Open Source Controller with internal THC

stivemaster13 wrote: Ha, well, you probably don't understand me. I'll explain again, giving an example. Here's a machine I'm making now and shortly before I finally release it. With it, what you see as a monitor is actually all in one, a touchscreen 17 "computer.
Imagine that the computer is in the dashboard with the drivers of the servomotors - it is on the back left and is not visible in the picture. The cable guides are 12.5 meters. Do you understand me better now?

I am sorry friend, I am not able to understand what you are trying to solve?

You mount your display on gantry.
What can go right?

All you need is a lightweight gantry,
A computer Screen fixed on the ground or table but not gantry.

For Marking Origin, you can use camera in PlasmaC or A wireless MPG with crosshair Leds with plasma head.

There are so many possibilities. I have made welding machines, Plasma machines and now working on Milling machine with ATC.
In future a laser too.

Hardware or software are not limitations in any way.

You dont want a mesa board, Fine.
Use a D525 motherboard, smaller than the mini pc you mentioned and a parallel port breakout.
you can build anything with that only.


Still if you want a pc on gantry. Use a small pc like odroid h2 on the back of monitor, and use mesa board in control panel and connect with ethernet, no 12 or 15M barrier there make it as long you want.

I know linux is difficult for some, so is building cnc machines. If you dont want to go in details, people can be injured due to you lack of control on your software.

It takes some time to get around things, for which these forums are helpful.
For the most time I never started a new thread, just reading old threads solved my problems.
Now I got bored from axis GUI, so I am building a GUI for plasma here on this forum. you can check MonoKrom

Possibilities are endless. Just dont reinvent the wheel.

If you dont want people to understand anything. Then mach3 is fine for most part.

But trust me friend, that software is 10 years old with no updates. A there are as many flaws as you can count.

You should be more interested in what you want to do.
If you can build or suggest something goodwith off the shelf coponents ( stay away from propreitory and licenced stuff), community will join you.

stivemaster13 wrote: Ha, well, you probably don't understand me. I'll explain again, giving an example. Here's a machine I'm making now and shortly before I finally release it. With it, what you see as a monitor is actually all in one, a touchscreen 17 "computer.
Imagine that the computer is in the dashboard with the drivers of the servomotors - it is on the back left and is not visible in the picture. The cable guides are 12.5 meters. Do you understand me better now?

This design could be achieved quite readily and I have a similar idea taking shape. I am not keen on the all in one. In fact I just removed one and replaced it a stand alone 15.6" touch screen and Odroid H2+ PC. This gives you more flexibility to upgrade to an industrial quality Touch screen for extremely harsh environments.


To achieve this right now with Linuxcnc, all you need is a Mesa card to do all your IO and control the servo motors (7i95, 7i76e or similar) located in your out of sight control box.
1. You connect your PC to the control box using a simple ethernet cable.
2. You place a Mesa 7i73 with your PC to connect all of your external controls at the operator HMI panel.
3. You use a second ethernet cable to connect the 7i73 via smart serial RS422 to the main Mesa card in the control box.
4. You could run a third ethernet cable from the UART on the Odoid H2+ back to the control box where it could be used to connect to the RS485 interface on a Hypertherm plasma cutter so you can control amps and pressure. It would be possible to eliminate this third cable if you use the Mesa UART at the control box end to talk to the Hypertherm but custom coding would be required.
5. If you need to run a seperate estop circuit you could consider utilising the spare cables in the ethernet cable in 1. to trigger a safety relay.
6. You use the second ethernet port on the Odroid or a USB wifi dongle to connect the machine to the local network/internet.

