On Monday 14 January 2019 21:53:54 Jon Elson wrote:

> On 01/14/2019 04:37 PM, Gene Heskett wrote:
> > Greeting Andy;
> >
> > The docs for G92 are a bit hazy, I find that with no arguments, it
> > still puts the machines 0,0 at where the machine is ATM, which is
> > not what I needed.  Is this intentional?
>
> Hmmm, that should NOT change any offsets, unless it is a
> special case of G92, with no axes selected.  Otherwise, how
> could you offset ONLY one axis, the others should not be
> changed.
>
> Jon
>
Andy said I should see an error, and its possible that these 
alignment .ngc's are not being used correctly. No clue what LCNC does 
with a syntax error in a subroutine, which is how this stuff is run from 
the hal_mdi. They came with no dox not even a README. So the intuitive 
thing to do, is start at the anchor point, homeing to the 0,0 at the 
bottom left of the workpiece. record that. Then run up the edge to the 
far end of the piece and bring it back to touching the edge, and tap the 
align x or align y which then calculates the error angle and applies 
that. But those two button routines also contained a G92 x0 y0.  And 
that jumps the home 0,0 to a foot away from where you started, It does 
apply the angle correction, but I've not tested that to determine if it 
corrects the error, or doubles it. So ATM I've taken them out as I don't 
believe aligning the machine to the workpiece should effect a change in 
the home position.

So until I'm cleared to actually do something so I can finish this 
frankenstein* build with my left arm, thats where I am with this. All I 
really want to do is make the rotation correction to align the machine 
to the workpiece without effecting the 0,0 position.  If that needs 
changed to keep the work on the table, that's what we have touch offs 
for.

This also needs a bunch of xml stuffs to hook it up and tally the 
sequences, and I got that hooked up yesterday.  Any discovered errors 
are a matter of swapping +  and - signs in the 2 subroutines, on 
designed to measure error along the X axis, the other along the Y. 
Mutually exclusive but a miss click on one is corrected by clicking the 
other. And of course there is a clear button to restore the default r0, 
which may not be, this particular code is rotated 90 degrees, more for 
testing this than for any machine limits reason, but that detected angle 
is a global variable the subs can access, so they apply the correction 
to the main.ngc's preset angle.  IOW, it won't over-ride what you put in 
the actual carve it main routine, only trim it for miss-alignment with 
the bed T slots. Thats the general idea at least, and the other reason 
for a fixed 90 rotation. However once this align thing is working, that 
almost becomes a never mind. I like that.

One of my next jobs for the lathe is to make a long contact probe like 
I've seen in some of this old tony's youtube videos. Need some drill rod 
about 3/32 diameter for raw material I haven't even ordered yet. Ticker 
sorta took precedence.

*frankenstein in this case means I'm using higher voltage (14->28) for 
improved rapids along with the 2M542 drivers from the old HF mill.  And 
in the final, lets make some swarf version, the only thing used in this 
things supplied driver box will be the vfd, which runs on wall power 127 
volt AC. I might even graft that to a real heat sink and put it in a 
smaller box along with its fan. That shelf would lose about 20 pounds as 
that steel box must be 14 gauge steel.

Thanks & take care, Jon.

Cheers, Gene Heskett
-- 



_______________________________________________
Emc-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/emc-users

Reply via email to