Another plasma component...

I have just pushed an update:

GCode Parser - Bug fix to correctly restore defaults on exit.

Test Panel - reduce arc voltage to minimum when torch turns off.

Keep it up Phill.

Actually Tommy could not help himself and started to spread the plasmac word
www.cnczone.com/forums/general-cnc-plasm...cnc.html#post2311734

Of course I helped him out!

What are mates for?

If anybody has some video or wants to give an endorsement of Phill's handiwork, feel free!

Another update pushed...

A fix for a bug in the torch off delay.

A change to hole sensing calculations and easier on the eye output from the gcode parser.

Phill,

For your info. The G41.1 and G42.1 are useless for plasma. I tested some of this function's on a circle and it's a nightmare to program.
The lead in's are alway's complaining. The contour's are not finished by the G41.1 and G42.1. There alway's stay's a spot left. WIthout a overcut Gcode command it's not able to make a full circle. And a Gcode overcut command does not exist. Maybe something for Andy to write. User's without a G64P0.01 command, maybe would not spot this accuracy problem.

Maybe it's better if you do a documentation update some day. To delete the G41.1 and G42.1 story's.
Later on we have a nice cam program that solves this. And sheetcam solves it, or Dxf2Gcode, or Inkscape.

Inkscape Gcode is for me also abra cadabra... Dxf2gcode is difficult for me too... The only one that i like is Sheetcam at the moment.
Hypertherm nest software is also no good option. It can nest, but that's the only nice feature it has.

Grotius wrote: And a Gcode overcut command does not exist. Maybe something for Andy to write.

Andy's suggestion was to create a new task module for plasma. In an Ini file, a task module and a io module are loaded. LinuxCNC only has one task module called milltask. But it has two io modules io and io2.

So Grotius, when you want a break from gcode generation, try this with the LinuxCNC source code and compile it..
1. Create a copy of the linuxcnc-dev/src/emc/task/ folder under the linuxcnc-dev/src/emc folder
2. rename the folder to plasmatask
3. Edit the Submake file in our new folder so it makes plasmatask, not milltask
4. in the linuxcnc-dev/src folder edit Makefile to build our new subfolder under the SUBDIRS directive (task is on line 127)
5. Build Linuxcnc in the usual way.
6. In your ini file, change TASK = milltask to TASK = plasmatask

Linuxcnc will now run perfectly but it will be running our new plasmatask module which we can change to our hearts desire.
You also need to make an IO module, so we can just go to the inuxcnc-dev/src/emc/iotask and copy one of the io modules there to iocontrol_plasma.cc and edit the Submake file. Then edit your inifile again so that EMCIO = io loads our new io module

This will be our new plasma specific halui module. So now we can write plasma specific code in plasmatask and new pins in halui and have plasmatask message plasma-io to update those pins and plasmatask might be able to be taught how to overburn.

Have fun! I would like to see the state tags branch implemented as a prerequisite to this. We'd need to comeup with a feature list for plasmatask.

For your info. The G41.1 and G42.1 are useless for plasma. I tested some of this function's on a circle and it's a nightmare to program.
The lead in's are alway's complaining. The contour's are not finished by the G41.1 and G42.1. There alway's stay's a spot left. WIthout a overcut Gcode command it's not able to make a full circle. And a Gcode overcut command does not exist. Maybe something for Andy to write. User's without a G64P0.01 command, maybe would not spot this accuracy problem.

If the post processor cannot do overcut then it can be done by the PlasmaC GCode parser

Do G41.1 & G41.2 behave differently to G41 & G42?