Re: What is the weak link in machine precision - Knee Mills

I spent a number of years doing metrology & calibration and began to
realize NIST's 4:1 calibration accuracy ratio requirements contained a
bit of specmanship in a century-old mechanical world with a 2:1
infrastructure. There are so many places on a complex machine where
you can cut corners, and true, some companies are flakier than others
by a long shot. SPI used to drive me crazy with their sloppy thread
gage tolerances. Some companies just take a wild stab at a spec on
average, others truly design to be accurately within it. Typically, I
found newer machines had a greater degree of modern engineering, that
is, they were not as overdesigned. A given chunk of metal might (1)
wear quicker, (2)have more flex, (3)allow annoying resonances, (4)not
be as dimensionally stable over time, (5)have actual measurements that
were borderline at best, and (6)be part of a poor overall design. You
just can't know without spending much time actualling measuring and
playing with every little detail of a machine until you truly *know*.
I found I much preferred the metal of an old, stable, solid,
*overdesigned* machine to a new one (barring serious problems, of
course.) *Sometimes* you can retrofit older stuff because it can
support greater accuracies and precisions than it was built for.
Other times, it is enough that an older machine will truly give you
the accuracies you seek under a wider variety of conditions than a new
one with too much "specmanship". With all this investment, it is
truly worth the money to buy some basic measuring equipment--a good
4-way caliper, dial indicator, and test indicator at the least. ;-)

Don

--- In CAD_CAM_EDM_DRO@yahoogroups.com, "metlmunchr" <metlmunchr@y...>
wrote:

> --- In CAD_CAM_EDM_DRO@yahoogroups.com, "zipdrive2k"
> <zipdrive2k@y...> wrote:
> Enco has a mill they state the specs are
> > 0.00078/ft. error for squarness and parallelism.
>
> Our friend Ballendo has often mentioned the overstatement of
> accuracies by sellers of low end machine tools. With this in mind I
> dug out the original verification documents for one of my machines.
> The limit for parallelism is .00078/ft, while the limit for
> purpendicularity (squareness) is .00118/ft. Notice this squareness
> tolerance is 50% greater than what Enco states. The actual
> measurements for my machine all fell within the stated limits. Some
> were substantially less than the allowance, while others barely made
> it under the wire. It's possibly worth noting that these numbers
> are from factory tests on a machine that sold for slightly over a
> quarter million dollars. Do you think it's possible Ballendo has a
> valid point regarding overstatement of accuracy? or does Enco have
> some secret no one else knows when it comes to machine
> construction? My offhand guess would be some of the better
> Taiwanese knee mills will equal Bridgeport's geometry numbers (I
> dont know what BP's specs are), but a further guess would be that
> most of the Chinese iron will fail to equal BP's numbers by a factor
> of 2 or more. If the upper limit was indeed .00078/ft, then the
> average machine could be expected to have the geometric accuracy of
> a Moore Jig Borer. I kinda doubt that's the case.
>
> Cliff