> what the typical noise level of a power transistor IS??
You can calculate the noise for an ideal transistor.
Silicon since before 1980 has usually come very close to the theoretical noise.
Noise can be analyzed into Voltage and Current. A power transistor has less stray resistance, and thus lower noise voltage at the same current; and it can be run at higher current for even less noise voltage.
However, a power transistor has lower current gain, so will have higher noise current at the same operating current, and still higher noise current if you use it at high current (which is usually why you paid for a power transistor).
Many mike amps use the medium/small Power 2N4401 device, which is sold for relay-switching and other small but tough chores, as their first stage. It is big enough to have stray resistance that is low compared to mike impedance, yet small enough to have high current gain.
As Gyraf says: if the first stage has significant gain, later stages can have quite high noise and it won't matter. Say your first stage has 0.5uV noise (combined voltage noise and the noise current in the source impedance) and a gain of 20. Its output has 10uV of noise. Pick a pretty hissy device for the second stage, say 2uV.
The combined noise "looks like" 10uV+2uV= 12uV. If the second stage were utterly perfect no-noise, it would be 10uV total. The difference is only 1.6dB: if noise is too high, 1.6dB less is not going to make you grin.
Actually the combined noise of 10uV and 2uV is 10.198uV. The hissy second stage only adds 0.2dB more noise than a perfect second stage.
Second (third etc) stage noise is usually irrelevant unless the first stage is very low gain or the second stage is incredibly hissy.
It is possible to screw-up. A common Fender guitar amp has a first tube with gain of 50, a tone stack with loss of 10, a volume pot that may have a loss of 10, and another tube. In this case the second tube dominates the noise (but not when you need to turn Vol all the way up). Fender has another plan with two gain stages before the tone stack, but it costs more and is not generally better for typical playing. Or not for noise figure: the dual stages do let a hot guitar overload the second stage, and then the grunge can be shaped with the tone controls, a useful tool for guitarists.
Nobody ever asks what the noise of a 6L6 output tube is. It would have to be insanely bad to be a problem. The first stage works with 0.020V signals, the 6L6 gets 20V signals, so a 6L6 would have to have 1,000 times the noise of a 12AX7 tube to add any hiss. A healthy 6L6 probably has noise similar to a 12AX7, a very sick 6L6 might have noise 100 times higher which is still negligible. When gitar amp noise does turn out to be power tubes, it is usually dirty sockets, oscillation, or a tube that is about to die, not honest random noise.
A clean 2N3055 is not hissy. A 2N3055 that hisses is probably dirty silicon and will fail. This was an issue in the early 1970s. We had to sort small transistors for noise. But the big application was in computers, where even low failure rates mean lots of downtime in a 10,000 transistor system. So in getting reliable transistors, we got quiet transistors. We still have noise from stray resistance and low gain, so some transistor types are better than others. But we don't have to weed-out the ones that hiss far more than they should.
I have a couple mike amps that use TIP120, a power transistor and a Darlington which is double the noise voltage. Both designs idle at 28mA. One has a small transistor in front, the other amp is just a TIP120. And these TIP120s are from 1978, when silicon cleanliness was not always the best. Both are quite quiet.
