Klaus von der Heide wrote:

[snip]
The problem is the discrete time the DDS uses. The high order bit
only changes at the clock edges. The result is a square wave with
heavy phase jitter.  For example, draw a sine wave (3 waves),
sample it at about 8/3 of its frequency in your sketch, and generate
the square wave from these sample values (all on the paper).
The one-periods and the zero-periods then are one or two clock cycles
long. That's what Andy in other words already said.
If a DAC is used then time is continuous and a clean square wave can
be obtained.

Hi Klaus,
                yes, that makes a lot of sense. It seems that, after all,
a DAC is needed... TNX for clarifying.

Time is continous... this was a great debate in the Theoretical Physics field
many, many years ago. But now I have lost contact with that ambient
and don't know what the present theories say. But this is totally off-topic...

73  Alberto   I2PHD