Re: [CAD_CAM_EDM_DRO] digitizer debouncing

Hi Carlos,

My acceleration/deceleration is based on the difference between the
starting (jump) feed speed, and the desired speed. It's not that the
probe routine WANTS overtravel, but you have to follow a profile to get
the steppers up to speed.

.25 might not be too bad, and I guess just go slow in Z. This code is
still quite preliminary, and I'm moving about .5" just to try it out.
I've never seen one run! This is just a little Sherline, so I don't
have much speed anyway. I think my decel is about .10 now. (it matches
my accel curve).

For the "work envelope" to hold the part, there may be considerable
movement needed to get to the part. If we already have a "rough"
contour to follow, we might be able to start closer. Haven't gotten
nearly that far yet. I'm just using a Gcode program to give my
"microswitch probe" (1 axis, 'X' only) some exercise!

Thanks for the numbers, I'll see what I can do. .002, that's fast!

Alan


Carlos Guillermo wrote:

>
> Alan -
>
> The Renishaw model I've laid out has about 1/4" overtravel in X-Y
> (depends on probe length), and about 1/16" in Z. I can't imagine
> the probing routine would want to overtravel any more than a
> couple thou., however, especially at those low speeds. Any more
> overtravel, and it would have to backtrack after every move in
> order to maintain high resolution. The accel/deccel should be set
> to a fairly high value, IMO, in order to finish the scan sometime
> this century. This is where the probe needs to be very rigid and
> light, and the overtravel/switch contact spring needs to be fairly
> stiff. Any (appreciable) oscillation at the probe tip would cause
> a different kind of "bounce" we'd have to debounce. Can you see
> what happens if you deccel within .001-.002" in your setup?
>
> Carlos Guillermo
> VERVE Engineering & Design
>
> -----Original Message-----
> From: Alan Marconett KM6VV [mailto:KM6VV@...]
> Sent: Tuesday, January 16, 2001 2:37 PM
> To: CAD_CAM_EDM_DRO@egroups.com
> Subject: [CAD_CAM_EDM_DRO] Re: digitizer debouncing
>
> Roman,
>
> I have to disagree here. 555 timers are lousy anyway, there are
> better
> "timer" parts. What may be needed is a schmitt trigger, something
> like
> the old 74LS14 I've mentioned before. No caps across the
> contacts. Run
> 5-20 ma. (1K pull-up) through the contacts.
>
> The real "debounce" is in the software! If you sample the line
> periodically (between steps), and take action on the 1ST one (and
> only
> the 1ST one), you have all the debounce you need.
>
> Did it this last weekend. No GATES/TIMERS! Last night I got the
> position latched, and decelerated afterwards to a stop. Did this
> repeatedly, with good results. Which raises a question. How much
> OVER
> TRAVEL do we have with the Renishaw "model" we are building? My
> little
> micro switch got a little "up tight" with a longer deceleration!
>
> Alan KM6VV