Cutting issues, Esab table with UCCNC and DC servos

[SM1]

Hi, I'm out of ideas now.

Background. I have converted an old Esab Ultrarex oxyfuel cnc machine with new servo drivers and UCCNC Axbb-e controller. I have also made a Z-axis with ballscrew and stepper motor and floating torch holder.

Plasma source, Hypertherm Powermax 105 + Proma THC 150.

I have tuned the DC servo drives so they are quick and accurate. I have moved the encoders from the gearbox to the motors due to bad resolution and trouble with backlash between motor and encoder. Now approx. 609 pulses/mm feed.

The cutting results gets really bad if I go over 1000mm/min feed rates. I have tried with different acceleration rates by it doesn't make any big difference in cut quality.

The rack and pinion + gearboxes have a total backlash of 0.28mm on X and 0.35mm on Y. This value is compensated in the UCCNC software.

Please have a look at the YouTube videos below and let me know what you think might be the cause of my problems.



When cutting below 1000mm the results are quite ok, but I think it should be able to cut better. The Proma THC is not super, but it controls the height quite ok now (but with much less voltage settings than the Hypertherm cut charts).

[weldguy]

Looks like something may either be loose or binding in 1 or more of your axis. I would turn the machine ON and hopefully the motors power up and hold the axis' in place, then try to wiggle each of the axis to see if there is any slop. Then I would power down and if you are able to move each axis through it's range of motion and see if you feel and binding.

[SM1]

Hi, no. Nothing is binding. It's running smoothly when I run with larger radius, straight lines in x and y direction. But the smaller parts with higher acceleration needs are not turning out as they should.

The repeatability and response from the servos are good.
The only things I can find that makes the problems are a combination of the 0.28mm X and 0.35mm Y backlash, UCCNC backlash compensation and constant velocity planning. It actually works better with Mach3 and it has another type of trajectory planner than UCCNC.

The latest results of cutting is shown here:


But then it's slow cutting. I will try with some smaller parts again later. Unfortunately I got slag on the cutting shield so the cutting arc was inclined, making like parallelogram shape of the edges.

I used Hypertherm 105A nozzles and 85A of cutting power. I used the chart settings for 85A. Is it OK to use 105A consumables and 85A current? Should I have increased the pierce delay and cutting and pierce height?

[SM1]

The reason why I only used 85A was that I was afraid of burning my main fuses. I only have 400V 25A at the moment (possible to load up to 1.8*25A for one hour but then I also have the compressor and my house electric stuff on the same mains).