THC antidive output turned off on 50 msec early before M5

[shad]

Hello Balazs!
I am try to use THC antidive output for the automatic voltage control. All works perfect during code execution, but I am notice that torch dive on the end of the cut.
I am make test and find that THC antidive output turned off on 50 msec early before then "M5" command issue and this causes torch diving to the plate. It's not good.
I am understand that this time depend from machine acceleration value and from cutting speed. But any time torch dive.
Is it possible to turn off antidive signal at the same time with the SPINDLE off?
Thank you!

[cncdrive]

Hi Andrew,

Could you please post this part of your g-code where this issue happens?
Then we will measure it out.

[shad]

Balazs, I am attaching program. On the every cuts it's happens.
One thing - the acceleration was set to 50 mm/sec^2 - antidive turn off in 50 msec before M5
500 mm/sec^2 - antidive turn off in 65 msec before M5

[shad]

I am using for testing UC300ETH-5 LPT controller.

[cncdrive]

Hi Andrew,

Thank you for the informations.
It is strange that this is happening like that, because the M5 is non-syncronous command, so the PC side software waits until the motion stops and only then sending the M5 command to the motion controller.
So, it is strange that the anti-dive output switches off before the M5.
I yet just speaking theoritically I see 2 possible reasons, one is that there is an issue with the ISmoving bit, and I remember a recent fix about that a few versions back which issue was that the ISmoving bit went off too early, so the question arrises, that which UCCNC version are you using now?
The other possibility I see is that the issue is not M5 macro related, but there is an issue with the Antidive check routine where it gives a false value too early, a bit before the motion actually stops.

Anyways, we will debug this soon, but meanwhile please confirm that you testing with the lastest, the 1.2039 version?

[cncdrive]

Hi Andrew,

We've checked why this issue is happening and the reason is totally different than what I thought.
I assumed that the motion did not end until the M5 on your scope photo, but it does.
So, the problem is that the motion ends and so the controller goes to idle and so the anti-dive goes off and only after the M5 executes.
Because the M5 is non-syncronous, but is a text macro there is a small time between the motion stops and the M5 command is sent to the controller and executes and in this short time the anti-dive is off, because there is no motion, the controller is already in idle.
We have an idea to add one more status bit to let the controller know when the cycle is started and stopped and make it as an enable signal for the anti-dive check routine.
This way the anti-dive could be kept on in that small time.
It will require some time for us to implmenet and test this, problably a few days.

On the attached image the anti-dive is the top channel and below is a channel direction signal and the bottom is the spindle (M3/M5),
so it can be clearly seen that the issue is when the motion already ended and between the M5.

[shad]

Hello Balazs!
Thank you for this explain, I am understand what happens. I am hope you can solve this issue and we can use this nice and powerful antidive future!

Also now I am understand why happens another issue - my customers said that on the every end of the cuts a divot is formed on the cutting part - if they not use lead out.
I think there is a delay between motion stop and M5 commands execution. In this case motion stopped, but M5 (and torch off) delayed on the 50-60 msec depend from cutting speed and acceleration settings.
Plasma arc is a continuous process and if there is no movement, it burns a hole in this place. 50-60 msec enough for this.
I am very sure we need to have any way to check end of the cut very fast as possible and turn off cutter. I know that this issue very important for plasma cutting quality.
May be you can add for example SpindleStartStop_event for this?

[shad]

Just now plugin check Spindle on/off LED for start/stop cutting. And I think the delay little more than delay between motion stop and spindle off because we have screenset delay.
It's will be great to know exactly when cycle is stopped - I mean about stop of the motion. But I think it's not easy....

[shad]

Balazs, Is it possible to activate THC antidive signal in any distance (for example 3 mm or distance can changed by operator) before end of the cycle, include M5? Just independent from any. Only distance before end.
It can be very useful, I think plasma users understand me
Thank you!

[Robertspark]

Andrew,

The problem with M5 is it's non-synchronous, and also unless you add in overburn to the end of your cut the torch will always decelerate to zero before turning off the torch via M5.

Hence why I use M10/M11 + M3/M5 as was discussed in another post and use the over-burn feature within sheetcam to extend the burn.
The catch with using this is you need two outputs and to be able to setup an OR + AND gate (i.e. fire the torch with either an M10 OR M3, and turn off the torch if EITHER M11 or M5 is triggered and the output from the UC controller drops out.

Likewise I use a snippet code within sheetcam to lock the torch (anti-dive) before the end of the cut a set distance from within sheetcam. The issue with uccnc is it will have to know where the end of the cut actually is (last motion before M5) and then implement anti-dive a distance before this.

I think this is going to be difficult.