Re: [CAD_CAM_EDM_DRO] PC ports, Encoder resolution and EDM PSU info

Dennis Fielder wrote:

> Three queries which I'm sure someone will know about...
>
> 1) Has anyone considered using the EIDE interfaces on the PC boards to
> interface out to the real
> world? Seems it's a very simple interface to use, with 8 off 8/16 bit
> read writable addresses?
> There's usually at least one spare on a low end machine, and it's easy
> to get 2 extra ports on an
> add-in card.

I have an IDE pc-card reader for getting images off a digital camera.
It
is connected to the 2nd IDE port, the hard drive is on the 1st IDE port.

I pulled the memory card out once while the system was powered on, and
it trashed my hard drive. I had to re-install windows to repair the
damage!
This sounded pretty scary for using devices outside the box, or
software/
hardware development.

> 3) re encoders on lathes to enable CNC threading, I see the
> Microkinetics units are 200 or 400 per
> rev, with pseudo-800 using all edges. Say with 1 thou periphery
> movement then 800 only allows
> 0.8/pi, or only a 1/4 inch diameter item. Seems very small to me,
> where have I gone wrong? I've
> seen encoders with 5000 steps, which seems more likely to be needed.

Huh? I don't follow your logic at all. First, part diameter has
absolutely
nothing to do with threading. The machine makes exactly the same move,
whether it is threading a 1/4" part, or a 4" part. Are you worried
about
surface finish? Maybe the surface finish would get worse as the
diameter
increased on a stepper system, but it probably has more to do with
stepper resolution on the Z axis than the encoder resolution on the
spindle!
OK, let's do a calculation. Since you mention inches above, I'll use
those units, even though your sig indicates a UK location.

Let's say we want a thread of 8 TPI (it seems a coarse thread would
show more 'stairsteps' than a fine one). Now, with a spindle encoder
of 200 cycles/rev, that gives 800 (real, not pseudo) counts/revolution.
We need to advance 1/8 of an inch for a count of 800 on the spindle.
If the Z axis leadscrew was an 8 TPI (common on a manual Imperial
lathe), that would be one full rotation. For a stepper system with 200
steps/rev, directly driving the screw, you need 200 steps for every 800
on the spindle. So, for every 4 encoder counts, you get one Z axis
step. Now, this system has a Z axis step resolution of 8 x 200 = 1600,
or .000625", which is moderate. For smoother results, you could
put a 2:1 or 4:1 belt reduction between the stepper and the leadscrew.

What I'm trying to point out here is that the resolution of the Z axis
(stepper or servo/encoder) will be dominant most of the time, not the
spindle encoder. Second, the update rate of the CNC system may
become dominant at high speeds. Lets say you are turning small
parts at 1800 RPM, or 30 Rev/sec. With a 200 cycle/rev spindle
encoder, that is 30 x 800 counts/sec, or 24,000 counts/sec.
With a CNC update rate of 1 KHz, it is finding that the spindle
has advanced 24 counts every time it looks! OK, nobody threads
at 1800 RPM, of course. But, how slow do you have to go before
the CNC is ready to see each Z axis stepper pulse? Well, in the
original
8 TPI example, the Z axis stepped once every 4 encoder counts.
At a 1 KHz rate again, that would need a spindle encoder at 4000 counst
per sec, or 1000 cycles of the encoder disc/sec. That comes to
1000/200,
or 5 Rev/sec. which is only 300 RPM. IF you go faster than this, a
1 KHz update rate can't send the step pulses smoothly, and will
have to bunch them.

Finally, of course, the machine is not going to be able to follow the
steps precisely. It will end up smoothing the motions out due to
inertia,
stick-slip friction, flexing of mechanical parts, etc.

And, if it is a servo system, the ability of the servo motor to continue
at
constant velocity between servo updates from the CNC control will make
it all smooth no matter what the encoder resolution.

Jon