Originally Posted by Steve Bruzonsky
Curt, THD is measured on the AC, that's why the UPS units mentioned above list THD in their specs.
Correct. THD, by definition, is the amount of energy in the signal that is composed of non-fundamental harmonics of the base frequency - 60Hz in the case of US AC power. There are actually two different definitions - THD, and thd; they differ in that THD is a %age of the total energy, while thd is a percentage of the fundamental energy.
I would still like someone of engineering acumen to explain how THD on the AC affects both analog and digital switching power supplies. Thanks.
In small amounts (<5%) it should have no discernible effect on the output of ANY solid power supply design, SMPS or linear.
At increasing levels, you can expect linear supplies to begin to have heating and noise issues, as the energy in the harmonics can cause excessive losses in the transformer. These inductive losses actually block the noise from getting into the DC side of the supply; the requisite capacitive filtering and voltage regulation stages will further isolate this noise from the output DC rails. Bluntly, if the capacitors can't filter out 5% THD, then they can't filter out the 60 Hz base wave that comes out of the rectifier. You might potentially have some inductive coupling (EMC) that would allow mid-frequency noise to get into the high-impedance circuitry elsewhere in the product, but this would be a product of very poor design in shielding and layout - not a problem with the PS. Now, you CAN have certain frequencies that will pass through a linear PS, and there is a point where the caps just can't filter out everything, but it tends to take quite a bit of 'junk' on the line to cause real noise to appear on the DC side. Transients are a much bigger issue with linears; they often dump a noise event into the device ('pop'); however, it doesn't take a $3500 power conditioner to deal with transients.
[Caveat: if the PS design has some bizarre resonant point that happens to line up with a non-fundamental frequency represented in the line voltage, it could cause overheating and reduce the supply's ability to deliver it's design-requirement current. In this case, the only good solution may be to use either a tuned filter or a double-conversion system.]
SMPS are a different animal entirely. SMPS devices actually like a square wave, as they can convert all of the energy in the harmonics, as well as in the fundamental. They actually create a tremendous amount of distortion in the switching process, then filter it all back out on the output side. THD should have NO impact on a SMPS. This is why the el cheapo 'shoebox' UPS units at the big-box office stores can get away with putting out a square wave (or 'modified' square wave) - all PCs and most other office devices use SMPSs, and are perfectly happy with that kind of power.
This is why you get so many people questioning the value of power conditioning for projectors that use a SMPS; there's just no good reason for it to matter. If it does make a difference, it is probably due to grounding, shielding, or some other system problem - not a line noise or voltage regulation issue.