[sdiy] BBD's in series and accumulative S/N ratio degradation.

[Tom Wiltshire]

On 23 Dec 2011, at 13:30, Mattias Rickardsson wrote:

> On 23 December 2011 13:36, Tom Wiltshire <tom at electricdruid.net> wrote:
>> you're supposed to filter the signal between stages to remove clock noise. The clock noise will (in theory) cause aliasing in the next BBD
> 
> Oh, is that so even if you use the same clock?

That's why I said "in theory". In theory you're not supposed to stick anything into a sampling system above nyquist, as you know - and the BBD clock is clearly breaking that rule.

In principle, you ought to be able to get away with it, since the clock of one and the next BBD should be identical.
In practice, it doesn't work that well.

If the edges of the signals didn't get degraded, and ignoring the other 1/f noise Magnus mentioned, you should be able to ignore even fairly large offsets caused by clock leakage until the end of the chain. But in reality, this is a bad idea. You might get away with it for a pair of BBDs, but I wouldn't try it with three, and four long BBDs is a definite fail. I tried it. I'd have avoided sticking three extra filters in my schematic if I could have, believe me. In my view, less is usually more. If you check schematics for long analog delay lines (for example; Yamaha E1010), you'll see inter-stage filters. In the end I reached the same conclusion they had, and put them in.

Massive filtering isn't over-the-top in this case because (as may have been mentioned here before) BBDs suck! You need all the help you can get to reduce noise to some level less than atrocious! My experience has been that each BBD in the chain needs to see the cleanest analog signal it possibly can in order to have a reasonable chance of seeing something related to the input coming out of the output - that means filtering the signal to remove clock spikes before you send that to the next BBD.

Tom