Unintended motion

Hi,

I'm having a tough issue with a weld cell. It's a DX100 controller, H frame positioner and 3 robots at the back.

The problem is that very occasionally, if the rear gate is opened for any reason, upon closing it, and starting the robot again at the front of the cell, the operator side tooling axis will try to move. It will rotate about 5 or 10 degrees before the cell faults out. When it faults, I get multiple faults from the PLC and robot, which is correct, that axis should not be moving when it's at the front.

After an occurrence, I will look at the pendant and it is still in Master job. I was expecting to see it had jumped to some job, but it is still in Master. I am wondering if it left master, ran some job it wasn't supposed to and went back into Master by the time I get around to the back of the cell to look. I can verify that when this occurred, the pendant was in Master job before the incident, and had not been accidentally left in a different job.

Looking at the program structure, I cannot yet see a way that from Master job, the robot could have gotten to a job that would have just rotated the operator side tooling axis. It may be possible, but I just can't see it yet.

I guess my questions are as follows:

1) Is there an SOUT that I can track in the PLC that will tell me if the robot leaves Master job? I know there are SOUTS for "Top of master", but I want to know if it leaves master job altogether and runs anything else.

2) Is there any kind of logging of which programs ran in a DX100 controller? I looked around the pendant and in the logging function, but it just seems to log faults

3) Is there any way, other than a MOV command in a job, that an axis can be moved? Something in the CIO maybe? I don't think so, but maybe there is a way?

I have looked at everything else I could think of, there are macro jobs but they don't have any external axis as a control group, there is a system task, but from what I saw, it doesn't contain any move instructions (it did create a fault when the front axis moved, as it looks like it has been programmed to do so), the interrupt job list is empty. My assumption is that the controller somehow jumped into an unintended job and got back to master by the time I looked. How I'm not sure yet.

There is no OP station to allow for jogging. I did however figure out that we do in fact have the logging function and I assume either a JANCDYIF01-2E or JANCD-YIF01-3E card installed as stated in the logging manual.

I figured out that the log will show any jobs called that included motion commands, classified as a "start" event and if you push select it will open up and show which job was run.

Now I have to wait for this problem to happen again (I can't figure out how to create it) and see if there was a job called that shouldn't have been. I looked to see if SIN 65-72, 129-136 were used in the CIO, but they are not, so it is pointing more to a job being called that shouldn't be.

I added answers to your questions below in blue.

Quote from 95devils

Weird. See answers in blue. Do you have an OP Station with joysticks to jog the operator side fixture? No

Sweep jobs being PSTARTed instead of CALLed? Yes the sweep job is a single job which is PSTARTed, I'm not sure why they used a Pstart since it is in a single job. It does work however. What is the danger in using a Pstart for a sweep job?

What about putting an DOUT ON and DOUT OFF at the top and bottom of each job. Individual signal for each. When the alarm occurs you could see what outputs are on to track the job flow. Could be painful depending on how many jobs there are. How often does this occur? Sounds fairly repeatable. Possible to do, but I think the logging function should do the same thing?

Quote from 95devils

That's great that you have the logging function with board. That tells you darn near everything. Most people I see don't have that.

Does the operator side visually move in what looks like normal motion when this alarm occurs? Or is it "where the heck are you going?" The operator side axis moves about 5 or 10 degrees quickly before the safeties detect that it moved in the operator space when it shouldn't have and creates a fault.

Is there a PWAIT SUB after the PSTART in the master task? Yes there is a pwait sub for the pstart

When this occurs are there multiple jobs running at the same time, assuming so since you said they were PSTARTed? The PAGE key will display all running tasks. No, when it is started, the robot was in Master job, and the "starting" is really just the plc external starting the robot after the robot was stopped (from the rear gate being opened in this case). The sweep job should not have started because there were no parts loaded, so the "starting" of the robot should have just started master job running again without jumping to any other jobs.

Can you post the master, sweep jobs, and any other jobs in between these two whether CALLed or PSTARTed? I will post in a bit, just have to take a backup

Just me talking here that may ring your bell. Three robots sitting at a neutral position. Probably a VMH positioner. Sweep job A or B running, cursor on a hard WAIT instruction for the operator to say ok to sweep. Operator opens rear door, does whatever, and closes rear door. Restart system and positioner operator side tries to move. I thought the same but when I actually saw it happen, there was no sweep job running at the time, just sitting in master waiting. Rear door was opened to address a gas issue with the robot, thus stopping the robot. After exiting and restarting the robot, the movement happened at the front, only on that axis. Had a sweep job began, all 3 axes should have moved. It seems like it's running just an S1 job, which there are in the controller, but for it to run that by itself from master, I can't see how that could have happened.

Do the robots try to move also? Not that I noticed, but I will check the next time it happens.

Where is the cursor in the master task when the alarm occurs? Door acts as a e-stop. What are the values of S2C422, S2C423, S2C424, S2C653, and S2C654? Can't recall exactly which line it was at, but only master was running at the time I looked (after it happened), no subs running at that point. But something could have executed before I had a chance to go around the back of the cell.

422:1

423:0
424:0

653:1

654:50

Display More

Looked at the programs and I think I would have to upload more jobs here than is worth it. I will wait until this event happens again and check the log data to see if it actually ran a program to cause the motion. Will update with what I find.