On Friday 01 March 2019 20:04:09 Kenneth Lerman wrote:

> I don't bother with centering the ball with respect to the probe. I've
> attached arms to my probe so that I can accurately rotate it 180
> degrees against a stop.


Since my probe when mounted is well below the motor, 3" or more and 
trailing only the wires what does the arm stop against?  Seems like we 
need a pix.

(and mine is now stuck in the 1/4 er11 collet, apparently the collet 
locked itself to the beginnings of the thread on the shank as I was 
pushing it up because its too [fill in fav expletive] long)

 I then do a calibration: 

>
>    1. I probe the center of a hole and record the X and Y values (call
> them X-A and Y-A)
>    2. I rotate the probe 180 degrees. Probe the same hole and call the
>    values X-B and Y-B.
>    3. The average of these values is the true center X-C and Y-C. X-B
> minus X-A is the offset X-O. Similarly for Y-O

If you used signed arithmetic in arriving at the average, addition at the 
end should be correct. I think...

>    4. Now, when I use the probe, I position it the same as it was in
> step one. I then add X-O and Y-O to the coordinates returned by the
> probe. (I may have the backwards -- I might have to subtract.)
>
> I've written subroutines that record the calibration values and record
> them.

I've written similar routines, using a spinning sewing machine needle 
threader reformed to fit inside the open end of a 3/32 brass pipe inset 
into a corner of a pcb pallet, calibrated the resultant offset from the 
corner of the pcb blank seated in the pallet, and done double sided 
pcb's accurately enough that holes drilled halfway thru the pcb, meet 
that same hole when turned over and the drill file run for the other 
side, neither of which is drilled deep enough to damage the pallet. Cost 
for the probe a nickle for 1/2" of 3/32" brass tubing, and $3.95 for the 
needle threader and 25 cents for 2.5" the same brass pipe the wire from 
the needle threader is now soldered into. Repeatability .0003 worst 
case.

But it doesn't work with alu due to the insulating qualities of the alox 
film on the alu. To make a 100% dependable contact detector to alu needs 
about 450 volts more in order to punch a hole in the alox. With brass or 
steel workpieces it works great.

> I've generally found that probing is repeatable to within a few
> tenths. (That's with a cheap -- less than $100 -- probe where the ball
> is visibly off center.)

I have to agree, but I'd like to see a pix of how you've setup the stops 
at 180 degrees.  And it occurs to me that a magnetic holder might be a 
good idea to keep the wired probe from being pulled off the stop by the 
wires as the measurement is in progress.
>
> I should comment that most people seem to probe by approaching in a
> direction at (relatively) high speed. Then backing off. Then probing
> again at low speed. I have no evidence that the common process is
> better than the one I use. I approach at "high" speed. I back off at
> low speed and note when contact is lost (technically, the switch
> closes again). That's the probed location.

That also should work well as long as the probe isn't rotating. With my 
rotating wire probe, contact may be only 5% of the rotation, its the 
discharged capacitor that remembers the  contact long enough for lcnc to 
recognize it, so I use the high speed, back off 20 thou and come back to 
a fresh contact at a much lower speed, which is the reading my routine 
call "it".

Like you, I've no data to say one way is better than the other. Its 
entirely possible I'd get plumb useable data with a slow backaway. But 
excessive backlash comp could easily mess with you.

I will state that that higher spindle speeds tend for more consistent 
results, but the diff between 100 rpm and 1000 rpm is less than .0002", 
so to me its a shrug.

> It just works.
>
> Ken
>
> Kenneth Lerman
> 55 Main Street
> Newtown, CT 06470
>
> On Fri, Mar 1, 2019 at 6:03 PM Gene Heskett <[email protected]> 
wrote:
> > On Friday 01 March 2019 11:31:45 Jon Elson wrote:
> > > On 02/28/2019 10:23 PM, Gene Heskett wrote:
> > > > Jon, have you considered setting a digital caliper to say
> > > > .2500" and locking it, then clamping it down well enough
> > > > it won't slip while you probe its gap.
> > >
> > > I homemade a ring with a pretty good inside bore, but it is
> > > not as good as a real inside mike standard.  Anyway,
> > > measuring a gauge block with the probe lets me do the best I
> > > can with my CNC mill.  I get a combination of machine
> > > backlash and probe accuracy, no idea how much of each is in
> > > there.
> > >
> > > Jon
> >
> > I just hooked mine up, pretty bad, and so eccentric, around .125" of
> > wobble at the ball, same diameter as the ball, its going to have to
> > be locked tight in order to meet the .005 accuracy it claims. 2
> > problems, when installed I have, when z is at top of travel, just
> > about the panel thickness of clearance with its default 2" long
> > probe, to a 3/8" thick spoil board, so I need to find a very short
> > probe, an inch or less. Once thats done and it epoxied into place, I
> > may be able to adjust the bottom spider to remove the eccentricity
> > at the ball.
> >
> > I measured a .250" caliper 3 times, came up with .176 once, (.251
> > with the ball), .187" once, (.262 with the ball, and .174 (.249 with
> > the ball)
> >
> > But it turned a few degrees so those measurements are exactly the
> > same as found on the ground behind the male bovine. :)
> >
> > Is it worth trying to fix? Or just another way to keep me out of the
> > bars?

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>



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

Reply via email to