I am happy to explore this further with you

Strange thing, the fact that I'm not English-speaking is obvious, but not understanding me to such an extent is already strange.
I tried to explain to Himarc3D what the problem with the HDMI cable would be, but in the end it happened that you don't understand me.
In fact, I have made over 50 machines - for plasma, gas cutting, metal and wood milling, lathes and welding manipulators and I am well aware of where and how the monitor, computer or controller can be placed.
With this machine I have deadlines that do not allow me to experiment, and believe me I have already conducted a lot of them, for example, more than a year continued my fiasco with a Russian - myCNC (LinuxCNC) with built-in THC. super unfinished, but for a solid price.
So this machine will be with Windows 7 + UCCNC.
The reason I write to you is far from my dream to appropriate someone's work, to have rights over LinuxCNC or something like that.
Rather, I'm interested in something else - we all know that the many Chinese manufacturers that have been breaking through recently, on cheap offline CNCs, have based their controllers entirely on LinuxCNCs. Is Western civilization aware of its backwardness and the fact that it has long since given up its initiative both in economics and in software development.
You will agree that the lack of development of the public LinuxCNC (On the official site the latest version 2.8 was uploaded less than a month ago) is very strange. Another strange thing is the lack of desire to produce hardware, that is, the lack of competition. I am directly referring to the ubiquitous MESA - for example, what exactly in the 7I96 card costs over $ 120. I use them and I am very happy. The quality of production is very good and works as the Chinese wrote. Well, it looks a lot more professional than a bare board right? If it could be used as FPGA of MESA ....
Couldn't we use the Chinese because they can best make cheap boards, and we could make our own drivers and use them?
Such a card has no reason to cost more than $ 50. When developing in the direction I wrote about earlier - PC + FPGA + THC / divider + relay block will cost less than $ 200.
Then a serious manufacturer, if he can think, understands that in order to have a good company controller, it is good to have a free version that allows people to buy cheap parts and assemble their own CNC controller. With the development of such a community, he will test the extras which he will then put in the paid version, which of course will have slightly more expensive parts in one box and most importantly with support !!!
We don't see anything like that, do we?

Have a nice day !

stivemaster13 wrote:

again I am explaining only 2 things

1) mesa is not compulsory. you can work with parallel port. I also think it needs some competetion. +1 you for this.
People are working for new driver boards, I assure you that. You are also welcome.

You are right at some places. I think PC should only send instructions and rest of work should be on PLC or fpga. That should work, but real-time kernel is more important. CPU gives us opportunities like openCV, PlasmaC, Constant surface speed and many more. FPGA wont be able to do that on its own.

2) use a small pc on monitor's back ethernet will be 1M. Use ethernet card like Mesa an ethernet can be much longer

Thank you .... you still didn't understand me. What I described is just one of the options, of course. For 10 years I have made all possible options without a MESA card, but with FPGA from other manufacturers.
You are not right about the capabilities of the FPGA - the Fanuc controller, for example, is a PLC / FPGA in one and can control the interpolation of 9 axes (standard, up to 12 in modification) and up to 32 in general, using a direct connection to the drivers, Ethercat, CanOpen, Mechatrolink, RS485 and so on.
In other words, if we set the interpolation to 5 axes, for example, to 5 separate cheap FPGAs - they will execute them absolutely accurately, not to mention the feed rate or surface speed. These are specialized circuits that perform the critical role for the PC, to clock the pulses in the case of the motors super smooth, regardless of their speed and complexity of movement.
So the whole problem is whether you can create a task for them properly. They can perform it with precision with a built-in PLC algorithm such as PID for example or whatever we decide there.
I give an example, on one of my machines - a metal mill, I have had such a controller XC609M for almost 2 years. Of course it works on Linux and inside it is made like this - the main controller is STM32 (from $ 5) and as FPGA for interpolations on each axis use another STM32 (from $ 2). One on each axis!
It works great, I can't say a bad word, except for one thing - the planning toll road, they have a problem with the CV. That is abrupt interrupted uneven movement when performing small curves.
Yes, but I have one DM800 , also with Linux. It is a bit simpler, but there is no problem with CV. Inside, it is again made in the same way - with STM32 in the role of FPGA
Think about it !

What is the benefit of what you are suggesting?