Re: Rotary to linear translation AHHA! (plus an idea for backlash free gearing)

"David A. Forsyth" wrote:

>
> because the tape is attached to the table and must move at the same
> speed, and when you deflect the tape with the wheel it now
> travels around a radius, and that circumferential distance is
> larger than the straight line distance, the tape must travel around
> the radius faster in order to maintain it's length since the tape is
> not stretchy (to that degree anyway).

OK, this makes more sense to me that stretching/compressing. It can be
drawn out, and calculated. Effects are large enough to be obvious.
An airfoil would make a poor analogy, air passing over either side of
the wing reaches the trailing edge at the same time. Air over the top
has to go further, thus has to travel faster.

Now, I would like to propose a backlash free gearing system that should
offer more than enough resolution for anyone's needs.

Make your pressure roller, gear, pulley, or whatever you are using to
generate rotary motion fairly decent sized. This will roll on small ball
bearings. You will need a reasonably hard roller attached to this shaft,
somewhat smaller than whatever is beign driven. Now, you have a hard
shaft, much smaller in diameter rolling against this drum, and being
pressured into the drum by two small ball bearings. Picure a planetary
setup with one of the three rollers vastly larger than the other two...
Your encoder attaches to the center (driven) shaft.
It should be easy to generate most any reasonable pulse count with this
setup, the main concern being that the driven drum and the hardened drum
are concentric.

Using the opposed disks rolling against a hardened rod, everything could
be totally enclosed in a housing, the shaft sealed with standard shaft
seals, and you have a reliable, backlash free, accurate position sensor.
Heck, I think I'd try to patent the idea if I hadn't just made a public
disclosure.

Jon