Re: doubling resolution strategy # 2

Ian,

The simple answer is:

With something KNOWN to be more accurate than what you are checking.
For example, a laser interferometer. A renishaw ball-bar. An
expensive "glass scale" linear encoder. and finally, something we all
can use, and most can afford; Gauge blocks (also called Jo-blocks)
and an accurately REPEATABLE dial test indicator. If the DTI is not
accurate then the comp table will be skewed, but with a repeatable
DTI,it will be universally skewed, so the effect will cancel out.

Since we will be moving "around" the gauge blocks, be sure you have
repeatability in z (if you're gonna 'hop' over 'em) or Y (if you're
gonna 'steer' around 'em). Also try to minimise the distance
travelled to hit zero (or your current "sought" location). I suppose
you "could" offset BOTH zero and the desired measurement, so
this "sideways error is nonexistent. also use the dti to be sure the
jo-block "stack" is NORMAL to the axis; otherwise abbe and other
errors will creep in...

Basically, you set up a known length (with an offset block at zero,
so all your readings are from one direction), and read the error off
the indicator. Not fast, but as accurate as your DTI and jo-blocks...

Of course if you can afford any of the "above this option" in the
list presented at the beginning of this post, then that will be a
good choice as well :-)

Hope this helps.

Ballendo

P.S. It's nice to see someone actually trying to find the "real world
accuracy of the machine. So many just "trust" or "assume", based on
resolution...

P.P.S. You could also design some "parts" which can be tested for
size/geometry/location. Based on the metrology of the parts, you will
know what your machine can, and can't do. The trick here is in the
design of suitable "test" parts... But really, if the machine makes
the parts YOU need to make, to the tolerances that YOU are capable of
checking, isn't that enough?

--- In CAD_CAM_EDM_DRO@y..., "Ian Wright" <Ian@i...> wrote:
> Hi,
>
>This is something I have been fighting with for the last couple of
>days. How DO you check the accuracy of either cnc or dro systems to
>the accuracies we seem to be capable of?
<snip>
>So, how do you work out the accuracy of a screw thread well enough
>to be able to make a good correction table? And how can you be sure
>your home-made dro is giving an accurate reading?
> Ian