Random arc signal lost

Clive wrote:

tommylight wrote:

Edited and added more info to the above post, sorry had to attend installing the roofing on the "hut" or "tent" or whatever that thing outside my shop is called.

Fridge in the winter and Sauna in the summer

LOL, that is so true.

@Rod,
To late, you should have backed me up when i was makig noise about this, but i never blame Phill, he just wants to please everyone and he likes challenges.

rodw wrote:

I'm sorry guys, I personally think you are off on a tangent debouncing your ArcOK. Show me the halscope plots showing signal bounce on ArcOK! I've never seen any evidence of that!

I would wonder if I actually have to use an oscilloscope, and whether or not halscope can catch it. Talking to an electrical engineer friend of mine, it can happen quite fast.

Please do not clutter Plasmac with unnecessary performance reducing crap! Leave that for the user to add if he needs it! I for one want the ArcOK to be taken in its raw state without any delays which software debounce inevitably adds. I don't want the pierce delay to be extended which will happen if a software debounce is added.. I want the plasma cutter to obey the cut parameters so that a 0.2 second delay is actually a 0.2 second delay not 0.22 seconds.

Currently, what fixed Cody's table is 5 periods on the 1000000ns servo thread, or 5ms. 5 milliseconds is .005s. So, as it is currently, it would bump your pierce delay from 0.2 seconds to 0.205 seconds. I am hard pressed to believe you can tell the difference a 0.005s debounce has on your pierce. However, without the debounce, the table, as I said, is unusable. But aside from all of that, if you dont want it, just don't add it to your config, or comment it out if it ends up there

Don't you think Hypertherm, Thermal Dynamics ESAB et. al. have electronics engineers that might be smart enough to ensure there is no signal bounce on ArcOK which is a critical system parameter?

I don't think it really has anything to do with their intelligence. It probably either has to do with industry standard, or cost. They're dry contact relays. They have mechanical bounce, and as such electrical noise. By your same logic, you could say "If minimum current fixes this problem, don't you think the engineers would be smart enough to mention it in their manual?"

Please refer to the Plasma Primer which contains this halscope plot which dates back to my Everlast plasma cutter..

I'm curious A if your halplot is shown with your resistor in place, and B if halscope is fast enough to catch it and C what it looks like when you zoom in much further (with halscope, not on a grainy picture).

Zoom in and locate the pink ArcOK signal (hint, its near the arc voltage). Can you see any evidence of signal bounce? I can't.

I would expect to have to zoom in MUCH further to see it. The time period of this plot is far too coarse to see the electrical noise from contact bounce.

I'm sorry Snowgoer. The fault I had with my Everlast ArcOK was correctly diagnosed by a panel of two silver haired experts in analog circuits. Your feedback just confirms that today, nobody understands the issue of relay contact oxidisation. Its not something that occurs in digital circuits. I had no clue either!

No need to be sorry bud. As I stated, the table wouldn't cut metal before, and it cuts metal now I would propose that by the same logic you apply to oxidation, perhaps the "silver hair" folks you mention aren't aware of digital computers fast enough to be effected by the switch bounce present in analog circuits. As stated, I suspect the resistor reduces the hysteresis present. Bounce when mechanical switches, buttons, relays, etc. is an electrical fact. In fact, that's why a debounce module exists in the first place.

It might not be the issue causing your problem, but you should be aware of it and the fact there is a 25 cent fix for it.

Table seems to work fine so far now (knock on wood?). I'd be remiss if I didn't point out that adding debounce to the arc-ok signal is ... free, and there's no chance of wiping out a ~$200 Mesa card due to a misplaced resistor.