[sdiy] BBD clock bleed

rsdio at audiobanshee.com rsdio at audiobanshee.com
Thu Oct 25 04:31:28 CEST 2018


On Oct 24, 2018, at 10:31 AM, Ingo Debus <igg.debus at gmail.com> wrote:
>> Am 24.10.2018 um 19:22 schrieb Uwe Schueler <uwe.schueler at uni-tuebingen.de>:
>>> However, the signal-to-clock ratio should improve by 6 dB.
>> here I don't agree: clock feed through that is (almost) identical on both channels should cancel out (almost) completely.
> 
> Yes, that I was thinking too. If the two BBD chips are 100% identical, clock feedthrough should be cancelled completely. Sort of like a humbucker pickup works.

Two BBD chips won’t be identical, because they’re way more complicated than a pair of coiled wires. The thousands of transistors in one BBD will vary quite a bit from the thousands in the other BBD.

> But how much of the clock noise is feedthrough and how much is aliasing? I don’t think you can reduce aliasing that way.
> 
> Ingo

There is no way to reduce aliasing after it has occurred. The only way to reduce aliasing is to prevent it in the first place by properly low-pass filtering the analog signal *before* it hits the BBD inputs. This anti-aliasing filter won’t help clock noise at all, but it will reduce or prevent aliasing, depending upon how well the filter removes frequencies above Nyquist.

A (second) reconstruction filter after the BBD outputs will smooth the signal, remove images, and reduce clock noise.

In a modular system, these filters could be separate from the BBD module, but then you have to patch a lot of pieces together to get a smooth, analog delay.

Brian Willoughby




More information about the Synth-diy mailing list