In article , artfulbodger says...

Could someone describe the kinematic mount?

The idea is you provide the miniumum amount of constraint to
define the geometry you desire.

My south bend lathe has the bed constrained at each end, by two
bolts. So it is possible to impart twist to the bed if one
tightens down the two headstock bolts, which then define the
beds front/back tilt, and then tighten down the two tailstock
bolts, which might want to pull the bed down to a *different*
front/back tilt. Likewise when the cast iron bed expands more
than the steel cabinet, the bed is going to bow up a bit in the
middle.

Hardinge gets around this by mounting their bed at the headstock
by pulling it down onto two spherical mounting points with heavy
springs. So now the bed has been constrained to have a certain
front/back tilt. It's still free to tilt left and right though,
pivoting around the other axis of the two spherical mounting
points.

The tailstock end has a single ball that protrudes from the botton
(actually a steel bearing ball, sitting in a countersunk recess) which
fits into an upwards-facing V-groove that is the third mouning point
on the cabinet. This means the tailstock end of the bed now is
restrained to give the bed the desired left-right tilt, and further
that any differential contraction or expansion of the bed with
respect to the cabinet (one is steel, the other cast iron) is
permitted because of the transverse V-groove on the lower mounting
point.

Of course the tailstock end of the bed is held down with the same
stiff spring setup as the headstock end.

Because of the single-point mount at the tailstock, there is no
way on earth there can be any twist imparted to the bed by forces
that arise between the cabinet and the floor. Likewise the
differential thermal expansion between bed and cabinet can not
bow the bed up and destroy the accuracy.

Hardinge finished their beds mounted on those same three spots,
so when they are installed in the cabinets, the surfaces of the
ways agree again with the factory versions. They're not twisted
or bowed.

Jim


--
==================================================
please reply to:
JRR(zero) at pkmfgvm4 (dot) vnet (dot) ibm (dot) com
==================================================