Another plasma component...

Grotius wrote: Mostly of time a following error can be solved by the ferror factor. Just attached a example to solve this. It works for servo (PID) and stepper systems. Interface type indepenent.

Thats the one I was trying to remember!

The joint following error on metric machines was the following setting, I had the same issue
"setp pid.N.maxerror .0005"

It’s obviously defaulted for imperial but way too small for metric

phillc54 wrote: I pushed another update:
- remove duplicated ini file entry - doesn't really affect anything, just a tidy up
- change resolution for arc voltage input scaling and offset
GUI files changed are:
configs/sim/axis/plasmac/native/plasmac_axis.py
configs/sim/axis/plasmac/gladevcp/plasmac_panel.py
configs/sim/gmoccapy/plasmac/plasmac_config.py
configs/sim/gmoccapy/plasmac/plasmac_panel.py

rodw wrote:
The on screen scaling does not work because of enforced upper and lower limits in the config screen as I can't enter an offset of 3659.375 (0-100 enforced)

Also the maths in my earlier post says the frequency counts per volt is 83.73 and you have to input the reciprocal of this which is 0.011943151 which is not logical. So I would recommend that the maths be changed to do the reciprocal stuff internally and also remove the enforced range on the scale. If that was the case I think it would work well with some explanation in the docs.

I have fixed the upper and lower limits, hopefully by enough to cover everything.
I don't want to change too much as I don't want the component itself to be dependent on any particular hardware.

I rigged up a 7i76e with a TCAD connected to the encoder input and two batteries for testing:
- run LinuxCNC
- open halshow and watch plasmac.arc-voltage-in and plasmac.arc-voltage-out
- set arc-voltage-scale to 1
- set arc-voltage-offset to the value read from plasmac.arc-voltage-in
- measure battery voltage
- connect battery to THCAD input
- set arc-voltage-scale to battery-voltage / plasmac.arc-voltage-out
could probably fine tune further but it was good enough for the purpose of the exercise.
My values were:
arc-voltage-offset = -3805
arc-voltage-scale = -0.000406

Cheers, Phill.

Im using a THCAD-300 which requires a voltage offset of 118200. I saw the max (upper) value on the arc-voltage-offset was 99999, so I had to increase it to 120000 so I could set my offset correctly. I realize that range isn't an issue for the THCAD-10s, but I assume anybody else with - 300 would run into the same issue.

@Bryan,

Has your THCAD-300 has jumpers on it? Then you can go a step back in scale to provide more full scale to linuxcnc. (= evolves in less accuracy, but no problem i think ).

OUTPUT FREQUENCY
The THCAD's frequency output can be optionally divi
ded by 32, 64, or 128 for
applications where the frequency is counted by soft
ware. Output frequency is selected by
placing a shorting jumper on one of 4 sets of pins.
Default mode is divide by one. The
calibration data on the THCAD is recorded in the di
vide by one mode..

www.mesanet.com/pdf/analog/thcadman.pdf

Uthayne wrote: Im using a THCAD-300 which requires a voltage offset of 118200. I saw the max (upper) value on the arc-voltage-offset was 99999, so I had to increase it to 120000 so I could set my offset correctly. I realize that range isn't an issue for the THCAD-10s, but I assume anybody else with - 300 would run into the same issue.

Yes, Grotius is on the money. The recommended setting is to set the THCAD to the 1/32 divide setting. Then you offset is only 3694.
By using this frequency divider, you get better accuracy as the THCAD has more time to average the results.

rodw wrote:
Yes, Grotius is on the money. The recommended setting is to set the THCAD to the 1/32 divide setting. Then you offset is only 3694.
By using this frequency divider, you get better accuracy as the THCAD has more time to average the results.

Ah, good points. Thanks Grotius and rodw.

Is there any benefit to switching it out for the -10 model? Was previously using it with raw arc voltage, now using it on a Hypertherm and could use divided voltage straight from the CPC