Servo Position in Paper Printing Application

[Patty2001 0]

Hi All.... I have paper printer machine 4 colour using 4 axis of MPL servo motor, kinetix 6000 and compactlogix sercos motion controller. Machine speed is 150 MPM. Printer roll #1 is master and roll #2, #3, #4 as a slave. I use MAG on roll 2 -4 to synch to master speed and position. I have a registration problem ( can't control registration) so it moves. I plan to replace MAG with MAPC. I need advise or comment from any one in this forum.... Thank you for advise. Patty

[michael G 2]

Leave the MAG there when you know the Registration distance that each slave must move use a MAM incremental (advanced MAPC ) to phase shift (change the registration position) by doing this it works at any speed and while accelerating and decelerating

[Patty2001 0]

Do you mean that if registration distance is 0.05 mm so will increase the slave position let say slave actual position is 100 mm so if will move to 25+0.05 mm? Also after we change this new slave position, we need to look to master position. What do you think if after registration change on slave position then use MAPC to lock slave position so slave_new_position = master_position + registration_mm ?

[Patty2001 0]

Do you mean that if registration distance let say 0.05 mm so will increase the slave position with slave actual position is 25 mm so will move to 25+0.05 mm? Also after we change this new slave position, we need to lock to master position. What do you think if after registration change on slave position then use MAPC to lock slave position so slave_new_position = master_position + registration_mm ?

[michael G 2]

hmmmm it seems to me that you do not understand the concept that I am proposing MAG all Slaves to the master - Use the correct ratio so that for one index of the Master you get one index of the slave (An index is the distance from one registration mark to the next registration mark) Now when you turn the master one index (MAM) the slaves turn one index exactly. Rockwell permits you to have multiple commands active at the same time on the same axis - This means that we can use MAM, MAJ, MAG and MAPC all at the same time and get some funny looking profiles In your case the MAG means that no matter what speed the master moves at the slave moves one index per master index The NEXT problem is that the registration marks on the slave do not align on the master registration marks Eg Slave 1 is always 5mm behind where it should be as the master rotates To correct this use a MAM incremental distance 5 mm - DO NOT STOP the MAG What the Internal Rockwell motion task did was to add the 5 mm move on top of the existing MAG. Effectively you are adjusting the phase of the slave Does this make sense to you? Edited 9 Sep 2011 by michael G

[Patty2001 0]

I got it. I got registration distance of printer roll #4 is 1 mm, roll #3 is 0.5 mm, roll #2 is 2.5 mm. I move these position using MAM so it moves to target position. Ratio beteen master - slave is 1.05. The registration is maintain in same position with 55 MPM if I increase line speed to 65 MPM, registration of each rolls start to change randomly between 1 - 2 mm. Any idea? Thank you.

[michael G 2]

Great now we are into the commissioning of what is going on The Main Point is the Trend is your friend, eg I trend Position, Vel, Torque, Registration you are looking for patterns to see what is going on - Is the motor Torque high (80% or above) while running fixed speed 1) Servo tuning - Must be correct before you look at the application Master axis must be smooth and follow the reference correctly - trend Position and Velocity Error and the Registration must always be 1 index no matter the speed or acceleration. Slave Axis - Velocity and acceleration feedforward usually is 100% (especially for your application) Slave Axis Registration when running independently - Registration must be 1 index no matter the speed or acceleration 2) does the registration running at fixed speed change (Try a bunch of different speeds eg 1/4, 1/2, 3/4 Top speed) - Monitor it for a while (trending the registration) If the Long term trend is rising or falling then the MAG ratio is incorrect If there is High variation but the value stays constant then look at how you are measuring the registration and verify the servo tuning 3) Does the registration offset have a fixed speed component - ie can you see a repeatable pattern that follows speed eg 1/4 speed Reg is 5 mm 1/2 speed Reg is 10mm 3/4 speed Ref is 15 mm Check axis properties Master Delay compensation and MAG is using command not feedback 4) How are you measuring the registration? A Picture with the design concept would be good plus how the signal gets into the program maybe i am on the wrong track - I am use to commissioning new sites. Is this a New site or a existing site if existing then how long was it running before you had troubles Can you post the PLC Servo code? Zip file of the ACD

[Patty2001 0]

