... I would be careful with those strong statements on randomness such as "passed FIPS 140-2 test", unless the exact working of those clocks under various circumstances are very well known.

At least the "voting of 5 results" makes the whole article suspicious: a biased "coin toss" produces a similar (if not exactly the same - I am lazy now to think about it more) bias than the coin toss itself.

One very often used source of physical randomness is the LSB of an ADC; yet many fail to recognize the exact conditions when this is usable, just one example, an adversary able to access the ADC's reference pin may quite easily induce some bias to the "randomness".

JW