[Joe Johnson]

The real test would be to truncate all your calculations to 16 bits and then to count the number of interations needed to get to within 1 bit of byte result (1 out of 256).

We only have 8 bits of resolution in the inputs. We have LESS than 8 bits of resolution in the outputs (I don't recall the actual numbers but I THINK it is something like 64 PWM steps in each direction -- not more). Finally, the sin and cos functions we need in our newton's method are only 8 bit functions.

There may be some case when more resolution can be utilized but I are skeptical.

If you got the answer to 8 significant bits, I suppose that would be plenty.

Joe J.