isn’t.

                          So if the sum of all times from input to output is greater than 286ms, this has zero chance of working. The lower it is below 286ms, the greater the chance it *might* work, but reality is rarely as good as theoretical limits. I’m not an app guy, so my opinions are dodgy, but I wouldn’t attempt it for anything more than about 100ms end to end.

                          Maybe one of the real app guys here can explain why I should stick to developing PLCs.

                          Comment



                          • #14
                            Originally posted by BobO View Post

                            That’s a part of it, but in this case likely not the critical factor. BRX onboard inputs and outputs are fast, probably 5ms round trip. Your load cell and associated signal conditioner, as well as the control valve, both introduce lag. If the full path time from input to output exceeds the spec, you are completely toast.

                            Since 10lbs was the requirement, and 35lbs/sec is the change you are trying to control, then a lag of 10/35, or 286ms is the absolute limit, assuming control output’s ability to effect the result is instant...and it isn’t.

                            So if the sum of all times from input to output is greater than 286ms, this has zero chance of working. The lower it is below 286ms, the greater the chance it *might* work, but reality is rarely as good as theoretical limits. I’m not an app guy, so my opinions are dodgy, but I wouldn’t attempt it for anything more than about 100ms end to end.

                            Maybe one of the real app guys here can explain why I should stick to developing PLCs.
                            The brx model I am using is BX-DM1E-18ED13. It says Minimum Pulse Width = 0.5 μs - High Speed.

                            Since I have two different types of analog signals, would the negative for both be used in terminal "1C" or common in the W (analog) terminal block section of plc?

                            Comment



                            • #15
                              https://www.automationdirect.com/adc...bx-dm1e-18ed13

                              Comment

                              Working...
                              X