> yep...think this is what I was looking for...thanks!!
>
> "M. Adam Davis" <EraseMEstienmanspam_OUTTakeThisOuTgmail.com> wrote:  I have some amount of rotary encoder info here:
> http://www.ubasics.com/adam/electronics/doc/rotryenc.shtml
>
> It covers very simple decoding (treat A as direction, B as step), then
> gives an overview of typical state machine decoding. It also covers
> the different types of encoders, and quotes a few people talking about
> some interesting absolute encoders.
>
> It's not likely the same discussion you remember but perhaps it has
> the information you need.
>
> And if not - ask away! The piclist can never have too many
> discussions on rotary encoders!
>
> -Adam
>
> On 9/27/07, alan smith wrote:
> > I know I have it someplace...stored under something obvious at the time....but there was some posts dealing with rotary encoder interfaces, and now of course I need to recall what was said....
> >
> > anyone remember?
> >
> >
> > ---------------------------------
> > Yahoo! oneSearch: Finally, mobile search that gives answers, not web links.
> > --

>
> If implemented straight like that it have a problem that is severe if you
> need to keep exact track of position:
>
> For example:
> In case of small movement/viration/noise inpulse, such that B channel
> sensor repeatedly go on and off but movemet is to little for A to change,
> Then your counter wanders off from reality!
>
> Unfortumately this bad examplke it taught in many schoolbooks :(
>
> The right way of doing this is to decode all changes in both channels.
>
> SAMPLING:
> One way of sampling is to use PortB interrupt on change, but if you
> already have a timer interrupt faster than any predicted movement you can
> hang the sampling on that to save on tie spent in context save/restore.
>
> DECODING EXAMPLE:
> Concatenate the status from the channelas from previous sample with the
> last sample into 4 bit.
> Use that in a jump table to jump to count:  no change (exit) / forward /
> backward / illegal (both channels changed)
>