Rotary Knife /w Registration

Originally posted by Peter Nachtwey:

I doubt drums can be changed on-the-fly.
The lock position is basically the offset I was talking about.

Click to expand...

You are correct. The drum is the drum and it is what it is. My guess is janner_10 is probably concerned about the ability of the seal drum to stay exactly on speed. His initial thought seems to have been that he would need to change the cam starting point relative to the web to change the relationship between the seal point and the perf point. That could potentially happen every time a seal occurs. That really isn't necessary unless you specifically limit yourself to modification to the cam as your adjustment methodology.

Keith

janner_10 said:

...however, I would like to dynamically change the cut position or master lock position every cut, the bag length won't ever change mid-cycle but without this there is no checks or balances IMO, this is the part I am having trouble with....

Click to expand...

This is the part of the OP I don't understand.


It seems to me that, assuming speed and stretch are constant, the perf event has to be offset by some phase angle with respect to the master drum seal cycle, but otherwise the perf event occurs at exactly the same rate as the drum rotation (or twice, if there are two seal events per drum rotation).


Why does the OP want "to dynamically change the cut position ... every cut?"


I think @kamenges is right about the concern being variable drum speed; if that is the case then the seal event at the drum is irrelevant, and what it important is the seal (or registration mark) detection at [pos 1]. Variation in speed would still be a problem, but its effect is attenuated by the shorter web distance from [pos 1] to the knife, compared to the distance from the drum to the knife.



Getting the knife speed right at the perf event a related but different issue, of course.

Originally posted by drbitboy:

...the seal event at the drum is irrelevant, and what it important is the seal (or registration mark) detection at [pos 1].

Click to expand...

That is correct. As long as it is far enough upstream to detect, calculate and adjust without having to go Ludicrous Speed it would work. Having a base profile to work off of is really nice so the system doesn't have to be completely reflexive. Adjusting by a few millimeters based on a detected event as opposed to having to adjust by 10's or 100's of millimeters is a much less disruptive proposition.

Originally posted by drbitboy:

I think @kamenges is right about the concern being variable drum speed;...

Click to expand...

That one was a pretty big assumption but at least plausible. The other thing could be that janner_10 is just trying to make the solution more universal. It isn't much of a stretch to think that if you have the guts in to make ONE adjustment you would have the capability to make that adjustment every repeat. And if you can, why not? The wonder of automation.

Keith

kamenges said:

It isn't much of a stretch to think that if you have the guts in to make ONE adjustment you would have the capability to make that adjustment every repeat. And if you can, why not? The wonder of automation.

Click to expand...

Amen.


@janner_10: how many bag-lengths between [pos 1] and the knife fire position?

Thanks for all the replies - just got back from the site after a good recce today.

Anything to do with the printer / ink jet doesn't exist anymore.

It's late here now - ill update more tomorrow.

Again, I do appreciate the replies.

if you are not printing the registration mark how are you going to where the seam is and when to perf the seam
I guess you could use the seam prox and count from their but I would question the accuracy on that. The closer to the perf that you sense the better

janner_10 said:

It's late here now - ill update more tomorrow.

Click to expand...

looking forward to it.

I was curious where the dancer (accumulator? festoon?) comes into the control system.


I passed this thread's link onto my brother (35y+ web handling), and he said he hoped the speed passed from the master (22DH41?) to the others (16M091 and 16M051) is the speed reference, not the speed measured; the latter adds noise to the process, the former does not.

Originally posted by GaryS:

I guess you could use the seam prox and count from their...

Click to expand...

From janner_10's original post:

Originally posted by janner_10:

The main drum also has a prox which gets made every time the seal bars touch the bag, the assumption is this creates a registration mark as it also is a DI to the Perforator Lenze Servo.

Click to expand...

You are correct that there are challenges to this, which is why drbitboy stated shortly after:

Originally posted by drbitboy:
Does the problem boil down to stretching in the material between the main roll and the final nip rolls at 17N041 (I assume this is a few mil of plastic)? Does the "fotocel" at [pos 1] try to eliminate or mimimize the effect of that stretching as source of uncontrolled variation?

Click to expand...

That is definitely something that will need to be considered. Based on the diagram it looks like everything after the seal drum is just a draw/geared axis so, while there may be some stretch, it should be fairly consistent after the seal drum. Once it is accounted for the hope would be it should stay steady. But stretch is definitely a concern and it remains to be seen if it will cause issues.

Originally posted by drbitboy:

I was curious where the dancer (accumulator? festoon?) comes into the control system.

Click to expand...

Good question. Given what is shown you wouldn't think one would be required. I guess its hard to tell if this is intended to be a dancer or an accumulator from a machine design perspective. Generally speaking anything with that many rolls would be an accumulator, assuming the drawing is functionally accurate. I have seen similar machine concepts that were an index-seal/perf design that needed to stop to perform the seal/perf operation. These would use a multi-roll accumulator between the sections. It is possible this is a modified design that used to index but is now continuous and the dancer was just carried over. The other possibility is that the upstream side can change speeds much more quickly than the seal drum can and the accumulator provides storage for that.

Originally posted by drbitboy:

...is the speed reference, not the speed measured; the latter adds noise to the process, the former does not.

Click to expand...

He is definitely correct about that. I would use a passed reference wherever possible myself. Sometimes its just not. Peter Nachtwey needs to deal with real references quite a bit and has developed some fairly sophisticated tools to deal with the noise that a real reference provides. But whenever possible a virtual master reference is the way to go.

Keith

the web tension / dancer signal would go into the Outfeed drive as trim
their is also a position signal going in there as well (Dancer Position)

So I thought I would update this thread as the machine went back into production this week.

It was a long 4 weeks on site finishing off the code and tweaking to get everything working.

The parts I thought would be easy, turned out to be quite tricky. The parts I thought would be tricky were done in a morning!

It was one of those great sites to work on, whereby the operators and management really wanted it to work and gave you all the help and
assistance that they could.

The web handling and dancers sort of took care of themselves, I geared everything to the Virtual Master and I started off with just P loops set at 0.5 and a + / - 10% trim, soon realised that it wasn't enough so most are at + / - 20% and the main drum is at 30%, that part was just fit and forget.

Servo drives and AC Motors are a pain in the bum, as you couldn't just enable them if the safety circuit was healthy, as they all started to overheat quite quickly.

Firstly I cammed the perforator to the main drive, but this just wasn't working. After watching the other machines start up, they weren't cammed at all, just geared together.

There was a prox on the main drum and a prox on the perforator, I used these to work out the bag length and just altered the gearing slightly to keep them in sync if you like. The only other factor in the gearing was the offset, which has to be changed every time the drum changes size.

The actual wind up operation was fairly straight forward, all the 'spindles' were in torque mode and with a couple of lines of code to restrict torque if ActualVelocity > 1250, stopped all run away issues. The biggest issue was recovering from mishaps without stopping the machine.

It tool about 20 minutes for 2 operators to initially thread the machine up, so if you made a mistake and it all went Pete Tong, they would have to do it all again. This was the most frustrating part of the job.

Anyway here is a video of it running and thanks for all the input earlier in the thread.

https://youtu.be/FIe2wFHxtfk

Fantastic to see your results in the flesh.