Issue with Y-axis stalling/binding consistently

[jonathan lustenader]

I have a program I am trying to run that on the exact same spot every time, the Y-axis stalls and the position in that axis is lost requiring rehoming the machine and trying again (the y-axis position shifts several inches).

Project is cutting some poster board to hold postcards. Laser freakout appears when trying to transition to the vertical highlighted area in this picture:

Annotation 2020-01-05 160601.png

Stalling is the high pitch noise roughly 3 seconds in.


Any thoughts on what could be causing this? Drawing was made in AutoCAD then imported into Lightburn.

Thanks

[Dave Sheldrake]

Is that howl when it stalls the belt slipping?

[Kev Williams]

Two of my lasers- my 'good' Gravograph LS900, and my Chinese Triumph 1390- have both had wire issues within the Y-axis drag-chain.

The Triumph started doing nutty stuff at about 2 years old, and only in certain spots. The problem was one of wires broke where it was soldered under the heat-shrink, it was okay until the chain would loop uphill; when the broken section was in the loop the tension on the wire would pull the connection apart and the X stepper would go bonkers, until the chain flattened out again.
Here's the bad wire:
tribrokewire.jpg

My LS900 has basically the same problem, except rather than a soldered connection breaking, a stepper wire just plain wore out, and started shorting out on the cabinet when the drag chain was in just the right position.
Here's that wire:
900brokewire.jpg

And the 900 would go bonkers when the wire was shorting out, much the same way the Triumph did...

However, your machine isn't going bonkers, it's just locking up...

But, I've had that problem too, the X stepper would get into some nasty vibration, and would lock up. So I thought I needed a new stepper, but turned out there was nothing wrong with it, which I found out when the new one I bought did the same thing. What I found was, the steppers on my machine need some work to do, they need some load, they do not like 'freewheeling'.
Here's a video clip, I chopped this out of about 5 minutes worth, and sorry about the incessant beeping --
but the vid shows what my stepper was doing totally un-attached from the main belt, it's only connected to the step-down gear, slewing at 500mm/Second...



So what did I do to fix the problem? Tightened up the belt... It might not be your problem, but it won't hurt to try

[Matt Schrum]

My first check would be broken wires or loose connections. I would guess that it's likely a mechanical issue and not software related. You can move the same job to a different area of the bed and see if that generates different results (and if it does differ at all, then you're looking at mechanical issues. If you get the exact same results, then maybe it's somehow a software gremlin).

[Kev Williams]

Just found my 'bonkers' video-- I made this vid of my Triumph before I knew the wire was broken, and assumed it was a stepper motor. You'll note that where it DOES go bonkers is limited to the one row of holes I'm cutting out, which became more apparent to me as time went on. I later determined the problem only happened between 16 and 18" down in the X-axis, which is when the broken wire was in the chain loop, getting stretched, and would disconnect randomly..

anyway, this shows how 'bonkers' and 'locking up' are different

[jonathan lustenader]

Thanks for the ideas, but sadly not it.

The belt is definitely not slipping, you can see the stepper stall when the howling happens.

I tried adding a little bit more tension to the y-axis belt on the right side as it was looser than on the left, but no change at all with that.

I reduced the acceleration significantly (factory default for the y-axis was 3500 mm/s/s - I dropped it to 500 over a few steps), which helped oh so slightly, but still happens on occasion. Not only is it still happening, when I let the entire program go (with the poster board actually getting cut) you can see where it happens in other areas of the program - exclusively in the y-axis. It does it randomly along the entire travel.

I had a similar issue on my CNC router that I was able to correct by adjusting the duration of the step pulse in Mach 3, but I can't find anywhere to change pulse timing in RDWorks or anywhere on the controller.

[jonathan lustenader]

So, nothing definitive, but good progress.

In the absence of being able to adjust the pulse duration like you can in Mach3 (my baseline in thinking just because of the similarity to issues with my router), I started looking at how the drivers were setup from the factory. Both axes were set for 32 microsteps, which is 6,400 pulses per revolution. Most of my experience with stepper based systems is using a Gecko G540, which operates at 10 microsteps, with perfectly acceptable accuracy and smoothness. Figuring that was a good starting point, I changed the driver to operate at 8 microsteps (1,600 pulses per revolution) and quadrupled the distance per step in the controller to match. I was then able to run the program without the stalling it had been having, but discovered a new issue at that point.

The program has narrow slots at a 45 degree angle (to tuck the corner of postcards into to hold them in place), and each of these slots had one of the two parallel edges have a stair step pattern to it instead of being a straight line. This was inconsistent to whether it was the upper or lower line of the slot, but always one of them. I suspect this is happening when moving in one direction or the other, but not both. Using RDWorks on the laptop I have hooked to the laser, I drew similar angled slots and told the system to operate at the same speed as the problematic program (150mm/s). The slots burned out perfectly, with no problems. Could this be something wrong with how Lightburn interpreted the initial .dxf, some other random issue in the software or controller, or something related to another buried setting; I have no clue.

So now I am going to take the initial .dxf file and redo the program in both Lightburn (what I used initially) and RDWorks, taking care to make as many of the settings as I can the same (feedrate, power, and whatever else I have the ability to change) in each program. I'll then re-run the file from each and compare. Since making this initial program, Lightburn has had two revision releases, so if there was a bug that caused this, hopefully it has been addressed. If the problem persists with the output from Lightburn and is fine from RDWorks, time to involve them in finding a fix. If they both suck, time to step back and rack my brain again.




Scratch that - RDWorks opens the .dxf file as strange twisty splines all over the place. Maybe neither one behaves well with splines. If the file from Lightburn keeps acting up after redoing it, I will revisit the initial file from AutoCAD to break it into basic arcs and lines (kind of annoying that when you join a line and a curve in AutoCAD it makes them into a spline).

[jonathan lustenader]

So not an artifact of the software, not directly at least. I need to work on acceleration settings I think. It's always the second line of the slot that has the stair stepping because the head is shaking from such an aggressive direction change.