QTPlasmaC mode2 - valid arc lost

After initial setup, I have wired the external THC with my parallel port controller and my plasma source, it's is a blowback type torch. My THC has a internal divider so I plugged negative and positive straight from euro connector.


Problem is I'm getting "valid arc lost". I guess it happens after THC starts sending UP/DOWN to QTPlasmaC,  because if I set THC "Delay Arc" to 1.5 sec, arc lost happens at 1.5 sec.   In the video below  THC setting "Delay Arc" was 0.5s (I know, too low, but it happens no matter the value, it only postpones the error)



I'm also getting "Probe tripped during non-probe move" but I think that is because the THC is making the torch go DOWN.
 

If I disable THC, it cuts ok.

The test material is 0.8mm galvanized steel. Tried several combinations of pierce height and cut height. Using 20Amps and 5 cfm. 
My Arc-OK UP and DOWN cable is shielded and grounded.  

Could you point me the things I should try setting and testing ?  Thanks.
 

This is the TCH view of the failed cut: 


This is the QTPC view:

Sure, should have done that already.

 

Thanks for looking at it.

The last messages:

Tue07 12:39 joint 0 on limit switch error
Tue07 12:39 joint 1 on limit switch error
Tue07 12:39 joint 2 on limit switch error
Tue07 12:39 joint 3 on limit switch error
Wed08 06:57 Unexpected realtime delay on task 0 with period 40000
This Message will only display once per session.
Run the Latency Test and resolve before continuing.

were because I had turned the controller off and put Linux to hibernate without shutting down QTPC

Sorry, I can't because I'm off site now.
But it is a simple conversational 20x20mm square
If it helps, you can see the g-code in the youtube video above.

 

Yes, I do.
It's from macrol.com.br (a local company here).

English manual is at:
macrol.com.br/wp-content/uploads/2020/10...WII-V3_4-english.pdf

Thank you, Phill, for your time and patience.

I will follow your advices and also start fresh with a new config, migrating block by block. I had some merges in the past that may be causing problems.

Will get back and share the findings.