Why real time kernel is needed with board like MESA?

I think you will find we have incorporated everything you talk about in the Plasmac config. Its just we may have gone about it slightly differently.
Probing down and then away has always been supported in Linuxcnc gcode with G38 but In Plasmac its done internally.
Yes, we sample the torch voltage constantly 1000 times per second and auto sample the torch voltage after a timedelay from ArcOK but defer it until we are at or near the desired cut velocity (your CV).. I never set it on the screen and always auto sample.

Linuxcnc has never had a pin that tells you the programmed feedrate (eg the target CV). Plasmac works around this by setting the feedrate from the materials settings on the screen through its materials file.

The current velocity (1000 times a second) is published on the motion.current-vel pin (mm/sec)
Version 2.9 (the development version) adds a motion.feed-upm pin which tells you what the feed rate is at all times(mm.min) This is the target Constant Velocity (CV you mention)
So looking for contant velocity is as simple as seeing if these two pins are equal

Plasmac has a void/kerf crossing hold. I'm told it works but few people need it. I have a morse sophisticated algorithm but its still a work in progress. It uses moving averages and looks at dv/dt (change in volts over time)

If you go to the last few pages of the plasma section on this forum, you will find a few components I wrote for torch auto sampling and other plasma functions you may be interested in.

I used Phills void sensing feature for the first time the other day, and it worked perfectly. I was cutting some left over 25mm plate and I had some large lead ins in the section I was cutting. It cut through the large lead ins without the torch diving. Before I would always have to restart the machine as the float switch was activated.

I thinks its foolish to say that the people working on LinuxCNC are not continuing to improve. They are not constantly redeveloping how the basic program functions, rather continuing to improve the features it has. And I for one think that is the most productive thing to do.

rodw wrote:

I think you will find we have incorporated everything you talk about in the Plasmac config. Its just we may have gone about it slightly differently.
Probing down and then away has always been supported in Linuxcnc gcode with G38 but In Plasmac its done internally.
Yes, we sample the torch voltage constantly 1000 times per second and auto sample the torch voltage after a timedelay from ArcOK but defer it until we are at or near the desired cut velocity (your CV).. I never set it on the screen and always auto sample.

Linuxcnc has never had a pin that tells you the programmed feedrate (eg the target CV). Plasmac works around this by setting the feedrate from the materials settings on the screen through its materials file.

The current velocity (1000 times a second) is published on the motion.current-vel pin (mm/sec)
Version 2.9 (the development version) adds a motion.feed-upm pin which tells you what the feed rate is at all times(mm.min) This is the target Constant Velocity (CV you mention)
So looking for contant velocity is as simple as seeing if these two pins are equal

Plasmac has a void/kerf crossing hold. I'm told it works but few people need it. I have a morse sophisticated algorithm but its still a work in progress. It uses moving averages and looks at dv/dt (change in volts over time)

If you go to the last few pages of the plasma section on this forum, you will find a few components I wrote for torch auto sampling and other plasma functions you may be interested in.

I have already read what you say and I must congratulate you for what you have done !!! This with CV and how exactly this pin is released, I did not understand at all, but now is not the time to ask you about it.
I don't want to look like a maximalist, but that's not the end, you know. I guess you have ideas that you're working on, but here are the things we're working on, why don't you like them too:
1. If the voltage recording is as close as possible to the drilling - immediately after the occurrence of CV, this is a very accurate method because it is not affected by the type of material, material thickness and (very important) wear of consumables - nozzle and electrode!
2. Something extremely challenging OpenCV contour analyzer of CAD file and G code generator for plasma and laser. Then it can already give a solution like that of Kjelberg / Hypertherm called - fine focus / true hole, namely - setting the speed in the holes 60-65% of that of the contour, when cutting the contour line of the hole M5 to come 1/2 kerf before the end of the last arc. So there will be no burning of the holes at the point where lead in and lead out meet.
3. And higher mathematics - 3D cutting and voltage transformations when changing the slope of the torch.
Success !

thefabricator03 wrote:

I used Phills void sensing feature for the first time the other day, and it worked perfectly. I was cutting some left over 25mm plate and I had some large lead ins in the section I was cutting. It cut through the large lead ins without the torch diving. Before I would always have to restart the machine as the float switch was activated.

I thinks its foolish to say that the people working on LinuxCNC are not continuing to improve. They are not constantly redeveloping how the basic program functions, rather continuing to improve the features it has. And I for one think that is the most productive thing to do.

You are wrong. You just don't understand people. I never accept that everyone understands plasma cutting as much as I do! And I don't treat those who understand less than me with contempt.
Especially, to ignore them - my product is only for specialists in plasma cutting. If you don't understand it, then you can't do anything. (This is about comparison - if you can't install LinuxCNC, you probably can't install your printer).
This took the most time in the development of our THC - not having to be an electronic engineer / plasma cutting specialist to install and use it!
I hope you understand me (with Google - so much).

