[sdiy] mixer

[jh.]

> Yes, that's what I was saying - it always adds.  It's best to do without
it if
> possible. If you're using low input resistances anyway, it would take a
pretty
> big capacitor to filter the noise.
> Also, if you just _have_ to use large input resistances (like with a high
source
> output impedence) you're better off using a FET input opamp even though
they
> have higher voltage noise, rather than a bipolar type.
> A long time ago in my engineering days I wrote an article on opamp noise
> calculations that got published in a British electronics engineering
magazine.
> If I can find the silly thing I can make it available.   National
Semiconductor
> also has an application note AN-104 "Noise Specs Confusing?".

I think I've seen this one, and you're perfectly right about FETs (with
low noise current) being better from a certain source impedance upward.
I think if you have low source impedance and if you can live with
some offset drift, better use a single transistor input stage, and not
a differential amp anyway: The noise across the two transistors
adds geometrically, so the single transistor input should be 3dB
better. In the early days, when amplifier noise was more of a problem
than today, there were chips which could either be used with
differential input or with single transitor input. Small power
amps from the LM38x series, if memory serves.

JH.



>
> "jh." wrote:
>
> > > Also, tricks used for error cancellation in precision circuits, like
> > having a
> > > resistor in series with the + input that is the same as the equivalent
> > resistance
> > > seen by the - input (eg, Rf || Rin) is actually detrimental, because
there
> > is no
> > > phase cancellation of noise between the + and - inputs - everything
adds
> > as RMS
> > > (the square root of the sum of the squares of all the individual noise
> > sources).
> >
> > This "trick" is for reducing offset voltage comming from input bias
current,
> > not for noise reduction. I think it even *adds* noise, so better bypass
it
> > with a capacitor.
> >
> > JH.
>
> --
> Scott Bernardi
> sbernardi at home.net
>
>