Measuring line noise and power conditioners

@erik_squires ,

Now you seem to be backing off your original post.

I never said I'm done. Just reporting my experience.

I guess I never thought that hash on the AC line was associated with any frequency....I assumed it was an equal opportunity offender of all frequencies regardless of where you were measuring the dB reduction.

Uh, RFI and EMI are the same thing. Usually up in the mid-high MHz and low GHz frequencies, you know, radio waves, electromagnetic waves. When you ASSUME something you make a fool out of me and Uma Thurman.

@geoffkait ,

Always appreciate your enlightment.

Of coarse, the logic in me would say, if these two terms are "the same thing", why are there two terms?

A point to keep in mind, which I would expect to be particularly important in the case of class AB and class D power amps (since their AC current draw fluctuates significantly with the dynamics of the music), is that AC current is drawn mainly in the form of narrow spikes, occurring during just a fraction of each 60 Hz period. Specifically, when the instantaneous voltage of the incoming AC waveform exceeds the voltage on the storage capacitors by the small amount that is sufficient to turn on the rectifier diodes.

The spectral composition of that current draw therefore includes frequencies that are much higher than 60 Hz, and filtering those higher frequencies out will tend to adversely affect perceived dynamics and other sonic characteristics. Which is no doubt a major reason why many audiophiles prefer to plug power amps directly into the wall outlet.

In fact Ralph (Atmasphere) has stated in past threads that in many cases these desirable frequency components can range up to several tens of kHz, and I have no reason to doubt that. And if I recall correctly Shunyata has a paper in which measurements are presented supporting that conclusion.

On the other hand, though, it is certainly possible that noise frequencies that are higher than 20 kHz can have audible consequences, by coupling into various circuit points within the components and via effects such as intermodulation and AM demodulation. So as usual in audio tradeoffs are involved. Too low a cutoff frequency will filter out desirable spectral components, and too high a cutoff frequency may not provide an optimal amount of noise filtering. And that tradeoff figures to be dependent on the designs of the particular components and on the spectral composition (i.e., the frequency components) of whatever noise may be present on the incoming AC. Which makes it all very unpredictable.

One thing I would feel confident of, though, is that a simplistic measurement of overall noise amplitude (the noise presumably consisting of a vast number of different frequency components, with most of them differing widely in amplitude and also in terms of their potential to have sonic consequences) probably doesn’t mean much.

Regards,
-- Al