On Thu, 25 Mar 2010 12:33 -0400, "Mark Wendt (Contractor)"
<[email protected]> wrote:
> At 12:18 PM 3/25/2010, you wrote:
> >On 25 March 2010 14:49, Mark Wendt (Contractor) 
> ><[email protected]> wrote:
> >
> > >         I'm not sure at this point.  I'm not sure if the errors
> > > showing up are cumulative or not.
> >
> >Not if they are due to backlash or microstepping nonlinearity.
> 
> I'll have to make sure one way or the other.  We have yet to 
> determine that it's truly backlash that's causing the problem.
> 
> 
> > >  Actually, I am looking for + or -
> > > .001" accuracy on the strips I'm going to be cutting.  That accuracy
> > > will be the height of the triangular cross section of the strip.
> >
> >But that accuracy is needed in the Z axis, which you have already
> >stated is spot-on. (I am guessing it is a ballscrew?)
> 
> Yes and no.  In order for the taper itself to be accurate, those flat 
> to apex (Z axis) dimensions have to be met at darn near the spot-on 
> length (X axis).  Remember, there's going to be 6 of these strips 
> glued up to make a rod section, and that rod section would look kinda 
> funky if the flat to apex dimensions didn't quite line up.
> 
> 
> > >  On
> > > the tips of some of the rods I make, I routinely hand plane down to a
> > > .025" flat to apex height, and can usually hit that measurement
> > > within a thou.  If the machine can't do that in production, it ain't
> > > gonna work for me.  So, I need to make it accurate.
> >
> >I don't think any rack-and-pinion system will be accurate to sub-thou
> >positional accuracy. However, as you are cutting a slow taper you need
> >far less X accuracy than Z accuracy.
> 
> Don't need to be sub thou.  Just + or - .001".  If the X axis is 
> consistently off .001" or .002" per inch, and it's cumulative, it'll 
> end up being close to .050" by the end of the strip.  I'm thinking 
> that's a bit much.
>

If it is truly cumulative, you can fix it by adjusting the scale.  If
you gain or lose 0.002" for every inch you travel, just change the axis
scale by that ratio.

You might want to invest in something like this:

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=320505470762

These sets were used back in the day on jig-borers.  There was a
v-groove
in the saddle, running parallel to the table travel, and another groove
in the knee, parallel to saddle travel.  At one end of each groove was a
sensitive dial indicator.  A boss on the table or saddle moved along
the groove at the other end.  Say you wanted to locate the table at
7.923
inches.  You'd set the micrometer head so that it was 2.923 inches, then
put it in the groove.  Then you'd add 4" of spacer blocks (a 3" and a
1").
Then move the table toward the indicator.  The table boss pushes the
entire stack, and when the indicator reads 0.000, you know you are at
the
right place.

There were two sets of rods and two micrometer heads so that you could
set both axes of the jig borer.  Nothing stops you from using all the
rods on one axis.  For your purposes, you don't need the micrometer
heads.
In fact the best deals on such sets are the ones missing the heads.

Most sets, like the one in that auction, have two 12", two 6", two 3",
two 2", and two 1" rods.  That lets you build stacks up to 48", in
precise
1" increments.  Rig a temporary V-groove down the length of your table.
It can be as simple as two strips of stock clamped down with a gap
between
them.  Fasten a block to the table at one end, and attach an indicator
to
your gantry.  Touch off and zero the indicator against the block.

Then move 1" away from the block, insert a 1" rod, and write down the
indicator reading.  Move another inch, replace the 1" rod with a 2" one,
write down the reading.  Move another inch, etc.  You can use various
combinations of rods to take a reading every inch over 48" total.

Once you have 48 readings, you can plot the error vs distance.  Fit a
straight line to the data - the slope will tell you the scale error.
The residual after you correct the scale error might be random, or it
might be cyclic, once per pinon rev, or whatever.

If you are really dedicated, get a few gage blocks in the 0-1" range,
and
you can make measurements more frequently.  A half-inch block will let
you measure every half inch.  A set of 0.1, 0.2, 0.4 and 0.8, or 0.1,
0.2
0.3, and 0.5 blocks will let you measure every 0.1" - although that
would
get incredibly time consuming over 48".

If wind up with errors that are non-linear but repeatable, EMC2's screw
error compensation can help.  You can enter up to 256 points along the
length of the machine.  That would let you correct every 0.2" if you
really wanted to.

John Kasunich
-- 
  John Kasunich
  [email protected]


------------------------------------------------------------------------------
Download Intel&#174; Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
_______________________________________________
Emc-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/emc-users

Reply via email to