My Newb Plasma Table Attempted Build

phillc54 wrote:

Hard to tell from the pics but are any of the corners of the "square" actually square?

Is the notch on the X axis or the Y axis?

Can you post the code you cut with, it looks ok when I do it here but...

EDIT: To me it looks like your X and Y axes are not perpendicular.

Phil, I made the adjustments on the float setup and that eliminated the choppy cuts at the corners. The step where the lead in/out is is still there. Attaching some photos. On the side where the step is, you can see that each side of that step on the X axis is square/perpendicular to the Y axis.

PCW wrote:

Can you post your current hal and ini files?

Yes sir, attached. It's odd becaue this issue just seemingly popped up out of nowhere. Before I changed the pulleys and belts. Cut one part as many others, and it was fine, then the next part and all afterwards have had the same issue since.

tommylight wrote:

Was that slope programmed or the machine did that on it's own?
Are you by any chance getting the latency error when starting LinuxCNC ?
When the drives are powered and enabled, how easy it is to turn the motors shaft by hand ?
What drives and what microstepping ?
If that slope was not programmed, one axis is binding, so if you can, remove the belts and move the axis by hand slowly while feeling for any bumps or binding.

Yesterday when I had it all apart to change the pulleys and belts I cleaned all of those surfaces up. Still, I thought maybe there is something making it miss so I tried several different areas of the cutting space available and it repeats the identical issue no matter where the material is placed to be cut. That makes it seem to me that it isn't some debris in the track causing a skip.

Just cut a 1in square using the wizard and it seems to happen at the lead in/out point.

I have no glue of plasma but i would guess that your driver catches some extra pulses as emf when you start. I would have a look into grounding.
What hapens when you fix a pen and run the same code without an arc
Just my 5c
Mike

tommylight wrote:

Check the DRO while cutting, is the axis causing that divot on the same position at the beginning and at the end of that cut ?

Just did this and the DRO shows at the beginning (after lead in on X axis) -.205 on the Y, then turns onto the Y axis cut down to 3.205, then back on X for 3.205, then up the Y for 3.205, then the Y axis position shows -.205 when coming back to where the divot/step is. So basically the DRO is not reflecting what the machine is actually doing there.

Mike_Eitel wrote:

I have no glue of plasma but i would guess that your driver catches some extra pulses as emf when you start. I would have a look into grounding.
What hapens when you fix a pen and run the same code without an arc
Just my 5c
Mike

Just tried this and it does the same thing with the plasma powered down completely. The proportions are a bit off because of the probing, but it's doing the same thing without any opportunity for interference from the plasma for sure.

i would write my own square in gcode. Its likely a bug in the shape gcodegenerated

Before I did that run with the plasma off and used a marker, I was going back over any changes to anything that were made around the time this started happening. I changed consumables to finecut, and although I don't remember for sure I believe before that 1st bad run I had updated linuxcnc/plasmac. Maybe I should go back to v2.8 (I'm on master now) and see what happens. I don't think it would be whatever was available from updating, but maybe my shoddy internet connection resulted in some errors in the download...? I think at this point it's worth a try.