Servo system with Mitsubishi PLC and equipment

[Paullys50]

For true synchronous control both the conveyor and label machine would need to be controlled by servos, by the same motion control module. You would setup what is called a "virtual master cam", this is how the sync is achieved. You control the master cam, and the two servos controlling the conveyor and label machine are driven off the master cam. This provides true synchronization. Much like a camshaft in a motor keeps all the intake/exhaust valves in-sync. Well, any mechanical machine.

The synchronization is so good that if you had 2 servos setup this way you could wrap scotch tape around the motor shafts, turn them on/off high speed/low speed..etc and the tape would not tear (assuming perfect alignment). Mitsubishi does this demonstration.

Here is another impressive demo.

 

[Peter Nachtwey]

That is nothing. It is easy to gear stuff if everything is pre-programmed and planned and the relative velocities and accelerations are known or 0.

Here is real synchronizing.
The person controls the dart and moves it as fast as they can towards the balloon. The dart will touch the balloon but not pop it. We have algorithms that allow us to match the position offset, velocity and acceleration so the dart will touch the balloon with a relative velocity and acceleration of 0 even if the dart is moving in an unpredictable way. Since the dart is being controlled manually the position, velocity and acceleration of the dart is always changing. There can't be any relative over shoot or the balloon will be popped.
http://www.mtssensors.com/?id=549
This is the unofficial video taken soon after we got the demo to work.
http://deltamotion.com/peter/Videos/Interactive mode.mp4
The controller was programmed over the internet. The person here that did the programming didn't see the official demo until he went a trade show.

Accelerating to 1.2 m/s is about what you see in the video. I believe the first video said 1.5 m/s was the fastest speed. Now everyone can see how about how fast the OP's motion will need to be.

 

[Paullys50]

That is nothing. It is easy to gear stuff if everything is pre-programmed and planned and the relative velocities and accelerations are known or 0.

I was simply pointing out to the OP, that what he saw in the Mitsubishi configuration that referred to "synchronization" is for configuring a virtual cam which served as the master, thus any motion cams linked to the master would then be synchronized. My post was a bit confusing as my explanation wasn't explicit to the OP's findings in GX Works3.

The OP's application is a simplified version of the demo I posted. If the conveyor motor and the label feed motor were both servos controlled by the servo module, a virtual master cam could be created to serve as the overall control of the two systems. The encoder on the conveyor would not be required and Start/stop/speed commands initiated to the master cam would ensure the 2 systems would be in sync.

The dart/balloon demo is quite impressive!

 

[qiwating]

just new come here and found the same question. thanks

 

[JRoss]

Sorry for the delayed response, I've been really busy with work.

The synchronous control is definitely what you want to use. Peter is correct that you need to offload this to the motion processor to make it work. You'll take the encoder into the connector on the simple motion module, and set it up as a master in the mechanical editor. The labeler servo will be a cam, which will be triggered by the registration sensor. I haven't used this in the FX5U hardware, but I have used the similar software in the rack-style QPLC. If you have any specific questions, post them hear and I'll be happy to take a crack at them.

A general comment on general comments.... The virtues or lack thereof of a particular piece of software aren't relevant to the discussion. The OP has been given a task and is trying to solve it using the tools he has been given. Being told what the software he has to use can't do isn't helpful.

Peter - The COP instruction equivalent is the BMOV (block move), though you need to understand the data structure to use it correctly. Labels are fairly new to GX Works and aren't handled very well, especially if you're used to a truly tag-based environment like C++ or RS Logix 5000. With Mitsubishi, everything in the background is bits and words. Assuming you have created a structure, you'll need to go into the global label declaration and specifically assign the data registers rather than let the compiler assign them. Then you can use the BMOV to do a bit-for-bit copy.

 

[doomsword]

I would like to thank you all, I presented my sketch of solution to boss and he said they pretty much came to the same idea (by that guy who was actually solving this, he's apparently on vacation until Monday). I will actually start working on it when that guy comes back, and he'll probably teach me (he's not at my location, he's at company HQ, probably will be a problem). He'll probably be my mentor, guy who used to mentor here was pushed out of company due some internal politics, guy who's sales engineer says it was a huge mistake, that guy had almost 40 years of experience and huge technical knowledge, he says I could learn so much from him.
My idea is like JRoss wrote, to use conveyor encoder as input axis, and use sensor signals to turn on/off virtual clutch when it's necessary. That synchronization manual helped me a lot, and idea of clutch by distance made me to check how clutch works in that virtual model in GX Works3. So thank you Peter.
Of course, I can't make cam tool until I get to understand physics of the system, probably I'm stuck with reading more manuals until Monday. I think I get the idea what they want me to be, to be real engineer engineer, to solve technical problems here, like that guy in company HQ.

 

[JRoss]

Glad to hear you have a mentor on this one! Have fun reading manuals. They're so much more interesting when you have an application you are trying to solve.

