Wile E Coyote Posted April 18, 2007 Report Posted April 18, 2007 I have installed 3 Mitsubishi 2Kw servos on an envelope press. The servos are HG-SF2024 and the Amps are MR-J2S-200B4. I am using them with a Q172 motion controller and the PLC is a Q02. The software is MT Developer for the motion control and I am running them in "Virtual Mode" as a "Draw Control". Everything works fine at a constant speed. The problem is when there is any acceleration or decceleration, all three servos do not follow the encoder exactly. There is a lag which moves the registration of the envelopes. It does correct after a short while. Has anyone had any experience with this sort of application. Mitsubishi people here are not able to help too much. Any help would be much appreciated. Quote
Peter Nachtwey Posted April 18, 2007 Report Posted April 18, 2007 I have found them clueless too. 1. What is the resolution of the reference encoder? The higher the better. 2. Are you using velocity and acceleration feed forwards? 3. How well do you have just the PID tuned? How do you know? 4. Can you slow down the rate of acceleration and deceleration of the reference encoder? 5. How bad is the error when accelerating and decelerating? Can you graph it? Quote
Wile E Coyote Posted April 19, 2007 Author Report Posted April 19, 2007 1. The encoder is a MR-HENC which is part of the package which comes with the Q172CPUN and encoder module Q172EX. The encoder plugs directly into the Q172EX. The resolution is 16384 PLS/Rev. 2. I am using "speed control" in the "Virtual Mode" of MT Developer programme SW6RN-GSV22P. This is what Mitsubishi recommended. If you look on the Mitsubishi web sight you will see how impressive the mock up of all this looks. There is an example of "Draw Control" which is exactly what I am trying to achieve. 3. The problem with all the parameters and PID settings is my problem. The manuals and help files are as clear as mud. Running in this mode and following the encoder I am not sure which parameters and settings are actually effective. 4. The accel and deccel is 8 seconds and 5 seconds, really nothing fantastic. 5. The movement on the paper of the envelope is about 7mm. If you can visualise a line which is the score for the fold, on acceleration the paper lags by about 7mm and then after 25 envelopes it has returned to its original position. In other words one can see the fold and the score move. This is at about 300 envelopes/minute. We normally run at 1000/minute. Have'nt even seen how bad it is at higher speeds. Thanks for your reply. Pulling my hair out because of the lack of help from Mitsubishi. Quote
Peter Nachtwey Posted April 19, 2007 Report Posted April 19, 2007 7 mm is awful. I would first tune the system up doing point to point moves at the same speeds at which you run in operational mode. This way the gearing is factored out and you can concentrate on the PID gains and feed forwards. Do you have a graph showing the target position, actual position and control output? It would be helpful. This should be easy. Is it possible that there is a filter on the encoder causing the phase delay? Quote
FNMdeJong Posted April 19, 2007 Report Posted April 19, 2007 I assume you use the mechanical designer for the virtual application. Do you also use, like the example a speed change gear? Check this smoothing parameter of the transmission module. Use automatic tuning for the drives (servo parameters, no PID settings to do when you use this and No feed forward gain) and try to increase the response setting. The acceleration and decceleration are resp. 8 and 5 seconds. What do you mean by this. Is this the acc and dec of the MR-HENC (reference speed), or are this the parameters in you're motion application??? If this are the values in your drives then: Set the acceleration and deceleration in you're application to 0. A speed change of the MR-HENC, will otherwise be followed up by an acceleration time of 8 seconds?? Can you monitor the command pulses, feedback pulses and droop pulses? Quote
Wile E Coyote Posted April 19, 2007 Author Report Posted April 19, 2007 Correct, I am using the mechanical designer. I am using a virtual encoder attached to the main shaft with 2 continuous gears and one spindle gear driving 3 output rollers. The first gear is set to a ratio of 14/1 and the second gear is set to 12/1 and the third gear is set to 8/1. The three axes are connected to three rollers for draw control. The acceleration and decceleration I mentioned is that of the main drive which drives the main shaft to which the MR-HENC is connected. I am using "Auto Tuning 1" and have the SV Response setting as high as I can before the Amplifiers come up with an error. I can monitor the command pulses and droop pulses. So far the droop pulses are about 13000 or there abouts on speed change. My problem is I need to know what parameter settings I can adjust which have an affect on the drives in "Virtual Mode" Many thanks for your reply. Quote
Wile E Coyote Posted April 19, 2007 Author Report Posted April 19, 2007 I do not have all the right gear to connect up the "OscilloScope" to the motion controller to get the actual lines of how they are following etc. Thanks again for your reply Quote
Peter Nachtwey Posted April 19, 2007 Report Posted April 19, 2007 If so then the drive is guilty of limiting the accelerations and decelerations. Remove this limitation. There should be some errors if the drive will not do what the controller requests. A good motion controller should have an oscilloscope feature built in and would generate errors if the drive will not do what the controller requests. Quote
FNMdeJong Posted April 19, 2007 Report Posted April 19, 2007 So, then this is a good motion controller Quote
Wile E Coyote Posted April 21, 2007 Author Report Posted April 21, 2007 I have put a firecracker up Mitsubishi's jib. All of a sudden I have all sorts of people from Mitsubishi swarming all over promising that all will be rectified by Monday evening or they will admit their equipment they recommended is not up to the job. I will post the outcome. Quote
Peter Nachtwey Posted May 4, 2007 Report Posted May 4, 2007 This is too bad. Apparently not. I find it hard to believe the Mitsubishi controller couldn't do the job. I do believe the tech support and the debugging tools were poor so the correct solution couldn't be found. I think we can do much better. I have videos if you are interested. It would be good if you had videos too. They are so easy to make now with todays cameras. Quote
TimWilborne Posted May 4, 2007 Report Posted May 4, 2007 Glad to see you finally found the right person. There are some pretty sharp cats up there at Mitsubishi's technical support. If there is a way to do it, they'll get you straightened out. Edit - I just noticed your in the UK, I can only speak for support in the US Quote
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.