MN3005, MN3007 BBD.. anyone built one?

[Jeremy Brookes]

>> I was puzzling over what the NE570N
>> compander was doing there. Is it normal
>> to compress the signal before it
>> goes through the delay stage?
>
>Compression at input followed by expansion at output is a noise
>reduction technique.  Here's my non-technical understanding of this
>process: Basically you compress the signal before it enters a device
>which introduces noise (like a delay line), then expand the signal after
>it leaves the device.  This (ideally) causes no change in the desired
>signal, but the noise which was introduced while the signal was
>compressed will become proportionally quieter to the desired signal
>after expansion.


Hmmm... I remember hearing something slightly different though it was
applied to digital delay and hence sampling.

What the compressing does is effectively give you a non-linear quantisation
interval so you get much finer resolution for low level signals round about
the zero volts point, with much bigger steps at the peaks of big signals.
Expansion on the output reverses the effect. What I think this means is that
low level signal suffer from less quantisation noise at the expense of the
noise being present in a high level signal. But then the high level signal
would mask the noise anyway. I think I remember this from an article for
building an 8-bit sampler with 12-bit apparent resolution to interface to a
Sinclair ZX Spectrum! (Wasn't that a Timex T1000 or something in the US?)