Re: [CAD_CAM_EDM_DRO] how straight is straight[rails]?

carlcnc wrote:

> I had a guy ask me if I can make a router within .001 straightness
> over 48",
> I am not sure this is possible with "normal" equipment.
> Taken literally that would mean an accuracy of .00002 per inch!
> {My ground straight edge is .002 per manf.}
> Do the "big boy" CNC builders achieve this kind of tolerance?

For routers, certainly not. For high-end CNC milling machines,
yes, they do. They need fantastically expensive Renishaw (and
other makes) of laser measuring equipment to prove the accuracy
of the axis alignment, the one thing that is harder to setup than
just getting the ways straight.

Precision, hand-scraped straight edges are usually good to
.0002" over their length, however long that may be. The big ones
have to be handled with gloves in temperature-controlled rooms,
and left to equalize for 20 minutes if you touch them, as thermal
warping gets worse as the straight edge gets longer.

> Just to waste some time I checked my 5x8 in X.Y. best I can get
> is +/- .005 over 6',using said straight edge.
>
> The rest of his request was " flat to .005,and less than 8K"

What does 'flat to .005" ' mean when straightness of .001"
is required? Flatness and straightness sound like the same
thing to me. Does he mean X and Y need to be straight to .001"
but Z can be good to only .005"?

What is he doing with such a machine? Does he need overall
positioning accuracy of .001" or just that any axis will cut
straight to .001"? There's a difference, due to cumulative
errors, and the effects of axis non-orthogonality.
For overall positioning to .001" over 48", in a milling machine,
I would think he's looking at about $250,000 absolute minimum,
and possible closer to $1M ! Like some of the machines that
run temperature controlled coolant through the ballscrews and
bathe the milling head to prevent any heating of the accuracy-
controlling parts.

Jon