Thermal noise in bipolars

[diarmuid]

Hello All,

I am trying to make a noise comparison for MOSFET vs bipolar.

- I know that bipolars exhibit lower flicker noise.

But what about their thermal noise?

They have higher gms so does this translate to higher thermal noise?

Thanks,

Diarmuid

 

[erikl]

If you want to compare, you should use amplifying stages with same gms. Otherwise this wouldn't be fair, I think, would it?

 

[diarmuid]

Good point.

So if we assume we are making a noise comparison of a mosfet vs bipolar (with same gm's), which would exhibit more thermal noise?

Strange that I cant find much data on the web regarding this question.

 

[erikl]

So if we assume we are making a noise comparison of a mosfet vs bipolar (with same gm's), which would exhibit more thermal noise?

The mosfet, I'd think, because it probably needs more current for the same gm.

Strange that I cant find much data on the web regarding this question.

Nor do I know such comparison. I'll leave the literature research to you ;-)

 

[diarmuid]

Yea, with increased current, noise would surely go up.

Someone should do a phd on all this stuff! Im finding serious gaps in noise analysis!

 

[WimRFP]

The BJT has thermal noise in its bulk resistances only. 1/gm appears as a voltage noise source in the base [ rt(4KTB*0.5*re), re = 1/gm ]. This noise source is in series with the noise generated by Rbb' and Ree' rt(4kTB(Rbb'+Ree')).

A FET has a temperature dependent drain noise current component that has to be transformed to the gate.

I don't know your frequency of interest, but at low frequency (especially for the CMOS), 1/f, 1/f^2, etc noise dominates the pure thermal (Johnson) noise contributions.

Whether the overall noise increases with increasing drain or collector current, depends also on the source impedance. I like the BJT because I can tweek the collector current to find a good trade-off between overall input current and voltage noise.

 

[erikl]

Someone should do a phd on all this stuff! Im finding serious gaps in noise analysis!

Sure; probably already done a hundred times - just find them ;-). Perhaps this Willy Sansen book is a good start?