Why real time kernel is needed with board like MESA?

I don't know if you understand - if we apply a technique that is used in one piece of equipment to another - it may not work!
That is, the burners are different and can create a stable cyclone of different lengths - according to what I worked with, I would first put this one - OTC - Daihen D12000 - this thing cuts with a 3% slope of 40 mm, then Duramax S105 - S125 - the same quality but up to 30 mm, followed by Trafimet S105 and PTM100 - slightly lower quality and again up to 30 mm.
So if the burner rotates the cyclone well, you may not need to change the height. If necessary, it can be changed by up to 10%, ie if it is 2.5 mm - to reduce it to 2.25 mm.

And about real time and holes I don't understand what you mean?In LinuxCNC it may be so, but in others even the most branded ones are not. For example, Okuma to Model 7000 does not give a code error message at the beginning, even if there is a move that cannot be executed. Just before this line stops and then writes an error.
In my opinion, the code analyzer should crawl the entire code to decide if it is possible to execute it, if there are errors, if it does not violate the soft limits, and so on. Then he will "see" the hole and will create such a task, which the planner will turn into simple small tasks to send to FPGA.
My opinion !!!

Thanks. So you rate the OTC Daihen as a high quality machine then? I had not heard of it in our market.
Yes, Linuxcnc by default will report errors before allowing it to run. Plasmac filters and modifies the gcode to apply hole cutting mode adjustments etc.

By working with G code you do not know everything that is happening for example you do not know the exact velocity right now, It may not be possible due to machine acceleration limits. By working in real time in Linuxcnc, we know everything instantly. So think of it as applying adjustments in your FGPA but that the linuxcnc "Super FGPA" knows a lot more about the job so can do a better job managing the cutting and apply adjustments that cannot be done in a FGPA that has imperfect information.

My OTC Daihen robotic welder is a high end machine and it better be for 17K,



I did not know they made plasma cutters, I will look into it.

Yeah, I like it.

We are getting somewhere.

thefabricator03 wrote:

My OTC Daihen robotic welder is a high end machine and it better be for 17K,

I did not know they made plasma cutters, I will look into it.

They make it too but i dont think they will invest time to improve like Hypertherm, daihen dont sell nothing for CNC plasma. Daihen made one laser weld hybrid machine to join alluminiun with steel and started sell january 2020
In Japan i think plasma is first released by panasonic and after daihen, weld first was daihen and 30 years ago was easy to adjust the 2 knobs so they made their name, now both panasonic and daihen are equal in weld. panasonic 130A is more cheap than daihen 120A.
tokiox.com/wp/%E3%83%80%E3%82%A4%E3%83%9...A5%E5%90%88/?lang=en

That's right, OTC Daihen is actually for many quality Japanese products (although this plasma, as well as their welding devices, are made in China).
The D12000 plasma is made for CNC and robots and is at, if not above, the level of Hypertherm's Powermax series. I sold mine and only then I came across a welding robot from OTC Daihen. It has a 350A source and welds perfectly. Inside, the machines look like Hypertherm - made with money!

As for the corrections - that's what I can not explain. If the FPGA is given a task for - distance (inside the FPGA are set the types of speeds G0, G1, jog and accelerations in each of them) and type of speed - then the result will be absolutely accurate performance ALWAYS. With perfect rectangular pulses with perfect acceleration accuracy and reverse acceleration, completely synchronous, no matter how many axes we are talking about - 3 axes, 5 axes or 15 axes !!!
In addition, FPGA can work with feedback check, so as not to make a mistake, even if the problem is after FPGA!

stivemaster13 wrote:

That's right, OTC Daihen is actually for many quality Japanese products (although this plasma, as well as their welding devices, are made in China).
The D12000 plasma is made for CNC and robots and is at, if not above, the level of Hypertherm's Powermax series. I sold mine and only then I came across a welding robot from OTC Daihen. It has a 350A source and welds perfectly. Inside, the machines look like Hypertherm - made with money!

As for the corrections - that's what I can not explain. If the FPGA is given a task for - distance (inside the FPGA are set the types of speeds G0, G1, jog and accelerations in each of them) and type of speed - then the result will be absolutely accurate performance ALWAYS. With perfect rectangular pulses with perfect acceleration accuracy and reverse acceleration, completely synchronous, no matter how many axes we are talking about - 3 axes, 5 axes or 15 axes !!!
In addition, FPGA can work with feedback check, so as not to make a mistake, even if the problem is after FPGA!

Question for stivemaster13

Do you have problems installing kernel?

I'm just clarifying things (upto a level i know).

You dont have to RT kernel install it seprately. It is included in latest iso. Downloads page clearly explain which iso to download.

Pc's other than power saving features are generally good for good latency ( I can be wrong)

Thankyou all for valuable information.

Programming new modules to fpga is not an easy work.
most people here are already doing best. What problem do you have with real-time kernel? achieving different results is a different matter.

I know FPGA can give good result. But programing them (I think) is a whole new world. Many new contributors are needed for that work.


I also propose for the sake of community that we should hire new good programmers full time. We should also find ways to pay those new and old programmers. Stivemaster13 is right at some points, that a little new effort is needed for the betterment of Linuxcnc. I have profitted from linuxcnc a lot. PlasmaC only open new doors for me. I should not hesitate from donating for this noble cause.

I am not, in any way, indicating that it is not good at this stage. Every opensource project need money to sustain in long term.

pinder wrote:

Every opensource project need money to sustain in long term.

Agreed. Thats the problem for Linuxcnc and what is holding it back. Every other open source program I use has a commercial arm or sponsor somewhere. In fact, I pay for the corporate version of some. Linuxcnc has no structure like that and have no way of even banking a donation let alone knowing what to do with it.

But then we come back to the FGPA idea. Maybe that could become a breadwinner but I fail to see any value in that unless it was a Mesa like system. Linuxcnc's biggest hidden secret is the ability to write custom components. Today I started writing a simple component to manage stepper drivers with a boost signal. So when a drive is at rest or at constant velocity, the current will be automatically reduced so drives will run cooler but still have awesome torque when its needed. Its features like that that can't easilly be built into a FGPA. I really think stivemaster1 should take the time to understand the linuxcnc paradigm. With his knowledge, he could do wonderful things with plasma on a real time linuxcnc platform that competes with the Hypertherm controllers where plasma control is embedded in the motion controller. I'm sure they are not using a FGPA, all the data is on a CANBUS or similar ethernet network so its trivial for them to manage their machines across an Ethernet network.