qtdragon - probing issues

Chris, where are you putting these changes? git branch ????

If I recall correctly, I can just do a 'git pull' to update, then the compile only compiles what has changed.....
 

git pull
git checkout qtvcp_probe_max_z_fix

Roguish wrote:

Chris, where are you putting these changes? git branch ????

If I recall correctly, I can just do a 'git pull' to update, then the compile only compiles what has changed.....

 

Yes, unless you do a make clean, only changed files are built by make. Because the changes are only to python and .ui files, they don't need compiling so a make finishes up immediately but I always run it just in case.

Chris, 

I worked this out. Inside probing  works correctly if X & Y hints are entered,

If hints are not added it goes a bit crazy for no reason, See the screen dump and video

 

photos.app.goo.gl/SgRoQ9yWaFavzKNB9

There really is no valid reason to do what its doing. 
It should never move away from the face to be probed on an inside probe
It should not move towards an already probed surface.
 

Great information - thanks Rod.

well here is something interesting:
If you don't add a hint, because of the math, the distance can become a negative value.
The python function expected the value to always be positive.

The first MDI command adds a negative sign, so you end up with two negative signs.
And linuxcnc happy accepts two negative signs, and converts them to a positive number.
And that is why it went the wrong direction on the first move.

I can add error checks for length hints.

so a case  of two wrongs don't make a right  .
This just came to me.

Assuming the direction is positive towards the desired surface to probe (whatever sign that is).
Perhaps if could be solved as follows:
if the calculated start position was negative in relation to the physical start position (eg further away), then the move is truncated by that distance and the current position is deemed the start position.
That should eliminate the initial wrong way move...

Sorry gone cross-eyed writing asciidoc for getting Linuxcnc.... Hope the above makes sense

I suspect this also explains why I could not probe small holes...

Chris and Rod,
well, looks like it's getting better.
What ever the outcome is, please, please include good documentation so users know what to do, and what not to do.  Probes are expensive.....
 

Could you write a small sample of how the docs would be most helpful for use?
There are help files but they probably lack detail.

Does this probing system revert to machine units before probing? I just published a fix to probe-screen-ng related to this.