Yet another BBD question...

[KA4HJH]

>Thanks for the description (and to the folk who e-mailed me directly!)  What
>does "limited range" mean?  That sounds simple, but do you mean that you
>can't achieve certain frequencies?  If so, why?

Because a BBD is an analog "sampler", it has many of the same 
problems that digital sampling has--aliasing, limited bandwidth, etc. 
The delay time is varied by changing the clock rate. The maximum 
delay time is constrained by how much bandwidth you want (just like a 
digital delay). Slow it down enough and you start to hear aliasing, 
even slower and you'll hear the clock itself as it comes into the 
audio range.

But the minimum delay time is limited, too. It takes time for the 
electrons to flow from one cap to another. Clock it too fast and the 
dynamic range starts to suffer. And every stage adds a little more 
error and a little more noise--the more stages the crummier the 
signal will be at the output.

Even so you can get some large time ratios with analog delays. The 
shorter Panasonic delays are spec'ed at 20:1, which is only a 
ballpark figure and depends on the application--and what you can put 
up with. The old Reticon delays could be clocked even faster, 
although they were noisier.


>Something that seems strange to me... wouldn't you need two lines, really?
>I mean, can a cap discharge and charge at the same time?  It seems to me
>that you would need to alternate two rows, one is charging, the other
>discharging, and your output is always from the discharging row.

You've obviously been thinking about this--a BBD is actually two 
alternating delay lines in parallel.

Which bring up the idea of put two *complete* delay lines in 
parallel, in differential mode. This cancels out some of the crap and 
doubles the dynamic range.

Another trick is to have one delay fixed to a short delay time while 
varying the other. This makes it possible to flange (comb filter) all 
the way out of the AF range (as the variable delay time approaches 
the fixed time), sidestepping the minimum delay time problem 
completely--at the expense of introducing a short, fixed time delay 
into your "straight" signal, as well as making it noisier. I'd like 
to try this myself.


>Are there other ways to do analog delays?

Reverb springs are analog delays, albeit with a lot of resonance.

Harry once suggested two tin cans and a piece of string. I have to 
give credit where it's due.

Terry Bowman, KA4HJH
"The Mac Doctor"

ICQ: 45652354