Mesa THCAD Divider settings

Whats the best settings to configure the onboard THCAD divider for LinuxCNC and why?

I'm using a Mesa 7i76E, THCAD and Dewey's eoffset_pid component for torch height control

I have an associate whose also using the THCAD wired to a microprocessor for Torch Height Control on another platform.

Somewhere I seem to remember PCW saying to use the 1/32 setting (which I'm using at the moment) but my mate was thinking I would be better off with a higher resolution full scale reading becasue of the high speed PID control.

I know the THCAD manual says to use the divided settings for counting in software but I am using the 7i76e's "high speed encoder". I assume then its just sending a number back to LinuxCNC along the ethernet cable.

Finally, my mate has pointed out that the THCAD uses a Sigma Delta voltage to frequency converter chip which is one of the best ways of measuring a noisy voltage. Is there any need then to applying additional filtering inside LinuxCNC? I've seen that the torch voltage in Halscope is pretty stable.

The advantage of using the divided mode is that you can set the encoder filter to a lower frequency
and gain better noise immunity on the THCAD ==> ENCODER INPUT path. If this path is short It probably doesn't matter

The divider ratio does not significantly affect the resolution

This is because of the way the encoder velocity estimation works, it does not just count input cycles, it counts input cycles/ time between the first count and the last, so gives high resolution numbers even at low input frequencies.

You might use a lowpass filter to get rid of noise outside the bandwidth of your Z control loop

Thanks Peter, my THCAD is mounted a few mm away from the end of the 7i76e and I've got an external 5 volt power supply powering both boards. I've put a fair bit of effort into shielding all external cables. So the cabling to the encoder is pretty much as short as I could make it so I might try the full scale setting.

I'm a bit out of my depth here so if you had a moment, it would be great if you could explain how to set up the filter in a bit more detail. I kinda get you can limit the bandwidth to the upper frequency you want to read but I don't understand how the gain works to set this limit.

Hmm, I think the master docs are much better and include a worked example.
linuxcnc.org/docs/devel/html/man/man9/lowpass.9.html

So if say I was not interested in signals above 200 volts, I work out the frequency for that voltage and calculate the gain based on the example.

The lowpass filter components "gain" parameter has a somewhat unfortunate name.
Its sets the gain as a function of frequency (the bandwidth), The gain of the Lowpass filter
is always 1.0 at DC

This (bandwidth) frequency has nothing to do with the THCAD frequency

For example if you had significant plasma voltage noise above 50 Hz and you wanted a
50 Hz low pass filter you would set the lowpass filter gain to 0.270 (using the equation in the filter manual page)

Thanks Peter. I'm quite excited here. Dewey has found a value in his eoffset_pid component that needs modifying on a metric machine. So I've updated it and his sim seems to be perfect with my machine config settings. I will be doing some test cuts in the morning and I am very hopeful we will have near perfect voltage tracking. He asked me this morning if I could do some controlled tests on the THCAD so he can try and incorporate the THCAD response into his plasma torch simulator.