Converting CNC-STEP HighZ-1000T to LinuxCNC with Mesa

No specific firmware needed to use step03 as charge pump. PCW handed over a bunch of commands in my charge-pump-thread which solved the issue. Fantastic!
forum.linuxcnc.org/27-driver-boards/3925...a-7i96-in-4-1-config

in the case you use the full internal controller
why dident yiu just go for the 7i92 solution witch brings you 2 full parport via ETH
likwise the one you used on mach4 smoothstepper

I also wanted a serial port to connect a 7i73 for interacting with a physical control panel.

Some progress, but not there yet.
I cannot get the machine to move.

I don't quite know how to debug this. I'll attach the config here if anyone has time to take a look.

- charge pump is working (measured with an oscilloscope)
- while homing X I can see some erratic pulses on the step00+ pin, but nothing moves
- All limit switches are "false"

I have a cheapo USB oscilloscope (Syscomp CGM-101) and very little knowledge about this, so I might look for the wrong thing.

Your step length setting is too short for most drives

I would set steplength and stepspace to 5000 unless you need higher than 100 KHz
step rates (and set direction setup and hold to 20000)

Thanks Peter.
I actually came up with this myself at the same time. I ran through pncconf for a 7i76 and got stepspace and length = 5000.

And.. IT MOVES! =))

What is driving these parameters in the wizards? I suspect that it comes from the StepGen settings? I ran the 7i96 config tool, selected preset Gecko 540 and got 20000. Did I make a bad selection or did the config tool give me bad values?

I thought "newsig emcmot.0.enable bit" newsig was retired

I think some settings are manufacturers numbers (but minimums)
and some have been updated to longer values that are more likely
to actually work...

Manufacturers minimum values should really never be used for working
values, and unless you bound the desired maximum velocity with the
steplength and stepspace, there is no reason to make the step pulse length
or stepspace any shorter than say 5 usec (5000 ns).

Likewise there is nothing lost by making the direction setup and
hold time 20 usec each (20000 ns)

By choosing these large timing values you can run almost any step/dir
drive. I think originally people assumed the drive minimum values were
optimum in some way, they are not...

Next roadbump - homing.

I think I'm going crazy. Nothing I do seems to satisfy the homing-monster. Is there any good reading about homing and setting "HOME" and "HOME_OFFSET" to correct values?

My scenario:

- limit switches are the same as home switches
- only switches at one end of each axis
- separate switches for each axis
X: limit at the far right (+570 mm), soft limit at 0
Y: limit at the "far away" (+1100 mm), soft limit at 0.
Z: limit at the top, i.e 0. Soft limit at -110 mm

The more I experiment with the parameters the less I understand.
- Jogging works perfectly and the machine sounds great
- Homing sounds terrible, as if the machine works against itself
- I get a lot of "joint n on limit switch error"

I tried to set debug=0 to get more information on the errors, but I cannot find any log files.

Also: My limit switches are inductive. How can I test these?

Have you looked at:

linuxcnc.org/docs/html/config/ini-homing.html

You should be able to verify the home switch action by watching the
home signal (or pin) with halshow