On Sat, 31 Mar 2007 15:39:19 +0100, "john jardine"
wrote:


"John Larkin" wrote in message
.. .
[...]
Seems about 30% THD?.

Is that what you get with 16 amplitude levels spaced, presumably, at
16 even time intervals? I'd guess somewhat better than 30%; a square
wave is only 43. Possibly you could make a better sine wave if the
resistors weren't binary weighted. More math than I care to sign up
for!

I'd envisioned a clean triangle. (3rd harmonic at 1/3rd of primary).

We just did a homebrew 8-channel DDS waveform generator, with 16-bit
math and 14-bit dacs, 2048 point sine lookup table. Sinewaves look
pretty good.

John


Nice!.
Sounds like you've been disenchanted with the performance of the AD chips.

We wanted the flexibility to do other/arbitrary waveforms, pulses,
modulation, and channel-channel phase shifts on later versions. And to
program amplitude. And it was fun to roll our own DDS.

John