I find the idea of 16 ferrites per cable rather disturbing - there really has to be a more elegant solution out there, speaking of which it's quite likely that IR's and mR's own filtering functions will skew the outcome of any such comparisons.
CAUTION: what follows is a long plea of ignorance and request for more knowledge
I had to deal with magnetic interference at 9kHz for scanning electron microscopes in semiconductor wafer fabs, and looked into generic shielding enough to have a glimmer of *how much I did not know or understand*. This was a "nice battle" to fight because the magnetic field caused the electron beam to wiggle to give wavy images: you could see, measure and document the bottom-line problem. Parameters of interest are frequency of interference, electric vs. magnetic fields, etc. In some cases a "faraday cage" will be an effective shield, i.e., a conductive surface completely surrounding the shielded area. I *think* this is what Enklein does for
http://www.enklein.com/usb.htm. I would love for the TheAttorney to purchase one and report back !-) *I think* that Faraday cages are good for really high frequency electric fields. For my 9kHz magnetic interference, a Faraday shield would not work, so, in addition to making the cover of high permeability metal, we used active cancellation: a Spicer system with a triaxial magnetometer, three huge coils surrounding the ~8'x8'x12' SEM, and a powerful feedback system. Talk about not so "elegant"!
I have found empirically that on the order of 10 ferrite cores on power cables are efficacious. Different ferrites are most effective at different frequencies. However, I do not know what frequency we are fighting, so I used 2 "flavors" of ferrites in some cases: shots in the dark.
(I am an analog guy, and have not cared enough about USB to play in that sand box, yet !-)
So, if someone out there knows either what we are fighting or the best way of doing it, please let us know. References to publications would be nice. Maybe Chord knows?