That's awfully complicated for a hardware RNG. A channel-hopping receiver? The classic solution is a radiation source, which quantum mechanics says is random.
"The chi-square test is the most commonly used test for the randomness of data, and is extremely sensitive to errors in pseudorandom sequence generators. The chi-square distribution is calculated for the stream of bytes in the file and expressed as an absolute number and a percentage which indicates how frequently a truly random sequence would exceed the value calculated. We interpret the percentage as the degree to which the sequence tested is suspected of being non-random. If the percentage is greater than 99% or less than 1%, the sequence is almost certainly not random. If the percentage is between 99% and 95% or between 1% and 5%, the sequence is suspect. Percentages between 90% and 95% and 5% and 10% indicate the sequence is “almost suspect”. Note that our JPEG file, while very dense in information, is far from random as revealed by the chi-square test.
Applying this test to the output of various pseudorandom sequence generators is interesting. The low-order 8 bits returned by the standard Unix rand() function, for example, yields:
Chi square distribution for 500000 samples is 0.01, and randomly would exceed this value more than 99.99 percent of the times.
While an improved generator [Park & Miller] reports:
Chi square distribution for 500000 samples is 212.53, and randomly would exceed this value 97.53 percent of the times.
Thus, the standard Unix generator (or at least the low-order bytes it returns) is unacceptably non-random, while the improved generator is much better but still sufficiently non-random to cause concern for demanding applications. Contrast both of these software generators with the chi-square result of a genuine random sequence created by timing radioactive decay events.
Chi square distribution for 500000 samples is 249.51, and randomly would exceed this value 40.98 percent of the times. "
Maybe it is more complicated as a concept, but the implementation is much more simple. They're using a MCU with integrated radio, it's almost free to add the RF source to the base system only using the avalanche diode.
A HV geiger tube and embedding a radiation source would be much more complicated / expensive to build, especially on small scale.
The channel hoping is just a bit of software. It costs $0.00 in additional hardware. Software complexity is a big deal in a TRNG. The KISS rule applies big-time here. The more complex the software, the more likely it is a significant flaw will be found and exploited. It remains unclear to me whether there should even be a microcontroller on a USB TRNG. Without one, we can eliminate the complexity from the USB key itself, but we have to move the whitening and health monitoring to the driver. It is unclear to me if making the driver more complex opens it up to enough additional attacks that the system is less secure overall.
http://www.fourmilab.ch/hotbits/hardware3.html