Re: [CAD_CAM_EDM_DRO] Re: gearmotors,encoders

jmw@... wrote:

> Jon, what do you think about the rigidity of a 1.25" screw 36" long--
> moving a 36" x 12" table plus the work. The encoder will have a
> thousand lines (as soon as I figure out how to recode 1968 line
> output to 1000 line).
> maybe only your ME knows for sure.

Yes, I certainly haven't calculated it. But, a 1.25" dia screw will be a
LOT stiffer than the .985" screw in my system.

Why would you want to lower the encoder resolution????? Assuming
a 5 TPI thread, 1000 cycles/rev (x 4 for quadrature decoding) =
5 x 4000 = 20000 counts/inch, or .00005" per count, same as I have
on my system. With the 1968 cycle encoders, you'd get
5 x 1968 x 4 = 39360 which is roughly .000025", or almost twice
the resolution. How are you going to 'recode' the encoder,
anyway? You'd have to change the glass disc in it, as well as
the analyzer gratings. You can't make those in your basement,
or at least it will be very hard!

If you have some idea to electronically interpolate the resolution
down, that would be the worst thing you could do! Because your
resolution would not go down from 1968 to 1000, but from 1968 to
8! So, you would actually only have the counts come out exactly
on the dot 8 times a revolution (I think 8 is the LCF of 1000 and 1968)
or a resolution of 8 x 5 = 40 or .025"
I can assure you no CNC system will work like that.

> Maybe on a sharp accel you'd loose a count ...

No, it wouldn't lose counts, jost be temporarily off by a few. But,
the fact that opposite ends of the leadscrew differ by a few counts
could cause the servo system to oscillate. On a high power servo
system not properly compensated for this setup, you could literally
break the screw in half this way.

Jon