[sdiy] tri-sin shaper and THD

[harrybissell]

Woah. Nice work !!!

I tried spice simulation with and without feedback.  My proggy
said the non-feedback case was about 3.5% THD and the feedback
case was about 1.05%.

That seems to coincide well with your data.

I think that your feedback method makes the overall shape better, but
does not affect the 'cusp' at the waveform peak.

Theer was some discussion in Electronotes (there it is again) about sine
conversion that was using an OTA for most of the conversion, then a
clipper to flatten the cusp.

Even using trimmers for gain and offset, the sine purity (without feedback)
leaves something to be desired. There are two 'low distortion' points. One
is
the shape that you showed in your paper, the other is an obviously
'square-ish'
wave, that does not have the cusps.  I don't like either.

HAVE to try adding feedback, it looks well worth the small extra effort

H^) harry

ryan williams wrote:

> that discussion of tri-sin converters and the mention of using feedback
> around the OTA caused me to write this note to show myself how it works.
> I posted it the other day, but I added some THD calculations to it and
> also another version with negative feedback.
>
> http://www.sdiy.org/destrukto/notes/tri-sin_converters_using_otas.pdf
>
> for anyone that knows about THD calculation. take a look at it and see
> if that is corrent (pg.4). Or if thats to much work, just tell me if
> this sounds right: I took the sum of the FFT of the converted sine and
> divided by the FFT of a pure sine then subtract 1. expressed in percent.
>
> I ended up with 1.87% distortion with feedback and 3.52% without. but
> the one with feedback looks much better (the sharp corner is almost
> completely gone). Those were the best numbers I could get.
>
> -ryan