PlasmaC torch starting problem (Solved with help from the famous three)

PCW wrote: Looks like the arc voltage may exceed your arc-ok high threshold
can you change the scaling and post it again?

OK is this better: Fail cut with air


fail cut


Fail cut started after the air stopped purging : no 50hz before start

phillc54 wrote: Wow, that is a mean spike you get 100mS after torch on for the second cut.

So there a maybe a few different ways around this.

1. install a current sensor in the torch lead and use that for an Arc OK signal
2. I could add a Arc OK on delay so the arc voltage needs to be within limits for nnn amount of time before Arc OK is true
3. I could add a Arc OK off delay so the arc voltage needs to be outside limits for nnn amount of time before Arc OK is false
4. I could add a Arc OK High inhibit so Arc OK won't go false if the arc voltage is too high

That would be very helpful Phil.

Could we try 4) first . I take it I would have to modify the code somewhere If so how?

phillc54 wrote: Wow, that is a mean spike you get 100mS after torch on for the second cut.

Thats not 100 ms! Its one second.

The division is the minor tick. 10 minor ticks per vertical line.

So Is plasmac not waiting long enough for the arc to establish?

This machine has significant lag from the torch on signal to actually igniting the arc. Much more than the two machines I have used.

All cuts are atypical against the hundreds of plots I've done and the failed cuts are clearly caused by Plasmac turning off the torch during the establishment phase.

When an arc commences it rapidly spikes up to 130% to 150% (sometimes even 200% or more) above the desired cutting volts. Then it should drop back to the cut voltage. on its own without THC action. This can take some time, maybe 1.5 seconds. This is natural, native behaviour and you need to set your cut volts to agree with that naturally occurring voltage. Even in the cutting example on the previous page I don't see this and it seems that the cut volts is significantly below this natural voltage so it steps down as the THC is enabled (I think this is what happened in the absence of a signal to say when the THC turned on). Thats by the by, the thing I see is that the gradual ramp up without a spike is unusual. A more typical result is the spikes Plasmac is failing.

If Plasmac fails ArcOK becasue a high voltage threshold is exceeded, then its algorithm is totally wrong as it fails to allow for this voltage spike which occurs naturally.

Plus Plasmac needs to allow for the significant lag from torch on to arc commencement on this machine. I don't know how Plasmac works there.

I still think this is another example of why synthesising ArcOK from a voltage is a bad idea. If you don't have an ArcOK signal, then you should add one that is based on current sensing. Eventually you will come unstuck as many have already but this is a more extreme example.

Sounds like if it not already done, simply masking the high voltage test for
a settable period of time after the torch on command would fix the issue.

Also just setting the high threshold to a bit under the open circuit voltage
when waiting for arc OK seems like it would also fix the issue

PCW wrote: Sounds like if it not already done, simply masking the high voltage test for
a settable period of time after the torch on command would fix the issue.

Also just setting the high threshold to a bit under the open circuit voltage
when waiting for arc OK seems like it would also fix the issue

Fix the algorithm, instead of this at line 469 in plasmac.comp
Try this:

recompile and install the comp with halcompile

My typos excluded, this should set the arcOK when it exceeds high voltage
and turn it off when it falls below low voltage.
Try high = 50% of cut voltage, low = 30% cut voltage.

You may need to reset ac_ok_out to FALSE at the end of the cut

And Arc Fail timeout of at least 2 seconds due to laggy torch start.