Here is a cut-and-paste from an article written by Ken Stevens of CAT in 1998 (w.r.t. to AC power it should be relevant even today):
--------------
The EI transformers superior handling of DC offset in the primary is certainly part of the story. Also the EI transformer handles the DC current imbalance in the secondary better as well (this caused by non-exact matching of diodes in a full-wave bridge or even more so by mismatched halves off center-tapped secondaries).
I believe, however, that the main reason for the better sound with the EI transformer is its vastly superior rejection of RF noise from the power line. Toroidal transformers have two major problems regarding noise. The first problem is the extremely low leakage inductance of even a sloppily wound toroid. I am not exaggerating to say that this leakage inductance will be more than 10 times lower in a toroidal transformer than in a simply wound EI transformer. Indeed, it may approach 100 times lower. Normally, because low LI yields wide bandwidth, one would assume that this is a good thing. Unfortunately, in the case of a power transformer, this gives the toroid an extremely wide passband for differential noise (i.e. noise which is on line but not on neutral) to get through to the equipment.
Some add-on line-filtering devices attempt to shunt this differential noise to neutral (converting it to common-mode noise) and/or "ground" (supposedly eliminating it). These devices can be somewhat effective on equipment that is extremely bad in this regard (i.e. those with toroids), but their effectiveness is limited in two major ways. First is the fact that the inductors and caps used have self resonances that make them ineffective beyond that frequency. Typically a good film cap will be limited to a megahertz or so and likewise for a similar-quality inductor -- totally ineffective for FM or CB. Of course you may use a tiny cap with a higher resonant frequency, but this will be a very small cap which will be less able to "short" the noise to the other side at any given frequency. For example, a 100pF ceramic cap may have a resonant frequency of 50 to 100 meg, but its impedance at 10 meg is about 160 ohms, not a very good "shunt" to ground at all. Bypasses don't solve this problem at all -- in fact, they make it worse because the parallel combination of two low-loss caps will have an extremely high-impedance resonance between their two low-impedance resonances, making the majority of the "in-between area" worse than if the bypass werent used at all.
The second major problem with using a network to shunt the noise to ground, or neutral, is that the ground wire in your wall socket is not a good ground at all for radio frequencies. The problem is that the length of wire running from your socket to ground is substantially greater than a wavelength at FM frequencies and therefore is disconnected from ground at those frequencies. Indeed, your home wiring is more of an antenna, actually picking up RF signals quite well. Of course you can't dump RF noise from line to a ground, which is just as dirty.
The next problem with toroidal transformers is their very high capacitance between primary and secondary windings. This high capacitance is due to the much larger surface area between adjacent layers of winding on a toroid versus a similarly sized EI transformer. Of course, you could put extra insulation between the primary and secondary layers of a toroid to reduce this capacitance, but the same amount of insulation would yield an even greater reduction in the capacitance of the EI transformer. All things being equal, the EI transformer wins big. Of course, this high capacitance creates a lower impedance path for common-mode noise -- noise which is the same on line and neutral (and ground also). At high frequencies, like FM, the noise is mainly common-mode because of the fact that the three conductors -- line, neutral and ground -- travel side by side through your house acting more like antennae for RF than anything else.
There are numerous advanced tricks for further reducing the two problems mentioned above, although one must be careful not to overdo it. If you went crazy with increased leakage inductance, for example, the resonance set up by the transformer and the caps after the full-wave bridge could drop too low for good regulation, resulting in boomy bass and dynamic compression.
-------------------------------
Personally, I've heard a few Denon CD players (used as CD players & as transports) & I seem to find them ALL to be rather "screechy" in the upper freq delivery. IMHO, of course. YMMV.
--------------
The EI transformers superior handling of DC offset in the primary is certainly part of the story. Also the EI transformer handles the DC current imbalance in the secondary better as well (this caused by non-exact matching of diodes in a full-wave bridge or even more so by mismatched halves off center-tapped secondaries).
I believe, however, that the main reason for the better sound with the EI transformer is its vastly superior rejection of RF noise from the power line. Toroidal transformers have two major problems regarding noise. The first problem is the extremely low leakage inductance of even a sloppily wound toroid. I am not exaggerating to say that this leakage inductance will be more than 10 times lower in a toroidal transformer than in a simply wound EI transformer. Indeed, it may approach 100 times lower. Normally, because low LI yields wide bandwidth, one would assume that this is a good thing. Unfortunately, in the case of a power transformer, this gives the toroid an extremely wide passband for differential noise (i.e. noise which is on line but not on neutral) to get through to the equipment.
Some add-on line-filtering devices attempt to shunt this differential noise to neutral (converting it to common-mode noise) and/or "ground" (supposedly eliminating it). These devices can be somewhat effective on equipment that is extremely bad in this regard (i.e. those with toroids), but their effectiveness is limited in two major ways. First is the fact that the inductors and caps used have self resonances that make them ineffective beyond that frequency. Typically a good film cap will be limited to a megahertz or so and likewise for a similar-quality inductor -- totally ineffective for FM or CB. Of course you may use a tiny cap with a higher resonant frequency, but this will be a very small cap which will be less able to "short" the noise to the other side at any given frequency. For example, a 100pF ceramic cap may have a resonant frequency of 50 to 100 meg, but its impedance at 10 meg is about 160 ohms, not a very good "shunt" to ground at all. Bypasses don't solve this problem at all -- in fact, they make it worse because the parallel combination of two low-loss caps will have an extremely high-impedance resonance between their two low-impedance resonances, making the majority of the "in-between area" worse than if the bypass werent used at all.
The second major problem with using a network to shunt the noise to ground, or neutral, is that the ground wire in your wall socket is not a good ground at all for radio frequencies. The problem is that the length of wire running from your socket to ground is substantially greater than a wavelength at FM frequencies and therefore is disconnected from ground at those frequencies. Indeed, your home wiring is more of an antenna, actually picking up RF signals quite well. Of course you can't dump RF noise from line to a ground, which is just as dirty.
The next problem with toroidal transformers is their very high capacitance between primary and secondary windings. This high capacitance is due to the much larger surface area between adjacent layers of winding on a toroid versus a similarly sized EI transformer. Of course, you could put extra insulation between the primary and secondary layers of a toroid to reduce this capacitance, but the same amount of insulation would yield an even greater reduction in the capacitance of the EI transformer. All things being equal, the EI transformer wins big. Of course, this high capacitance creates a lower impedance path for common-mode noise -- noise which is the same on line and neutral (and ground also). At high frequencies, like FM, the noise is mainly common-mode because of the fact that the three conductors -- line, neutral and ground -- travel side by side through your house acting more like antennae for RF than anything else.
There are numerous advanced tricks for further reducing the two problems mentioned above, although one must be careful not to overdo it. If you went crazy with increased leakage inductance, for example, the resonance set up by the transformer and the caps after the full-wave bridge could drop too low for good regulation, resulting in boomy bass and dynamic compression.
-------------------------------
Personally, I've heard a few Denon CD players (used as CD players & as transports) & I seem to find them ALL to be rather "screechy" in the upper freq delivery. IMHO, of course. YMMV.