Wouldn't adding negative feedback increase bandwidth just like emitter degeneration, without necessarily adding noise?
Either way, you get contributions from resistor thermal noise. Games can be played, of course, to optimize your system noise by playing around with resistor values. But there is a way to reduce the effective "temperature" of a resistor by using negative feedback in a way that at first glance almost seems like magic.
It's been a little difficult to find relevant articles on this particular subject via online searches because 99.999% of search "hits" I encountered have nothing to do with this subject, but I was able to find this.
Basically you start with a higher-value resistor that is connected to a signal-source which replicates the signal on the other side of the resistor but is 180 degrees out of phase with it (and is amplified). The additional voltage swing makes the resistor "look" smaller, and, in addition, the feedback reduces the effective noise of the resistor. Depending on your requirements, the improvement could be pretty substantial.
This scheme has some limitations, though. Since you've got an amplifier in the loop, it has bandwidth limitations. As the signal (or noise) frequency goes up, at the some point the amplifier's gain starts to go down -- so the effective "temperature" of the resistor increases. Its effective value as a resistor increases as well, which might result in other problems. The other, more obvious limitation, is that the scheme isn't universally applicable to every resistor in your circuit. It has been used to reduce the effective thermal noise of a 47K ground-terminating resistor in a MM phono preamp but as far as I know, that's about it, in terms of audio applications. Going past that may be possible, but I'm not sure if the added complexity would be worth it. The noise contribution of the first gain stage generally is the main determining factor of the overall system's SNR so why go beyond that?
The other final limitation is the En and In of the amplifier you're using in that negative feedback loop, they never will be zero. So you can't reduce the effective "temperature" of your input resistor below a certain point.
It's been a little difficult to find relevant articles on this particular subject via online searches because 99.999% of search "hits" I encountered have nothing to do with this subject, but I was able to find this.
Basically you start with a higher-value resistor that is connected to a signal-source which replicates the signal on the other side of the resistor but is 180 degrees out of phase with it (and is amplified). The additional voltage swing makes the resistor "look" smaller, and, in addition, the feedback reduces the effective noise of the resistor. Depending on your requirements, the improvement could be pretty substantial.
This scheme has some limitations, though. Since you've got an amplifier in the loop, it has bandwidth limitations. As the signal (or noise) frequency goes up, at the some point the amplifier's gain starts to go down -- so the effective "temperature" of the resistor increases. Its effective value as a resistor increases as well, which might result in other problems. The other, more obvious limitation, is that the scheme isn't universally applicable to every resistor in your circuit. It has been used to reduce the effective thermal noise of a 47K ground-terminating resistor in a MM phono preamp but as far as I know, that's about it, in terms of audio applications. Going past that may be possible, but I'm not sure if the added complexity would be worth it. The noise contribution of the first gain stage generally is the main determining factor of the overall system's SNR so why go beyond that?
The other final limitation is the En and In of the amplifier you're using in that negative feedback loop, they never will be zero. So you can't reduce the effective "temperature" of your input resistor below a certain point.
Emitter degeneration is a type of feedback, like ejp already wrote. I don't understand why he calls it voltage feedback; according to the original 1930's definition it is current feedback, but that term has been redefined at least twice, so no one knows what it means anymore. Anyway, that's just semantics.
To reduce gain and increase bandwidth without adding noise, you theoretically need non-energic feedback. That is, feedback via wires or transformers, not via impedances. (Ideal gyrators could also do the trick, but have the disadvantage that they don't exist.)