Re: [CAD_CAM_EDM_DRO] Re: Just How Accurate Is CNC?

softtechinit wrote:

>
>--- In CAD_CAM_EDM_DRO@yahoogroups.com, "John Heritage"
><john.heritage@v...> wrote:
>
>
>
>>0.001" is around 0.0254mm.
>>
>>The newest commercial machine, driven by linear motors, can
>>achieve a tolerance of 0.001mm.
>>
>>
>
>
>Thank you John for taking time to respond.
>
>The mill in question is an Nantong factory built CNC workbench mill
>(about 800 lbs. with electrical control box). The mill is virtually
>new, however the owner has replaced the stepper motors and ball
>screws and populated the control box (which was empty) with
>components that are generally available (and that he sells).
>Commercially distributed versions of similar mills from Nantong
>reputedly achieve positioning and repeatability in the neighborhood
>of 0.0001". As I said earlier, he is estimating only 0.001 on his
>machine.
>
>

That is a clear case of specsmanship. Yes, one axis can easily be moved
in .0001" increments. That is just resolution. But, a machine of any
appreciable size will not be able to hold such accuracy over any length
without going to extreme effort. Note that steel expands at the rate of
roughly 6 parts/million per degree C. So, for every inch of length, it
expands 6 micro-inches. If it is a foot long, it expands by .000072"
Therefore, a steel machine will be hard pressed to hold a tolerance of
.0001" on any piece one foot long, if even one degree C of temperature
change to the machine or the workpiece is allowed!

>It appears to me that his modifications were not too successful.
>
>

Or, more likely, he is being realistic, while the sellers of the
original iron
were talking blue sky!

Anyway, without referring to the work envelope, "accuracy of .0001" is
totally meaningless! And, nobody can deliver a stated accuracy over
some particular envelope without some pretty expensive tools to measure
it. Just knowing your leadscrews are accurate to X precision, and that
your CNC control can move to Y resolution doesn't even BEGIN to
address the real workspace accuracy. Do you know how to measure
orthogonality error? Check out the cost on a Renishaw ball-bar tester,
or a laser interferometer.

Jon