It's based upon listening tests, and a thorough re-think as to what is going on.
We have two SQ differences - improved warmth, and improved soundstage depth. The high frequency ferrites don't improve warmth so much (it's pretty much the same), but they improve depth much more. And there are reasons for this. Depth changes are due to signal correlated RF noise which gets into Dave's ground plane, then is demodulated by the analogue electronics. One thing I know from the Dave development is that the perception of depth has no limit to how small these effects may be; in other words, small signal amplitude accuracy has to be perfect. Any change in small signal accuracy no matter how small is audible - and frankly that's really weird and very strange when you think about it. Warmth on the other hand is just any random RF noise, creating intermodulation (with the wanted audio signal) products in the analogue electronics, these intermod products are random noise, and hence we get noise floor modulation, and that's the source of the brightness.
Now with digital electronics we have something called ground bounce. This is current from switching activity within the FPGA or chip that goes into ground, and this current causes the ground wires (and ground plane as a ground plane has inductance too) to bounce up and down due to the ground inductance. We have two forms of noise - clock noise, which for the M scaler is 208 MHz; and noise due to logic level transitions as the internal data signal propagates through the FPGA gates. These delays are from 100pS to 500pS, so are in the 1 to 5GHz region - centred around 2 GHz. This noise (unlike clock noise) is signal correlated, and this would explain why use of 2GHz ferrites improves depth perception. Also, the galvanic isolation is effective at 208 MHz, but is ten times less effective at 2 GHz.
As to the air gap - perhaps this improves the ferrites parasitic capacitance; it's the parasitic capacitance that reduces the effectiveness of the ferrite at higher frequencies - if there was no parasitic capacitance on ferrites or on the galvanic isolation, we would not be having this sensitivity in the first place.
Rob
