To get real random noise, try using a transistor "backwards", as a zener diode. Then look at the voltage- it's quite "noisy", esp. if you use a decent-sized series resistor (try 100Kohms).
When I was in graduate school, a colleague built a gadget based on this principle as a source of Poisson-distributed pulses for testing the post-detector electronics of X-ray and extreme-ultraviolet astronomical instruments. I don't remember the precise source of the noise, but it was similar to the "zener" trick cited in that it was based on quantum phenomena rather than a mathematical generator. The idea was to get random pulses of varying amplitudes, amplify them up and use only those bigger than a threshold to generate output pulses. The instrument also averaged the output rate and used it via feedback to adjust the amplifier, so as to obtain a desired average rate of outputs. I think that for a while at least there was a commercial random-noise generator available that used this principle, though I don't remember whose (and it's been long enough that it probably doesn't matter). Try scientific-instrument catalogs, et cetera. -- Jay Freeman