Thanks for the suggestion ZASto. I found your link to the Big-Tex script on another thread (for those interested it's
https://www.dropbox.com/sh/7c5lmhl58lyhzn7/AACy4fPjUVi1RIX4Nh0Sm884a?dl=0) and took a look. Although there's some good stuff in it, I'm not really keen on switching screensets, etc that it requires. And anyway, I'm thinking of taking a new approach to zeroing and am part way through scripting my solution.
[A little background: I'm a retired software engineer machining aircraft prototypes as a hobby. I run big tools (1/2" shank) in metal and just recovered from a 'wrong-z crash' that hurt not just the tool, it hurt the machine (bent $pindle & z-$crew) so I'm being extra paranoid... After talking with some experienced machinists who actually know what they're doing, I learned that zeroing directly is for newbs and the right way to do it is to zero onto a gauge-block, removing the block whenever jogging into the material, then checking clearance by sliding it in. Much safer! But that wasn't the cause of the problem, running a crappy script was. (Not blaming here, except maybe a little for the Mach developer(s) who should have provided a template for 'the right way to do it', I bet that would have saved many tear$!).]
So, my
big idea (ok, one of them) is this: I use a spring-touchplate with about 1/4" spring travel. For the manual tool change scenario (clicking the Auto Tool Zero (ATZ) button, not executing a T# M6 code) I want to manually jog down to touching a 1" gague block
on top of the touchplate, then click ATZ. The script will wait for contact to break (or possibly contact then break), then G90 G31 Z-1.125" (roughly speaking), limiting Z motion on the probe and reducing the risk of overrunning the plate if anything is wrong. Yes, paranoid - but paranoid enough?
For a 'T# M6' tool change, I'd like to do a superset of this, moving down a known distance from the tool change position (using absolute positioning mode) and initiating the above stuff from there.
I know this is a little unconventional so I'm wondering what the more experienced folks think?
If it turns out well, I'd be happy to share the scripts, collaborate with others, etc to perfect it for all scenarios and prevent others from the $1000 cost & month of downtime I experienced from using but not fully understanding what was going on in that [
name redacted to protect the guilty] script.