Hi, Thank you for your reply. Following is things that done. 1. Tune Master axis with position loop configuration, position error auto tune and position gain = 151.778, integral = 5.77, velocity gain = 389.321 2. Tune Slave axis with position loop configuration, velocity and accel feedforward autotune and position gain = 295.67456, velocity gain = 764.57263, velocity feedforward = 100%, accel feedforward = 100% 3. Motor Positiv torque limit is 70%, constant torque = 100% 4. Measure the registration with MAG ratio = 1, registration distance is 1 index to master. 5. Run the machine with 30 MPM, register seeting using MAM instruction (Roll #1 is master, orange colour), roll #2 is red colour, roll #3 is green colour and roll #4 is blue colour. Orange, green and blue is stable with variation arround 1 mm, red Colour is not stable (vary approx 4 mm -randomly). 6. If we increase speed to 80 mpm, during accel, registration is change then after steady speed, registration on roll #3, #4 to #1 is stable with variation 1 mm (oscilate arround 1 mm). 7. transmition from servo motor to printing plate roll and ink duct roll is using driven gear. Plate roll circum is 755 mm and servo roll circum is 500 mm. 8. Rewind tension is very stable. We use two nip rolls, one before printer and another after printer. What do you think that this variation is cause by mechanical isseu or software issue? For servo configuration, what do you think is we use dual command loop (velocity and position command instead of posiiton loop? Thank you for your advise. Patty

[Patty2001 0]

Hi, Thank you for your reply. Following is things that done. 1. Tune Master axis with position loop configuration, position error auto tune and position gain = 151.778, integral = 5.77, velocity gain = 389.321 2. Tune Slave axis with position loop configuration, velocity and accel feedforward autotune and position gain = 295.67456, velocity gain = 764.57263, velocity feedforward = 100%, accel feedforward = 100% 3. Motor Positiv torque limit is 70%, constant torque = 100% 4. Measure the registration with MAG ratio = 1, registration distance is 1 index to master. 5. Run the machine with 30 MPM, register seeting using MAM instruction (Roll #1 is master, orange colour), roll #2 is red colour, roll #3 is green colour and roll #4 is blue colour. Orange, green and blue is stable with variation arround 1 mm, red Colour is not stable (vary approx 4 mm -randomly). 6. If we increase speed to 80 mpm, during accel, registration is change then after steady speed, registration on roll #3, #4 to #1 is stable with variation 1 mm (oscilate arround 1 mm). 7. transmition from servo motor to printing plate roll and ink duct roll is using driven gear. Plate roll circum is 755 mm and servo roll circum is 500 mm. 8. Rewind tension is very stable. We use two nip rolls, one before printer and another after printer. What do you think that this variation is cause by mechanical isseu or software issue? For servo configuration, what do you think is we use dual command loop (velocity and position command instead of posiiton loop? Thank you for your advise. Patty

[michael G 2]

Ohh this is getting very hard very quickly - You are asking me to determine what is going on remotely with no program and no trends, No idea of PLC scan time or what else is being controlled by the PLC. We have advanced though - you seem to have improved the steady state error however for a printing press you are still out by a factor of 100 or 1000 you have also got the gearing ratio to 1:1 which is what I would have expected. How did you determine the scaling of the master and slave axis using the RSlogix Calculate ? - for a printing press I would be using micron not mm Motor Positive torque limit 70% - this number seems wrong - why this value? To assist I need at least Trends showing at least servo Position, Velocity, Position error and torque You give me the Auto tune values but do not show me a trend of that axis running by itself Again the first step is to be happy with the servo tuning - Each axis Must be able to do a MAJ at 80 MPM with the Position Error near 0.001 mm No Paper Then Each axis must be able to accelerate and decelerate with Position error near 0.001 mm get this right first Anybody else got printer experience - mine is limited and a long time ago so I cannot remember the accuracy needed

[Patty2001 0]

Hi, I forgot to save or print screen trend, so i will send to you. We will try this unit again on next shutdown during our maintanance day. I will take all data needed. Attached is our logic for your review. For 70% torque, if we used below 50% torque, servo motor is not enough power to turn printer roll so why we increase to 70%. Do you mean that this torque is to high? PrinterLogic.pdf

[michael G 2]

Torque : I have the Torque limits from about 150% to 250% - Positive, Negative and bipolar By limiting the Torque you are not permitting the Servo to use all its available power to maintain position. In your application position is more important than torque, So that you do not tear the paper when correcting the registration correction is usually performed over some ratio of index distance depending on the size of the correction Typically I only limit torque in winding applications, stationary holding situations or drive train Torque limitations (specified by the mechanical engineer). If you can please Zip the ACD file up and post The PDF does not have all the property settings of the servos, motion group update time Michael

[Patty2001 0]

PapperPrinter_V1A.ACDHi Michel, attached is our logix files. We will adjust torque to its default (+/- 150%) What do you think if we use Dual Command Servo Loop (Velocity Command and Position Command) instead of Position Loop? Please advise. Thank you. Patty

[michael G 2]

If you cannot get Position loop to work then using Dual Command will not help I think that Dual command is a better configuration but have not had much need for it. Will look at the Program later thank you

[Patty2001 0]

