Manual tool change + tool lengh touch off

andypugh wrote:

What I am not sure how to do is to make it all happen automatically when you issue an M6 command.

I think this should/could be a job for the postprocessor.

Has anyone experience with touch off probe hardware:
-Just a normal switch (could be unprecise)
- an induktice switch ( I could imagine that the induction is influenced by the size of the tool?)

There are quite a lot of touch-probe designs out there. The classical Renishaw type is probably as good as anything (three rods around a circle spring-loaded into the gaps between ball bearings. It is all wired in series and so any movement in any direction breaks the continuity.

www.indoor.flyer.co.uk/probe.htm is a good example of the design.
www.cnczone.com/forums/digitizing_laser_...ade_touch_probe.html is a very cheap and easy version of the design.

For a tool-length probe you would leave off the ptobe rod.

If your tools will always be conductive (not necessarily a safe assumption with some of the coated tips, though ordinary carbide conducts well) you could just land on a spring-loaded plate and let the electrical contact be the tool-to-plate contact.

So far I installed a touch probe and started experimenting with the gcode. But there are some errors I can't cope with:

If i follow the tool-length-probe.ngc example i touch of with

G38.2 Z0 f50 (f50 is modified)
when I try to run
G91 g0Z.2 (off the switch) I get the error:" Probe tripped during non-probe MDI command"

This could be solved by running the probe slower (f30) but if I want to have a big safety distance this could take a long time
Is there a better command to release the switch that ignores that the probe is pressed?

Second Error:

I touched of my tool the first time, joged a bit with the manual control, changed the tool length and run a G30. If I try to touch of again(G38.2 Z0 f30) I get "Startpoint to close to probe point"

Ok, I found out that G38 need to be run in G90 mode. But even if I try to release the switch with G38.5

Code Target state Direction Signal Error
G38.2 Contact Toward workpiece Yes
G38.3 Contact Toward workpiece No
G38.4 No Contact Away from workpiece Yes
G38.5 No Contact Away from workpiece No

which shouldn't cause an error, I get one

I am afraid I am completely out my depth here, as I have never done any probing.

I might get the chance to experiment later, but I have a lot of other things to do.

It is probably worth taking this query to the emc2 mailing list, as developer coverage is better there.

From the docs it really does look like you should be able to do a fast G38.2 onto the switch and then a slow g38.5 off of it. If that is not working then I suspect there might be a bug that needs squashing.

I found the reason for this error with the hal-meter:

If the "probe-switch" is open I've a clear FALSE if the switch is closed it toggles between FALSE and TRUE even if I press the switch so that I'm sure he's not at the break over point. And also with a second switch. I've already send the signal through an low-pass filter so I'm 99% sure that hardware couldn't be the reason.

Any Ideas?