Advanced Search

 Great! Let's move ahead.

Let's talk about 0-10 volt range first.
The Mesa THCAD is a voltage to frequency converter. The THCAD10 has a 10 volt full scale. The THCAD 5 has a 5 volt full scale. The newer THCAD2 can be configured to either have a 10v or 5v full scale range. For your hardware, the 10v range sounds best.
We read the frequency using an encoder A input (usually on a Mesa card) and then scale it internally back to a voltage. Usually we set the THCAD's on board frequency divider so we read 1/32 of the frequency when sending it to Linuxcnc.

To receive a frequency, we would simply connect your hardware direct to an encoder A input and not use a THCAD.

It would be easy to apply a mathematical function to the encoder frequency to convert to a voltage in a real time component written in C.

There are other options to read both voltage and frequency. For example, appropriate Ethercat slaves for both voltage and frequency are available.
I think for Ethercat I would prefer 0-10 volt.

Finally, I might mention that for plasma torch height control, we adjust the height using external offsets (this offset is external to the motion controller). A scaled count (much like a MPG) allows us to take the input and output a height adjustment of the error. But because you output a distance, it might be possible to manage height control via one of Linuxcnc's PIDS.
 
Anyway, if you have a contact us option on your website, let me know, and I will reach out to take things further.

fantastic.

There is still a possibility of implementation, so I would like to adapt it to LinuxCNC as much as possible.

I am not sure if it makes practical sense to process data multiple times, which loses measurement accuracy and introduces additional errors at each stage C/f -> U -> f ...

i can made frequency output directly in sensor.

I see that the THCAD5 card processes voltage to frequencies in the range of 100 kHz - 1 MHz, but I can adapt to any frequency (lower frexuancy could be directly monitored in many "less efficient" controllers and send via cable easyer ), it can also be a PWM output or serial communication, which I can send a ready-to-use height without the need for the CNC system to process the signal, which in my opinion would be the best option.

In your opinion, what would be the best option?

Send one to Rod, her can make it work easily as he has THCAD that turn 0-10V to frequency, and he is very detailed when testing stuff.
I do not have a laser (i do have a big box of lasers, but He-Ne ones at low power) so i can not help with testing it. I can attach it to a plasma torch and see how long it lasts with all the interference plasma causes, but that is it. Oh yeah, Rod can also test this, as he has plasma machines.

Hello LinuxCNC community,We are a group of engineering students working on our semester project – building a CNC plasma cutter using LinuxCNC as the control system.Since we are developing the machine from scratch and learning LinuxCNC at the same time, we would greatly appreciate the opportunity to discuss and receive guidance from experienced users. We are particularly interested in understanding potential challenges, key considerations, and best practices for setting up LinuxCNC for our machine.Brief Overview of Our Machine:

• CNC plasma cutter with X, Y, and Z axes
• Z-axis with a ball screw and linear guides
• Floating head with a proximity sensor as a probe input
• Plasma cutter with a voltage divider and THC (Torch Height Control)
• LinuxCNC control, likely using Mesa hardware
• Torch Height Control (THC) integration via Mesa THCAD-2
• Voltage divider (50:1) for plasma voltage feedback
• Probing system for automatic height sensing before cutting

We have already explored various existing solutions, but we would love to hear from those with practical experience in setting up and running a CNC plasma cutter with LinuxCNC.Some key topics we would like input on:

• Choosing the right hardware (drivers, Mesa cards, sensors)
• Configuring LinuxCNC (HAL, INI, probe input, THC integration)
• Common challenges and pitfalls when setting up a plasma cutter
• Best practices for calibration and fine-tuning

If anyone is willing to share their expertise or discuss this with us in more detail, we would greatly appreciate it. A virtual meeting would also be very valuable for us.Thank you in advance for your time and help.Best regards,
Ibrahim Al-N and the 4.semster students at Absalon Kalundborg (Denmark) 

It is possible to do ohmic sensing from the tip. At least it is on a Hypertherm 30XP using relay ohmic, I imagine it would work the same with a THCAD. I have been doing this for a few years now without any issues. The only mod to my torch from memory was to insert a diode, the inspiration was from   here  and  here .

Got a THCAD2 from www.welectron.com in Germany a few days back, not installed yet as it was freezing lately.
Updates when i get to using it a bit, although i see no apparent reason for any difference from THCAD10 and THCAD300 i used before.

For now I am using the period as a base for calculating the frequency. Looks good so far. I will have to test for a while, but I think it should be fine now. Thank you for all your inputs!

there is no comparison. Linuxcnc's QTplasmac, a Mesa 7i96 and a THCAD2 will give you a high end system for almost nothing. Its feature set is very extensive. The difference is that the THC is managed internally by Linuxcnc, not an external THC so the system knows much more about what I'd going on and has much more ability to provide consistent results.
This is quite an old video but it shows what you can do

And the cut recovery features

nothing comes close!

Hello, my name is Daan Bent, 28 years old, and I’m from the Netherlands.
At the moment, I’ve started working on the mechanical design of a DIY plasma cutting table, with an option for a 4th axis to also cut tubes. For now, I want to focus on cutting plates, but I’d like to have the hardware prepared for a 4th axis later on.This brings me to the first problem I’m facing: I’m hesitating between Mach3, myplasmac, and LinuxCNC software/hardware. Given the forum, most of you will probably suggest going for Linux. My programming experience is very minimal, but I’m willing to learn and find it quite interesting. Do you think it’s doable? The advantage of myplasmac and Mach3 is that they seem easier to use in my opinion, but the options with Linux seem endless.Here’s the start of the design for my table, tips are of course very welcome! Over the next period, I’ll be focusing on the design of the mechanical part and diving deeper into the hardware and software.

 

I think one can not take for granted that EL3061's signal ground is the same potential as the workpiece ground. If you are lucky it might work, but to be on the safe side the torch signals should be isolated and free to float.

I had a quick look at the manual and I saw something that indicated the drop to zero was an index mark. My guess is that one of the PDOs needs to be set to remove it.

As I said earlier in the Ethercat world, a 0-10 volt analog input is probably a better choice but I have not tried one (assuming the plasma cutter has a voltage divider).. The EL3061 for example offers 500 volt isolation which is similar to the Mesa THCAD. To scale the voltage you would just need to use a mult2 component, send the output to QTplasmac and set the Qtplasmac offset to zero and the scale to 1.

If the pulse length method works for you it is even better. 100 ns resolution between two rising edges gives 0.1 Hz resolution and seems to be updated every cycle.
 

Yes, indeed. However since the drops seem to be random, I am not sure if it is related.

The manual also states that no frequency measurement is possible if the counter is blocked by the gate, but that in this case the period can be measured regardless.

And further: If an encoder signal is only present at input A/A and the frequency/period is to be measured, the terminal must be set to “Enable FWD count” in CoE 0x8000:03 [} 155].

   

I looked through the EL5101 docs and - did I get it right? - it measures frequency by counting number of edge rises during a 10 ms window? Could be where 10 ms comes from. Seems low resolution also, frequency will be in steps of 100 Hz.

Yes, it consistently appears to be around 10ms.

However, I believe I’ve found a workaround. I have noticed, that the EL5101 outputs the encoder period which is an absolutely stable signal. So I wrote a custom component to calculate the frequency based on it, which in turn can be used in PlasmaC. In the attached Halscope screenshot, you can see the troublesome frequency signal with drops from the EL5101 in red, the encoder period in cyan, and the calculated frequency in green.