 

[ROAMER_AUS]

Povezi ih lancem

 

[doomsword]

Just an update on current status, I just got demo-case with same equipment and my task is now to solve this problem.
You gotta love this job.
EDIT: I also start to think this forum might be best place to get some real help, we're gonna be in touch.

EDIT2: We might as well start, when I read program with configuration from my OP, I got notification that firmware of Simple Motion and programming tool used for Simple Motion module (part of GX Works3) differ, in sense that some functions can't be supported with my version of tool and version of firmware on Simple Motion module. And those functions are being used. Who can update firmware version, where do I ask in case of such issues?

 

[doomsword]

I think I have solved it, it was about setting parameters of PID regulator and some other stuff using autotuning. It works now, but it sounds really bad. I think it's still not done, I'll have to explore control theory, what I hated back at college. Who'd know that phase compensation and advance, and parameters of PI regulators actually mean something in real life.

 

[morens]

same problem here

hello

doomsword , i working on a FX5 plc with a fx55-40SSC-S module. im working on a project for indexing with 2 axis.
but by the moment i just learn how to move in jog mode lol!!. i try to use the tables much of the configuration on the gxw3 for the module its the couse of the "issues".
Richg now im working in how to make a home return in order to use the data position table for the axis.

i hope you find your solution.

best regards

 

[doomsword]

hello

doomsword , i working on a FX5 plc with a fx55-40SSC-S module. im working on a project for indexing with 2 axis.
but by the moment i just learn how to move in jog mode lol!!. i try to use the tables much of the configuration on the gxw3 for the module its the couse of the "issues".
Richg now im working in how to make a home return in order to use the data position table for the axis.

i hope you find your solution.

best regards

Hello, I managed to get much understanding of that module + I'm getting two day training for that module in a week (I had few 'trainings' and I think it's just mostly for impressing people buying Mitsu equipment with flashy videos. But I'll make them spit out real knowledge). After a month or so dealing with it I can say that anything related to motion isn't easy.

We got that machine in my office, I managed to get it running, logic that drives the servo works. I managed to set synchronization, after coming to deal with painful shortcomings of that module:
I had to set a virtual axis just to unlock second high speed input. Then I came to realize cycle time of that module is 1.77ms, which might not be fast enough (they talked about response time of 1ms, I now think this is pure fantasy).

I managed to make everything run on module, mostly by using address mode for both virtual clutches, since you can't start and stop clutch using separate signal for each action, I had to make some tricks to make it run properly.

I use PLC to determine start/stop of clutches in advance, since real 'real-time' operation is impossible, I write that information 1 labeling cycle before it's executed since I don't have enough time.

And finally, yesterday and today I managed to set up some kind of conveyor belt simulation, where i use other motor to rotate synchronous encoder which simulates impulses from conveyor. And it works, those two things are in sync.

I managed to make it to run nicely up until 25-30% of required max speed. It still runs decently up to 35-40% of max speed. Last few hours of this work day I spent analyzing responses of system in MR Configurator2.

And I came to conclusion that 750W isn't enough, this thing can't make enough torque to counter dynamic torque generated when accel/decel-ing this machine. At least few kW in my opinion. Will ask boss about this tomorrow, I'm pretty sure you can't accelerate this thing in 1ms (or even 5-10ms) to 100% of max speed. Also this problem is analyzed by experienced Mitsu motion engineer in Poland (center of Eastern European hub), first thing he asked was how strong is our motor.


Feel free to ask me questions, I'm anyways intended to be a Mitsu tech support for my country.

 

[Peter Nachtwey]

Who'd know that phase compensation and advance, and parameters of PI regulators actually mean something in real life.

These courses weren't even taught when I went to college.
BTW, there are 3 things you need to know.
1 System identification
2 Pole Placement
3 Observe or Kalman filter control ( maybe ) but the first 2 are a must.

 

[doomsword]

These courses weren't even taught when I went to college.
BTW, there are 3 things you need to know.
1 System identification
2 Pole Placement
3 Observe or Kalman filter control ( maybe ) but the first 2 are a must.

I (should) know first 2, but 3rd is unknown to me. I have bachelor in Automation where we learned first two things (albeit it was done poorly, no one explained us how that esoteric mathematical theory can be applied to real cases).
3rd is taught at master in Automation, but I switched to Electrical Machinery and Automatization (which is about motors and how to run then with PLCs, inverters, how to design motors, structures for control of electromechanical systems etc., because I was afraid that I'm not getting taught anything that can be used for anything beside getting a PhD in Automation, and I'm not a PhD material. )

There is a class which I could enroll next semester, called Theory of Estimation, but from what I've heard it's executed poorly and it is really pain in the ***. They learn about Kalman filters there.

After two months here I'd say that there's nobody to teach me real stuff but me, I can get tutorial how to click stuff in GX Works 2/3, which anyone could learn by himself if he knows basics about PLCs, like I do (luckily, I had superb class where we learned a lot about PLCs, from wiring to implementing PI regulator in PLC).

P.S. how did you learn it without college? How is it possible, 'trainings' I'm getting are flashy stuff meant for those poor guys who operate machinery in 120 decibel environments.