It provides a relatively high level of noise, typically around -105dBm/Hz up to cca 20MHz, and falls off with cca 12dB/octave above that.
The noise level is not calibrated, and will depend on the particular transistors used. It is not stabilized regarding to supply voltage and temperature, but provided these are kept constant, it will be fairly constant.
The first transistor ("upside down") is used as an avalanche noise diode, through B-E diode breakdown. The other two transistors provide some amplification and impedance conversion.
The output can be checked with an oscilloscope. Depending on the brightness setting, there should be at least 10mVpp of "fog", typically it should be around 30mV. (to see low voltages on a scope, connect directly to scope input, don't use 10:1 probe)
The DC working points depend on the BE breakdown voltage of the first transistor, typically around 6V. This can vary a volt up or down, but the circuit should still work.
Just check that the middle transistor can "breathe" - that is its collector is at least a volt below supply, and its emitor is at least a volt below its collector.
It makes sense to put the whole circuit into a small metal box, with a coax connector, and some (feedthrough) supply filtering.
Apart from the testing of the SIDI IF circuits, I guess this noise source could also be useful for testing "Jupiter type" HF receivers.