Stefan Sczekalla-Waldschmidt wrote:

> Using one line as Clock an the other as direction did not work
> for me. This may work in a enviroment without or with less noise.

  The concept is flawed.  If the "clock" line senses one edge only,
then a "jitter" which would toggle this line back and forth would quite
wrongly produce an apparent continuous rotation in one direction.

  For this to work, the "clock" must be processed on both edges, with
the sense of the "data" being considered opposite in each case.  Even
so, this approach loses half the definition of the encoder.  You
really need to detect *any* change, possibly by cross-clocking hardware
on both inputs.
--
  Cheers,
        Paul B.