To: atm{at}shore.net
From: "Matthew L. Brown" 
Reply-To: "Matthew L. Brown" 



>>If the 3 grooves come close enough together, meaning less than the
>>spacing of the balls, the balls will fall into the grooves the same place
>>every time.  You don't need high machining accuracy to get excellent
>>repeatability.  Of course the grooves have to be long enough so that the
>>balls can find the one position that they will all drop into the grooves.

If the grooves have to be short, as they might for a telescope cage, then
you'll need to keep tolerances only reasonably good, say 1/8" on
position on a 10" diameter scope.

One way to think of it is, mentally drop one ball into a groove (all the
grooves in this example are oriented radially away from the center).  That
ball is free to slide in that groove.  Now rotate the part with the balls
in it so a 2nd ball drops in the groove.  The two balls are still free to
slide in their grooves, though not independently, by rotating and
translating the part.  If the grooves are infinitely long, you can always
move things so the 3rd ball drops into its groove.  Real world grooves have
finite lengths, so it is possible that one or more of the balls runs out of
groove before it drops in.  This is where you'll need a little ordinary
precision.

I use this all the time where I work, and it really does get micron
repeatability for stiff, clean parts.


>Not quite...this would assume that the 3 grooves intersect the
>3 balls exactly. The true kinematic mount with 3 balls has a hollow
>cone for one point, a radial v groove for the second, and a flat surface
>in ~ the plane of the 3 balls for the third. The cone/ball constrains
>all 3 linear directions, the groove/ball together with the cone/ball
>constrains
>2 of the 3 rotational directions, and the flat surface constrains
>the 3rd rotational direction.

--- BBBS/NT v4.00 MP