AW: [sdiy] Random Voltage Generation

[Czech Martin]

I've read about using a low excess noise resistor in
order to get thermal noise. I know a fellow that build such a
thing for his thesis, to get better random numbers.
Of course you need high gain amplification and good shielding.

Let me see, RMS noise voltage of an ideal resistor is (if I'm not
altzheimering here):

U=SQRT{k*T*R*B}

B=20kHz (1/s)
k=1.38E-23 J/K (V*A*s/K)
R=10 Meg Ohm (V/A)
T=300K

U=28uV, the peak level is about 3 times higher 86uV.
Putting 10 R in series, we will get only about 3 times
higher...

Perhaps this can be done, exploiting the very low input noise
currents of JFET or MOSFET amps.

I wonder how white this noise may be, and how much 1/f
there is.

Has someone made experiences what this means in music
applications other than the audio band (trigger source etc.).

m.c.