Thanks so much to everyone so far who has responded to my posts! I've only skimmed the surface of audiophilia, but I've been into electricity and electronics literally since I was a baby, and I feel at home here. It's great to talk to knowledgeable people who say things that I mostly understand but that still force me to think and do a little research. It may be time to dust off some of those old audio projects that have languished for years. It's definitely time to visit the storage locker and drag out the Froy loudspeakers that I built from scratch more years ago than I care to think about...
IME, the biggest issue with SMPS, notably class-II types, is the BCOD (Blue Cap of Death). Typically 1nF(!) bridging primary to secondary to (more) easily comply to EMC. Injects a ton of half- or full-wave recitifed mains (plus RF hash), much worse that the simple leakage current structure of transformers (where the coupling capacitance is ofter lower except for really large toroids). Even when the secondary is bonded to PE the nasty SMPS leakage still pollutes the PE locally.
For the same reason, Y-caps in mains filters are polluting the PE.
The only way to reduce the leakage is to remove the BCOD and place really large common-mode filters (CM-coils and ferrites, staggered in operating frequencies) on both sides of the SMPS. Bulky and costly. And it still does not help much for class-II
@keantoken,
My experience with ZSRP is that you must make sure it doesn't pick up any magnetic field, be it any air-borne fields or be it fields coming from high AC current paths with non-trivial loop area anywhere nearby. A ZSRP adjacent to a "generic power plane" is pretty much useless, had to learn that the hard way...
For the same reason, Y-caps in mains filters are polluting the PE.
The only way to reduce the leakage is to remove the BCOD and place really large common-mode filters (CM-coils and ferrites, staggered in operating frequencies) on both sides of the SMPS. Bulky and costly. And it still does not help much for class-II
@keantoken,
My experience with ZSRP is that you must make sure it doesn't pick up any magnetic field, be it any air-borne fields or be it fields coming from high AC current paths with non-trivial loop area anywhere nearby. A ZSRP adjacent to a "generic power plane" is pretty much useless, had to learn that the hard way...
Yes, the SMPS trafo leakage capacitance has 180V being switched across it which means you have a high impedance noise source which will for the most part ignore whatever common mode filtering you use. So the solution is a capacitor across the trafo insulation, right at the trafo to provide a short current loop to short out these currents before they reach any long wires or radiate from a large area. But then the capacitor conducts rectifier and mains noise into ground. There are ways around this but manufacturers see no gain in paying for any more capacitors than necessary to meet consumer standards. The best solution for this current is to earth the chassis so that it can be DC decoupled from input to output without sharing a conductor with ground, but then you have to deal with the DC ground loop through the socket ground, which may cause audible hum. Manufacturers will often prefer double insulated because at least SMPS ground noise >20KHz is not directly audible and easy for a consumer to identify as a defect, plus a 2-prong cord is cheaper...
It's probably wrong for me to use the term ZSRP as it implies the plane is used as a reference voltage, which will be inferior to differential tracing below the BW of the ground plane. I guess that means I am just referring to a ZSP, connected to the circuit at only one point as I said. If it is used this way then logically any magnetic fields intersecting with the plane must be reduced, and there are no forced currents to generate their own field below the plane BW.
It's probably wrong for me to use the term ZSRP as it implies the plane is used as a reference voltage, which will be inferior to differential tracing below the BW of the ground plane. I guess that means I am just referring to a ZSP, connected to the circuit at only one point as I said. If it is used this way then logically any magnetic fields intersecting with the plane must be reduced, and there are no forced currents to generate their own field below the plane BW.