Re: [CAD_CAM_EDM_DRO] Re: doubling resolution strategy # 2

Ian Wright wrote:

> Hi,
>
> Thanks to everyone for the ideas. I don't yet have the machine together
> enough to make parts as a couple of you suggest but, while I had thought of
> this myself as a possible way of checking things, I realised that all it
> will do really is add in another whole set of errors. For example, how would
> you know what is tool deflection, inaccuracies in the diameter of a tool,
> offset errors etc. ? I know this is probably going a bit over the top but I
> am, after all, hoping to make some quite accurate bits for wrist watches on
> my little mill when its done.

Not over the top at all. If you are drilling a couple small holes to mount something
on a plate, it may not be real critical. Scribe some lines, center punch and drill.

But, if you are making hydraulic valves, small engines, gas turbine parts, optical
instruments, precision clock parts, or whatever it is that needs to be made
precisely, then you need to know ALL the sources of error, and think about their
effects on the part. Tool deflection is a killer in some cases. You may need slots
or pockets with small a inside radius, but the depth of the slot or pocket argues
for a larger diameter cutter to achieve less tool deflection. That puts you in a bind,
and you may have to make tradeoffs here.

> I am fairly happy that the problems I have are
> with measuring rather than the machine itself but, with the amount of talk
> on the resolution of machines on the list - accuracies to better than
> 1/10thou etc., I thought perhaps people were actually working to such
> tolerances and verifying that their machines actually did what they assume
> they are capable of. Perhaps not.

I can cut slots with about .003" accuracy from one wall to another without
extreme measures to compensate for the variations. Anything beyond that,
and I have to make measurements and then tune the program to shave off
another tiny amount. I'd strongly suspect anyone who claims to machine
anything to .0001" accurate dimensions of trying to fool us or himself!
(You can do this with careful measuring and fine finishing, but you can't
turn out parts accurate to .0001" with an end mill or lathe bit, just by turning
the dials to the right setting, and expect the part to be on target.)
Leadscrews accurate to .0001"/foot (or over the full length of travel) are
VERY expensive, and that may be static accuracy only, for anything that
will fit inside a Bridgeport-sized machine. Real accurate screws that can
deliver that kind of accuracy under load may need to be 3-5" in diameter,
and cost $100,000 EACH, for a meter of length! You just about can't get
linear scales of that resolution, without interpolators. And, to get scales
that are actually ACCURATE to that resolution, you're again getting into
SERIOUS money. They DO put scales like that on CMMs, but they only
work in temperature-controlled enviroments.

Jon