Re: [CAD_CAM_EDM_DRO] Re: Mill retro-fit info available...

scottleatham wrote:

> Accuracy... Well, for the 2/100,000ths, I used a friends Starrett
> Bench Micrometer accurate to 5/1,000,000ths ($800 for an indicator is
> a little rich for my blood). I then cut some T-6061 Aluminium on
> either side with an small end-mill (about 3 inches apart), and
> measured the distance between the cuts on either side. Did this a
> few times and took an average. Then did it again with the same
> already cut part to check for repeatability. Just the surface
> distorion was a factor that had to aceraged as well. I had to play
> with backlash settings quite a bit to get it repeatable, but in the
> end I nailed it. So the 2/100,000ths accuracy is over a range of 3
> inches. I noticed that accuracy sightly degraded when repeative cuts
> were made further apart (around 5 inches or more, to about
> 4/100,000ths). Although the servo drives would start to compensate
> for progressive error over time and improve the accuracy. Feedrate
> between moves was approx. 100ipm.

Well, cutting materials, you are adding tool deflection, and it makes things

a lot harder. The approved technique is a ball-bar tester. I think I could

make one of these, although it would be quite a complex and precise
gadget to build. You make the CNC machine orbit in a circular move, and
while one end causes a shaft encoder to deliver rotation around the circle
to the device reading the move, a linear motion detector on the bar reads
deviation from the length of the bar. You then plot this (with the length
expanded) on a theta-radius plot, and you can see axis jitter, sluggish
axis performance, backlash, non-orthogonality, etc.

>
> For the 500IPM, the machine is rated to 472IPM, to verify this I used
> a yard stick and stop watch with the switchs remotely wired at each
> end. I doubted I'll get 2/100,000ths accuracy when it's flying
> around at 500ipm, I have not performed that test yet.
>
> Ironically, I've got a hobby machine (a Shop Task) that I converted
> to CNC with stepper motors and Ah-Ha system to drive it. It was
> pretty accurate for a small machine (2/10,000ths). However after
> adding ball-screws so it would run faster... I could only get
> accuracy to about 4/500ths repeatable accuracy! So I've got some
> work to do on that machine!

Well, unless you use precision ground ballscrews, that is not a surprise.
I have high precision ballscrews in my Bridgeport, thay look good to better
than .001" over the entire length. These were a lucky find, I'll never find

any more like them.

Jon