New day new problem

the 100k and 2k are original as manufacturered by hypertherm (my machine is a pmx 45 non XP so it is supplied with a fixed, non adjustable 50:1 voltage divider ratio.... not like yours being ajustable via dip switches)

I would leave yours set at 20:1 ratio and I would leave it using the thcad10 (these won't be your problems with a loss of transfer signal or the cut stopping short)

10v X 21 = 210v full scale.... cutting voltage of about 130v / 21 = ~ 6.1V

up to a max of about 140 or 145V  = 6.6 to 6.9V
I would rather mine was 20:1.... but it's not... my options are
​​​​​​50:1 (standard)
10:1 (neuron thc voltage divider board)
raw voltage

leave it as it is ... but I would suggest some logging of signals to try to work out why or where the signals are being lost or stopped. 

(off to do some documentation reading on logging as it's one thing I have not read up on)

Thanks for clearing that up and I will be making the change to the hypertherm to the 20:1 and the Arc voltage scale and see what happens. I wonder if the "Backup Config" ( .taz file) log would give the info you are looking for to get info on the error message

If a cut is interrupted by an error then it is most probable that motion doesn't stop at the exact point the error occurred, nor will motion automatically return to where the error occurred.
There is a cut recovery panel that replaces the jog panel and you can reverse along the path a bit and even off the path as well to create a leadin.
linuxcnc.org/docs/devel/html/plasma/qtpl...html#qt_cut-recovery

There is also a lost arc delay halpin that may be useful in your case:
linuxcnc.org/docs/devel/html/plasma/qtpl...html#_lost_arc_delay

I agree with Robert 20:1 is the way to go.
The only other thought I had is that perhaps there is not enough current through the Hypertherm relay to ensure its reliable operation.
This topic has been done to death on the forum a few times. I think its now covered in the Plasmac docs with a couple of solutions. A pull down resistor from the Mesa input to ground would cover it. I bought some 100 ohm 10 watt HS10 chassis mount resistors for this which will allow 240 mA through the relay which is rated from 100 mA to 10 amps.

rodw wrote:

This topic has been done to death on the forum a few times. I think its now covered in the Plasmac docs with a couple of solutions. A pull down resistor from the Mesa input to ground would cover it. I bought some 100 ohm 10 watt HS10 chassis mount resistors for this which will allow 240 mA through the relay which is rated from 100 mA to 10 amps.

It sure has, why don't you offer the alternative options? There is no need to use such a low value resistor as you keep proposing.

1. Because the alternative options are documented as I mentioned.
2. Because I can't remember what the alternatives are but can  remember what I have done.
3. Because a value close to the resistance suggested in the Plasmac docs worked for a while in my Everlast but proved not to be enough later in its life in the hands of the next owner.
4. Because its very had to find collaborating data for the use of capacitors to solve relay oxidisation despite my highest regard for PCW.
4. Because despite 20 posts on this thread, nobody else had thought to flag the most likely cause of the problem.

rodw wrote:

Because despite 20 posts on this thread, nobody else had thought to flag the most likely cause of the problem.

I would expect that if it were a wetting issue then it would crop up at the start of the cut not in the middle of the cut.

There are a few patents on capacitor discharge contact wetting
one is : US9355791B2

phillc54 wrote:

rodw wrote:

Because despite 20 posts on this thread, nobody else had thought to flag the most likely cause of the problem.
I would expect that if it were a wetting issue then it would crop up at the start of the cut not in the middle of the cut.


+1