First, let me give the punchline that Alex (likely his designer John) later on have agreed that his hypothesis was wrong. Please see:
http://www.audiosciencereview.com/f...view-and-measurements.1829/page-15#post-46792
Here is the much longer answer.
Switchmode power supplies due to their nature of turning on and off, create tons of electromagnetic interference (EMI). This switching noise starts at Kilohertz frequencies but goes way up to megahertz due to its high power and decomposition of a square wave (odd harmonics forever). In western world, such devices must limit this kind of interference or they are not legal to sell.
What is bad is that the radiofrequency energy actually comes out of the DC wire of the power supply! Even though the switching is before the isolation transformer, there is enough capacitance to let high frequencies to travel across that isolation barrier and as such, make the DC wire as an antenna radiating everywhere.
Turns out there is a very cheap solution to the problem. You put a capacitor from the output of the power supply to its AC input. The impedance of the input is pretty low for high frequencies and as such, it works to substantially reduce the radiated energy.
Now, you don't want that capacitor to ever fail short because if it does, it will connect the AC input to output representative massive risk of shock/death! For this reason, special safety capacitors called "Y" (and "X") capacitors are used here which in failure mode, become open (there is a film in there that burns through). So if there is an incoming surge on the AC line, all that happens is that you lose your EMI reduction above.
Now you know why the "y capacitor" reference has been used time and time again.
Here is the issue with the Y capacitor though. It not only takes the output noise and dumps it in the input, but also works in reverse, allow some (small) amount of AC mains current to travel the other way onto the output of the DC power supply.
There are regulations on how much is allowed before there is a risk of shock. If you ever touch an appliance and you get a "buzz", that is the leakage current that is allowed. That is, it doesn't kill you so it is OK.
Problem we have in audio is that we have "single-ended" equipment, i.e. RCA Jacks. This means that the reference for the signal, i.e. what is "zero," is the chassis of the equipment. This is a bad architecture and the reason people sometimes get hum.
Getting rid of that hum requires that we stop the current flow between the chassis of the two pieces of equipment. Because once it flows, it is indistinguishable from audio signal itself.
In the case of switchmode power supply as used in this context is that it is already anxious and ready to pump out AC current. By design, it is "leaking" current and it is doing so on the ground pin of the DC connector which becomes the ground of the audio circuits.
Wow. Are you with me still?
Now let's get into Alex's hypothesis that there is a switchmode power supply in the Audio Precision Analyzer and hence, it also contributes the same problem. The fundamental problem with that theory is that he doesn't know it is using a switchmode power supply for its operation. I don't either. But this becomes immaterial if you read further.
If you have read this thread fully, you will see that I tested sources powering my USB DAC like my main computer and laptop which too have switchmode power supplies. These supplies showed none of the AC leakage.
The reason they did not is because they have better designed power supplies than the $10 wall wart supplied here. Putting in the Y cap above is a hack, an effective one but still a hack. It allows the power supply to have lots of noise to start, and then tries to filter it. A better but more expensive solution is to use other power supply topologies that have far less noise (e.g. switch when the current is zero). To that end, if the AP has switchmode power supply, it too would be of highest quality by nature and not source of its own leakage.
Now, the theory that Alex/John put forward was even more complex than this. They were attempting to say -- poorly I might add -- that not only there was leakage in the AP but that it was adding to the ones contributed by their supply. Well, this was easy to test for as I have an AC power generator that lets me program its frequency. So I set that to 70 Hz and let the AP run at 60 Hz. That way, the AC leakage contribution would be different from each other. The result resoundingly showed the leakage to be entirely that of 70 Hz and hence, the power supply that came with Iso Regen:
It is true that if we had used balanced system, we would not worry about this. But 99% of the people are not.
If your amplifier/pre-amp is connected to the DAC using RCA jacks, then you would see the exact the same leakage as my Audio Precision Analyzer sees.
Summary
Cheap switchmode power supplies employ filtering that by design creates AC current leakage. In typical unbalanced (RCA) audio systems that we use, we are highly susceptible to such leakage because it establishes the reference for our audio signal. As such, this leakage current travels all the way through the DAC and presents itself on its output.
Solutions are to use linear power supplies, avoid these devices if they present no other value, or much better switchmode power supplies.
None of this points to a flaw in the design of the test fixture that I am using. I am on purpose emulating what happens in real life.
P.S. Welcome to the forum.