Hi Michel, any comment or advise for printer program? Thank you. Patty

[michael G 2]

oh $#!* - My evaluation is you are miles behind in understanding the motion system basics - Get Professional help from your RA distributor and some training Also perform a Motion analysis on the system I think that your loads are wrong - free download from rockwell web site Motion Analyzer Problems that I see in a very short review of your program Axis Properties - The BIGGIE Tuning The motor load inertia ratio gives you an indication of how control able your system is - I aim at less than 3 for positioning systems. Printer Roll 1 - 4 this is marginal - Peak Torque of 106% I would prefer to aim at 150% Peak Torque Printer Roll 2 - 7 - I would regard this as on the High side Note as Peak Torque is a 63% - Either the GB is wrong or bigger amplifier is needed (eg BMO2) Printer Roll 3 - 23 - Way too high This is reflected in a Peak Torque of 63% Either the GB is wrong or a bigger amplifier is needed (eg BMO2) Printer Roll 4 - 25.5 - Way too high This is reflected in a Peak Torque of 63% Either the GB is wrong or a bigger amplifier is needed (eg BMO2) IF you manage to get acceptable position control at fixed speed then the master MUST use full S Curve motion so that the slave axis can keep up AND the Maximum Accel /Decel rate CANNOT excede the worst slave Accel/Decel rate WARNING - There is a high possibility that the servo's will be PULLED by the paper nip rolls - this will cause registration skips Sercos Scaling you are wasting scan time doing the scaling yourself - Get the Axis to do the scaling I prefer to use real units eg mm or degrees - it means that the values that you look at make sense. Also to improve accuracy use the largest numbers possible eg use 1000 for 1 mm or 1000 for one degree Use the "Calculate" button on the "Drive/Motor" Tab Realtime axis information - Position and Velocity Feedback are not required (automatically updated) I use Position Error and Torque Feedback Also get rid of the GSV's on the Axis - that was required in earlier versions but not version 18 Loop Configuration Fix the Torque Problem first Either Position Servo or Dual Command Servo will work [*] you are using a Compact Logix Processor L43 - Control Logix L7x may be better for your application [*] Sercos update time is 2 ms Set to 8MB Baud Rate (do not allow 4 MB Baud rate) 0.5 ms update (may have to be 1 ms but documentation says that for K6000 this is ok) [*] Motion group Coarse update time - currently 2 ms Alarm is Disabled - Monitor for this error - it gives you an indication that something major is wrong initially set to 4 or 5 ms - then monitor the Motion task Attribute and target Max time to be minimum of 1/2 the coarse update period ( eg Coarse update period of 5ms then Max is 2 to 2.5 ms) Remember to check this after adding motion instructions to the PLC program. The reason for the 1/2 is to allow time for the rest of the PLC program to execute. [*] Firmware Make sure that you have the latest firmware for V18 2094 get to 1.111 - has an effect on the sercos recovery time [*] Scan Order and priority Maintaining registration lock is the most important thing that you are doing However Registration Lock can only be updated once per index so one way is to put the code into a 10 ms task and have everything else in a continuous task Another way is a Event task based on the motion group task - (Affects coarse update time) and everything else in a continuous task [*] Trends Again I implore you trend the torque - In your case if the Torque flatlines at the Max Peak while running then you are asking too much of your mechanics Typical trend that I use for tuning Note: as you are using Rev as the axis units you cannot see if the tuning is ok on the trend Tune_PR1.zip All of your rolls should have the same units and scaling Trend the following (Master.ActualPosition - slave.ActualPosition + index) Mod Index - this is the Phase (Index is 360 deg, 1 Rev or circumference in mm) It Tells you how closely the Slave is following the master (Note you may have to learn about motion group strobe and Axis.StrobeActualPosition to get this correct) [*] Coding Tips Where possible use MAG using "Command" (I only use Actual if the reference is an encoder) Never Engage a MAPC with a straight line Profile while running - Puts Huge Accel force onto the slave axis Registration Position Change I do not see how you measure the registration error - It looks manual On this System I would use MAM incremental, Profile "S-Curve", Accel and Decel Jerk 100, Jerk Units "% of time" Learn About UDT's and use them - they self document and permit code reusability Patty2001 - for the next two weeks my time to assist will be greatly limited by a project I am commissioning. I also feel that in your case you need some one more local who really knows what they are doing to assist you (I.E. a Site visit) - I know the the Rockwell guys in Singapore are experienced

[Patty2001 0]

Hi Michel. thank you for your advise. We got printer registration with good stability up to speed 190 MPM. Following action was taken. 1. Change the Speed to 8 MBPS and sercos scan time to 0.5 MS 2. Found mechanical back lash on Gearbox and replace with new gearbox 3. Adjust torque limit to +/- 150% 4. Re-tune all axis and get consistant Position Error maximum 80 Micron 5. Run the printer and found actual registration arround +/- 200 to 300 micron (measure manually) Thank you for your support. Patty

[michael G 2]

Glad you got it all working - thank you for posting what finally happened Edited 1 Oct 2011 by michael G