Plasma Torch Height Controller

[robertspark]

I've been dabbling with the idea of a THC for years....

So I finally thought I'd have a go at something after being a plasma user I thought I'd have a good idea of what I would want to see in a THC (and I have fee time on my hands)

I thought I'd pose a question.... what would you like to see or entice you to buy a torch height controller?

The THC I've come up with basically takes a raw torch voltage signal and will output a simple THC UP / THC DN signal based upon whether it is below or above the setpoint.

This THC does not have any knobs, buttons or a screen as it is expected that it could be controlled from the PC front end or via a touchscreen.

Originally I was going to transmit the target voltage to the THC via a PWM spindle signal, which I may still do / offer as a wiring option, but then it got a little complicated as I wanted to be able to control the hysteresis (dead-band) of the THC and also offer a means to control the sensitivity of the THC, given it is essentially a competitor and will have some averaging / software filtering within the code within it. I decided that I'd provide a modbus interface to provide the SETTINGS to the THC. the THC will still output THC UP and THC DN. and it could have an input for THC INHIBIT and also for ARCOK.

Cost as I see it is a factor so I did want to do this as a low cost solution (hence avoiding knobs, buttons and screens).

All the settings for the THC could be output from gcode to set the voltage, hysteresis and sensitivity so via a postprocessor from within sheetcam.

Modbus is very noise immune, UCCNC and Mach3 have plugins for it so it can be implemented cheaply and easily.
the high speed digital THCUP/THCDN and ARCOK (if you did not want to use the one on a plasma cutter CPC connector) would also be high noise immunity by design.

I don't have a cheap plasma cutter (HF start), but I'd like to test it and offer it for HF start plasma too. I have designed the raw voltage sense circuit as such.

I did not want to do it as a stand alone THC (yet) and thought I may start out with a THCUP/THCDN and ARCOK variant first, and then expand it to Step & Direct and maybe stand alone.

This may go nowhere further than the prototypes I'm working on, but I thought that it may provide a little pocket money as time is cheap for me at present.

[weldguy]

Sounds like a worthy project and I am sure you have the ability to pull it off. One thing I have noticed is that you are very tech savy and most plasma table users in my experience are not. Most would not have the ability to implement what you describe as they do not have your skills. A stand alone with simple installation would be ideal for most. Super curious to follow your progess and see what you come up with. Super cool project

[robertspark]

thanks much appreciated

[Dfabltd]

Hi Rob
Great that you are taking on this project, and their is a market i believe
You have hit the nail on the head with everything you're doing
I don't believe the need for a stand alone unit once the modbus is easy to setup
In reality all must users want is toolset in sheetcam select say 10mm load file in uccnc/mach4 and press cut

I struggled with my mini thc only for yourself i wouldn't have it working now,
if you can create something a bit more streamlined and with a set of instructions that average joe soap plasma table builder guy can read your onto a winner

[Rodw]

Robert, Good luck with your venture. If you can't build a THC, nobody can.

Your modbus interface sounds interesting. What baud rate will it run at?

I have one suggestion for you though. Support arc voltage sampling. Allow the user to send the time delay before sampling the voltage. If a 0 is set, disable sampling.

When I was playing with one of the Linuxcnc developers who had written a THC component (that did not make the grade), I wrote a component that did this and I just hard coded the delay I wanted. I really did this becasue I wanted to make sure the THC control started with a set point exactly on the torch voltage to avoid any unexpected corrections ( I had some massive crashes during testing!). I used this hard coded approach for ages and even now where I have so much more ability to control things, I still use torch voltage sampling and never look at the torch voltage on the display. I think the delay will vary between tables based on the XY acceleration settings but I settled on 1.5 seconds for everything as it did not seem to vary with material thickness.

The only thing you probably won't be able to do is to defer sampling until it gets to the Feedrate. This ensures a valid sample is made if the timer expires on an arc or short segment and the machine has not been able to get to the feed rate. We can do that in Linuxcnc becasue we know the current velocity.

[robertspark]

if you can create something a bit more streamlined and with a set of instructions that average joe soap plasma table builder guy can read your onto a winner

That's the plan >> design following KISS (Keep it simple stupid)

[robertspark]

Robert, Good luck with your venture. If you can't build a THC, nobody can.

Your modbus interface sounds interesting. What baud rate will it run at?

115200 at present, but the modbus loop runs at 20Hz inside uccnc so that is more important so I may slow it down depending upon the amount of data once I've had a look at the finlised frames being sent and recieved

At present I am using Serial RTU over RS485 and not ethernet (TCP).

I have thought about using Serial ASCII so that the modbus could link to hypertherm RS485 communication port, but the baud of that is really slow at 19200.
https://www.forum.linuxcnc.org/media/ku ... otocol.pdf

I have one suggestion for you though. Support arc voltage sampling. Allow the user to send the time delay before sampling the voltage. If a 0 is set, disable sampling.

Yes, that is included already as Sample & Hold

When I was playing with one of the Linuxcnc developers who had written a THC component (that did not make the grade), I wrote a component that did this and I just hard coded the delay I wanted. I really did this becasue I wanted to make sure the THC control started with a set point exactly on the torch voltage to avoid any unexpected corrections ( I had some massive crashes during testing!). I used this hard coded approach for ages and even now where I have so much more ability to control things, I still use torch voltage sampling and never look at the torch voltage on the display. I think the delay will vary between tables based on the XY acceleration settings but I settled on 1.5 seconds for everything as it did not seem to vary with material thickness.

The only thing you probably won't be able to do is to defer sampling until it gets to the Feedrate. This ensures a valid sample is made if the timer expires on an arc or short segment and the machine has not been able to get to the feed rate. We can do that in Linuxcnc becasue we know the current velocity.

I am not planning to do this realtime, but it is in the back of my mind, but at 50mSec (internal loop frequency of uccnc for macro loops) it can be done via a simple digital input (not over modbus). Its not to compete with LinuxCNC + plasmac as that has it sown up with the THCAD, but its the Mach3/Mach4/UCCNC user with a IDC26 pin connector spare port (idiot proof design idea) or LPT port.

Much appreciated for the comments

[Rodw]

Robert, use an embedded controller that includes a spare UART and use that to talk to the Hypertherm. That takes it out of the timing loop.
You might be able to sample the voltage as a moving average over 50-100 readings and send the average back via modbus. That additionally would apply some filtering which would not be a bad idea.

Linuxcnc's Plasmac is no longer a THC, it is a full dedicated plasma controller. It even has a conversational mode with a shape library in the GUI (that I never use). Any comparison has to be made with the total controller package (ie Mach/THC or UCCNC/THC). Its that tight integration with the all seeing, all knowing motion controller running on the Linux PC that blows Mach with an external motion controller board out of the water.