We have continuously tried to understand what you want to say.

But you always insulted people.

Be humble.
If you know plasma so well, why dont you show us some sample videos or something useful which we can use in PlasmaC.

If you have problem translating, post in your native language.

Offense ??? I did not intend to offend anyone!
I just said I've been doing this for about 10 years and I know some things better than a person starting now!

What exactly do you find offensive?

As I said, I am ready to share any of my experiences, as I have done in previous posts.

stivemaster13 wrote:

2. Something extremely challenging OpenCV contour analyzer of CAD file and G code generator for plasma and laser. Then it can already give a solution like that of Kjelberg / Hypertherm called - fine focus / true hole, namely - setting the speed in the holes 60-65% of that of the contour, when cutting the contour line of the hole M5 to come 1/2 kerf before the end of the last arc. So there will be no burning of the holes at the point where lead in and lead out meet.

I hate to burst your bubble on this, but this is as simple as setting a rule within the PP. If you run Linuxcnc with PlasmaC, the PlasmaC component is already set up to do this internally. This is old news too.

If you would take the time to thoroughly read the PlasmaC User Guide and PlasmaC Primer documents, you would find this information in detail.

Most of what you say has been covered in Linuxcnc. including the hole mode velocity reduction. I like the idea of constant velocity so I will do some plots to review this behaviour as its something that I have been thinking about and meaning to review for some time.

I think the true hole technology Hypertherm uses is a bit more sophisticated than simply slowing the torch to 60-65% on holes. Thats just the first step. This velocity reduction should not be arbitrary but determined by the hole radius and the velocity at whch it can be traversed. The next evolution for Linuxcnc I am working on is to dynamically measure hole and arc radius in real time while cutting. We can calculate the applicable centripetal acceleration limit and the minimum radius the machine can traverse without slowing down. We should then be able to compensate for the wider kerf if this limit is exceeded by offsetting the torch position in real time so accurate dimensions are maintained. You can't do this without the real time environment Linuxcnc provides and it will remove hole mode gcode from the cut file.

It does not matter if one of us knows more about plasma than you or not. You have to understand that collectively we know far more than you as our knowledge is pooled together in a collaborative way in this open source project. Someone did say that probably less than 100 people on the planet really understood plasma and most of them kept their knowledge confidential. He put me in that group so welcome to the 100 club!

I think there are some on this forum who know more than I do. But their knowledge might be in a different field to my knowledge so once pooled we are truly experts.

That's right you're right. Here are some details about the openings and methods of HYPERTHERM. There things were like that in the recent past - if you mathematically calculate how long the burner travels while cutting a hole and then if you cut a straight line from a contour you get a number. That is, in order for the cutting speed to be equal, it must be reduced in the holes to about 70%. This was what Hypertherm's controls were doing then. However, then they found (of course not only them - most innovations in this business actually have Kjellberg) that the cyclone of the plasma jet is not actually stationary but wanders when changing direction and therefore it is better in the holes cutting height should be more small, the velocity is about 65% of that of the contour and the plasma must stop shortly before reaching the kerf crossing. At Kjellberg at this point at the end of the hole (1/2 of the thickness of the kerf) the plasma decreases, does not stop. I can not say what curve reduces the amperes (I did not think to measure).
Kjellberg do the same in sharp turns - reduce amps in real time (up to 20%).
However, this is aerobatics and very much depends on the equipment and whether it is compatible. In no case does it apply to any equipment.
The plasma cutting machine can cut very well without controlling the power of the plasma in real time.
This for the second cutting height of the holes can also be done with the tool number !!!
I want to make one clarification - all the time as holes - I mean small holes or precise holes (bolt holes) such under 30mm.

This is interesting. Yes its unfortunate that there are not many plasma cutters that allow reaI time control of current. There was one mentioned recently with 0-10volt control. I think the Thermal Dynamics current control would be easy to intercept and replace with a digital potentiometer. I do know one user who has done this.

I do not think the amps will alter on a curve as plasma is a constant current circuit. With current control, it would be interesting when cutting a hole to disable the torch height control and switch to changing the current to keep voltage constant. In theory, the arc's power would remain constant so kerf should remain the same width at the lower speed. That sounds like what Kjellberg is doing.

With Linuxcnc, it would not be necessary to change tools, we could use an external offset to lower the cutting height in real time without gcode. Then the height adjustment could be dynamic and vary with conditions. How much do you lower the torch by?

I really would like to see a system where there was no gcode changes required at all but instead let the motion controller manage everything. This pull request to the linuxcnc core I wrote was the first step to allow this. github.com/LinuxCNC/linuxcnc/pull/900
This publishes the arc radius, and variables that say if its a hole or an arc and a direction heading which could be used to calculate X & Y offsets to adjust the torch position to account for the wider kerf at the lower cut velocity. It these kind of complex calculations that apply position and velocity overrides where Linuxcnc's real time control comes into its own...