[sdiy] improved dynamic range diffpair-vca

[Magnus Danielson]

From: "JH." <jhaible at debitel.net>
Subject: Re: [sdiy] improved dynamic range diffpair-vca
Date: Wed, 25 Apr 2007 02:06:53 +0200
Message-ID: <003201c786cd$a295ea60$0200a8c0 at jhsilent>

> >Yes, but considering the configuration the distorsion will be the killer
> >anyway
> 
> You set an amount of distortion that you consider tolerable.
> This results in a certain signal level on an ordinard differential pair.
> That level, divided by the noise of the pair, is the SNR.
> 
> With the proposed circuit, and the same amount of distortion, you can have 
> twice the signal level.
> And you'll get SQRT(2) times the noise.
> Resulting in 2 / SQRT(2) = SQRT(2), or 3dB,  SNR improovement.
> 
> If this is true, it's true for any amount of distortion you choose.

It really differs what you define as noise in this case. If you only consider
white noise, pink noise (i.e. random sources) etc. then you are certainly
correct. Another form is "anything but". This includes hum and its overtones,
harmonic and non-harmonic distorsion etc. In this form noise is all unwanted
signal. This form is usefull for actual dynamics.

Also, you are assuming that the transistors are the main source of noise.
They don't need to be.

So it all depends. Saying it is 6 dB SNR improvement is a sloopy way of saying
it, but for the same drive-level will the distorsion be another 6 dB down.
That is certainly usefull. The noise floor moves up by 3 dB thought, but that
may be a much less problem (and I think it is).

If you want a really high-dynamics VCA the Drawmer 4-transistor setup should be
a win. I beleive that the THAT VCAs still use that configuration, but with the
refinements that comes with sitting on the same silicon etc. Haven't measured
the THD of that one thought.

Cheers,
Magnus