I think I may be suffering from NCUD (Noise Calculation Understanding Disorder).
When calculating the noise due to input current and source resistance, why do we not use input bias current Ib but only input noise current In which is much less?
Surely the Ib goes thru the Rs and creates a voltage noise...? In an inverted config the feedback Rf is in parallel with Rs but not so in a noninv config.
For instance many opamps have In less than 1pA but Ib up to 1uA. Why ignore Ib for noise calculation?
Thanks for enlighten...
When calculating the noise due to input current and source resistance, why do we not use input bias current Ib but only input noise current In which is much less?
Surely the Ib goes thru the Rs and creates a voltage noise...? In an inverted config the feedback Rf is in parallel with Rs but not so in a noninv config.
For instance many opamps have In less than 1pA but Ib up to 1uA. Why ignore Ib for noise calculation?
Thanks for enlighten...
If Ib and Rs are both perfectly constant, Ib flowing through Rs only causes a constant DC offset rather than noise. Random variations of Ib and Rs can cause noise, but those are covered by the input noise current spec and the resistor 1/f noise spec, respectively.
Aha that makes sense. I didnt think Ib was without variation. If Ib is perfectly constant then I see how it is. Thanks.
In reality Ib is not perfectly constant, but the random variations are described by In - which is the part that matters for noise.
For example, suppose your amplifier input is the base of a bipolar transistor with a DC base current of 1 uA. The average input bias current is then 1 uA. However, currents through semiconductor junctions are affected by a phenomenon called shot noise that causes random variations. Those variations have a spectral density of In = sqrt(2 q Ib) ~= 0.566 pA/sqrt(Hz) where q is the elementary charge (1.6022E-19 C).
Measured over a bandwidth of, for example, 20 kHz, that's about 80 pA RMS of random variations on top of an average current of 1 uA. The 1 uA of average current just causes a DC offset, the 80 pA of random variations on top of it cause noise.
For example, suppose your amplifier input is the base of a bipolar transistor with a DC base current of 1 uA. The average input bias current is then 1 uA. However, currents through semiconductor junctions are affected by a phenomenon called shot noise that causes random variations. Those variations have a spectral density of In = sqrt(2 q Ib) ~= 0.566 pA/sqrt(Hz) where q is the elementary charge (1.6022E-19 C).
Measured over a bandwidth of, for example, 20 kHz, that's about 80 pA RMS of random variations on top of an average current of 1 uA. The 1 uA of average current just causes a DC offset, the 80 pA of random variations on top of it cause noise.
Thanks. In is the bubbly part of Ib.
I may be wrong but this is not so well described in literature as thermal noise seems to be. Most likely I have simply started drifting as I get to the relevant paragraphs describing input current noise so I guess I should reread those parts.
I may be wrong but this is not so well described in literature as thermal noise seems to be. Most likely I have simply started drifting as I get to the relevant paragraphs describing input current noise so I guess I should reread those parts.
Yes, that's essentially it.
What I wrote is actually slightly inaccurate, because In also includes some other terms - terms that are negligible in most practical cases and that require long calculations involving chain parameters. If the literature you read includes those, I can imagine that the key message got lost.
What I wrote is actually slightly inaccurate, because In also includes some other terms - terms that are negligible in most practical cases and that require long calculations involving chain parameters. If the literature you read includes those, I can imagine that the key message got lost.
Whether shot-noise is generated depends on the details of the current path and the charges flowing - sometimes concerted current flow dominates, where the forces between moving charges are able to smooth the variations out (no/little shot noise), sometimes the current acts more like statistically independent charges - ie exhibits shot noise.
In metals the charges are very close together and tend not to act as independently as in a semiconductor, so metals don't show much if any shot-noise.
In metals the charges are very close together and tend not to act as independently as in a semiconductor, so metals don't show much if any shot-noise.