[sdiy] Spare op-amp useful in gain stage?

[Richie Burnett]

Putting multiple amplifiers in parallel lowers the noise level.  You get a 
3dB increase in SNR for two amplifiers, and a 6dB increase in SNR for four 
amplifiers.  The same signal is being processed by all amplifiers so the 
multiple coherent outputs add together arithmetically at the point where the 
amplifier outputs are combined.  But the noise generated in each amplifier 
is "orthogonal" to that generated in the others so the separate noise 
sources add vectorially.  You soon reach the point of diminishing returns 
though, for example 4 amplifiers in parallel will give you a 6dB increase in 
SNR, but to get a further 6dB increase in SNR you need to go to 16 
amplifiers!

Cascading two lower gain amplifiers to replace a single high-gain stage 
allows a much wider bandwidth to be achieved if this is a design goal, but 
this wider bandwidth will then let through more noise.  It really depends 
what parameter you are trying to *improve* about your amplifier by throwing 
more op-amps at it.

-Richie,


> Hi!
>
> I was in a discussion about what to do with a spare OP-amp. These
> fellows come in duals and quads, and sometimes the quantities don't
> end up perfectly and you stand there with an extra unused op-amp.
>
> Is there any clever way of using two op-amps instead of one in, say, a
> noisy high-gain stage in order to reach lower noise or something like
> that?
>
> Douglas Self's small signal book suggests paralleling two gain stages
> ("multipath amplifiers"), having their outputs tied together with 10
> ohms to the destination, winning 3 dB SNR. Are there other clever ways
> of using a spare op-amp, particularly in a stage with high gain?