Some thoughts and some findings using PlasmaC

Home work :
-Do some cutting of whatever
-during cutting trigger a limit switch
-press F2 and resume by using "run from here"
=Result=
-probing will ignore probe speed
-torch will move to high hitting the limit switch at the end of the cut, not always
-THC will not follow the material properly usually lowering the torch to much---subjective/not confirmed !!

Please let me know if anyone else can reproduce this.
Thank you.
Last 2 days did some cutting on 3mm mild steel plate with 386 probe/pierce per 2x1m plate, 6 plates total so the issue of Z axis moving past limit was still there at 3m/m setup speed, but it was bearable, roughly once per 100 cuts.

Tommy,

I will try to reproduce in the next day or two. One question I had... are you on the latest version? I had a similar issue and I thought I was on the latest version but was not ... I hadn't seen it again after updating (doesn't mean it's not there).. but just wanted to make sure we are all on the same sheet of music.

I have been trying on a sim but don't see the issue. I tripped joint0 and joint3 limits and both seemed to behave ok.

Definitely can reproduce it on my machine (haven't tried on sim).

It's exactly like Tommy described, but I would add some detail as to how I got it to hose up:

Start program
After first probe, trip a limit switch (I've been doing Joint_2 (the Y next to me as I can put metal in front of it easily)
F2
Click G0 positional move before M3 command and do Run From Here
Let the torch probe again, let it go until after it probes the next M3 command
When it starts cutting, stop it
Restart the program
Let it probe and start to cut
Press stop

At this point I got a Joint_3 following error, my table sounds like Z ran into a hard stop (just a terribly sudden Z move command I am guessing with no acceleration). Everything past that caused Z to behave badly. I ran the torch to near the top, but the next time I pressed cycle start, it slow probed from the top to the metal and at the next retract ran UP into the hard stop and crashed (and gives me a joint on limit switch error). Since I had to manually move the Z off the limit, I had to restart linuxcnc (or rehome) to keep going.

Another note, I got it to do this 2 times. Once as above, and another that just gave me a joint following error and did not crash after that.

Things to note:

1. I think this is highlighting that there is still something going on with the Z offsets and movement timing. At times, Z movements from when under the control of PlasmaC are still causing Joint_3 (Z) following errors, it sounds as though the acceleration is not accounted for. Like it's commanding too big a move too fast.
2. From what I see of what Tommy described, he's using bad practice. The user shouldn't start a program with the torch so far down (almost touching the material) near the soft limit (I'm guessing). It would be better practice to jog the torch near the top before doing "run from here" to recover from an error of this magnitude (Phill, please correct me if I am wrong). I am guessing this happens because he is starting the program near the lower soft limit. I suspect I've not seen this error because I added a Z movement to my XY zero button and I try to start 5mm from the top (per the user guide ).

Well Snowwy, you are much better at this than i am, thank you for confirming that.
I do agree that this should be an offset thingy, and i do no mind homing the Z axis again if and when that happens as tripping limit switches should not happen during normal work. The other thing when the Z goes past the soft limit happens due to arc lost at the end of the cuts (lots of them due to cutting thin material with big size nozzle) so it pauses and usually resumes fine, but sometimes it will not resume as it somehow got over the limit, stopping the program execution by pressing ESC and resuming from line works fine.
As for your guess, i did that only once so far and stopped and resumed after raising the torch, i have both machines set to have the Z axis home 20mm under the limit switch, they both have 200mm of travel on Z and one is cutting at the very bottom of the travel while the other is cutting roughly at 120mm under the limit switch, so there is plenty of space.
To rehome Z axis in case the Z axis trips the limit, switch to joint mode, click on "ignore limits", move the Z axis down with the on screen buttons till past limit, press home on the keyboard, this will home just the Z axis and switch to world mode on it's own.

Did some tests today regarding void sensing and the results are pretty messy,
===setup===
Cebora 120A plasma source set at 50A (does 60 measured), 1.4mm nozzle
Feed rate at 3600 and 1800mm/m (can cut at 4.4m/m with the same current/nozzle)
- cut 10 columns of 200mm length spaced at 30mm
- cut 15 rows of 350mm long cuts
- changing the void sense override on the config panel from 1 to 500 does change the results, but after going on and off several times at the beginning of the cuts ( even when not over a void sometimes ) it just stays on after passing the first to third column/void, remains on a bit after the torch is off.
I did save the hascope data and a screenshot, but they are on the machine PC outside so i will upload them tomorrow.

On another note, PlasmaC does not have ARCOK off grace period, setting the debounce to 500 did not improve much the issue of getting arc lost at the end of small cuts requiring user intervention to continue.
"mine" had a settable period in the ini file at the very bottom, had it set to 0.2 so it never gave such issues.
Again this is cutting thin material with big nozzles, so not for everyone.
I will try to add a timedelay tomorrow, see if that helps.
Is there a debounce at the end of the signal ?

tommylight wrote:

On another note, PlasmaC does not have ARCOK off grace period, setting the debounce to 500 did not improve much the issue of getting arc lost at the end of small cuts requiring user intervention to continue.
"mine" had a settable period in the ini file at the very bottom, had it set to 0.2 so it never gave such issues.
Again this is cutting thin material with big nozzles, so not for everyone.
I will try to add a timedelay tomorrow, see if that helps.
Is there a debounce at the end of the signal ?

When cutting holes in thin material, I have used G-Code to ignore the arc-ok signal .1" (or so) from the end of the cut with VERY good results. That lets it arc out whenever it wants to, and does not stop the machine. I simply turn it back on before the next cut starts. Rather than use large debounce numbers, I would recommend coding the enabling and disabling of "ignore arc ok" to just ignore it near the end of your cut. No intervention will be required

Have a look here

Of course, use synchronized with motion as immediate will "break blending" (it will slow down and stop your cut for a very small increment).

Let me know how it works!

While i am on a roll, the reverse run is magnificent for the issues i am having as i can always go back and forth and fix any mishaps, BUT, the biggest feature is being able to continue a cut even when the arc was lost on a bend or radius !
With toma_thc config i had to use run from here from a straight line in gcode or from the beginning of the cut when there was no straight lines, so that meant having the torch start and go over an already cut piece forcing me to speed up the cut in that section or raise the cut voltage to prevent the torch from diving into material and dragging rendering the nozzle unusable if i made a mistake during the above procedure.
Now i can do a run from here at the beginning of the failed part and press pause during probing, use the fwd button to get to the uncut part and press resume !
Or to be on the safe side, disable the torch, run from here, pause, check if the torch is where it should be and use fwd/rev to get it there, enable the torch and resume !
Very convenient !