Why does better power = better sound?


Why does improving power quality improve sound quality?

I’m not asking to start an argument about power cords or wall outlets. Please let’s not go there. I’m asking because I’m hoping to learn some technical explanations for the effects of power quality on sound quality. I think I already understand how…

1. greater current availability = greater dynamic range
2. reduction of RFI/EMI = better signal to noise ratio

…but what about these…

3. ???????? = greater perceived resolution
4. ???????? = more realistic instrument timbres
5. ???????? = more precise imaging

Are differences in resolution, instrument timbres, imaging, etc. somehow reducible to current availability and/or powerline noise? If so, HOW are they reducible?

Again, I’m hoping to get into technical specifics, not polemical generalities.

Thanks in advance.

Bryon
bryoncunningham

Showing 11 responses by almarg

A further thought about power regenerators. My guess is that the root cause of the reduction in dynamics that they are sometimes reported to cause is simply that the 120VAC or whatever that they put out is different (and most likely lower) than the voltage at the wall outlets.

A voltage reduction can make a difference in several ways, to a greater or lesser extent depending on the design of the particular components, especially the amplifier. It could reduce dynamic headroom, reduce maximum power capability, change internal operating temperatures, change bias points, and increase distortion on high volume peaks that approach the reduced clipping point.

On the other hand, if the line voltage happens to be lower than what the regenerator puts out, the voltage increase provided by the regenerator at those locations could decrease distortion on high volume peaks that approach the clipping point. That decrease in distortion could conceivably be perceived as a reduction in dynamics. In a different context (that of SET amplifiers) Atmasphere has commented in the past that since the 5th, 7th, and 9th harmonics of a note's fundamental frequency are significant determinants of our perception of loudness, an increase in those distortion components that occurs primarily on high volume transients will result in a subjective perception of increased dynamics.

Uru, thanks for your comment. Lynne (Arnettpartners) and Bruce (Bifwynne), given the unpredictability and system dependency of it all, I have no particular suggestions beyond what the others have said. Bruce, dedicated lines certainly seem like a good idea, that is amply backed up by anecdotal evidence. I finally got around to having one installed last year, but I upgraded my amplifier at the same time so I don't know how much difference it may have made, if any.

Best regards,
-- Al
Hi Bruce,

I have no particular knowledge of an audio-oriented conditioner that would handle multiple lines at the panel. But in any event I would expect that installing conditioners at the system end of the runs would be preferable, because they would then be able to filter out RFI that may be picked up by the wiring between the panel and the outlets.

It would probably make sense to purchase one conditioner initially, and try it out on each of the different lines.

Not familiar with the Gardner "Pastorale"; thanks for mentioning it. My "go to" version is an imported Japanese CBS/Sony remastering, on LP, of Bruno Walter's famous 1958(!) performance with the Columbia Symphony. I purchased it during the 1980's. Wonderful performance, of course, and remarkably pleasing sonics aside from a bit of steeliness in the strings at times.

Best regards,
-- Al
Hi Lynne,

What is relevant is the maximum amount of current that is demanded by the audio system. As long as a 15A rating would provide a reasonable amount of margin with respect to that demand (and I suspect that it will in your case, assuming that you will not be using a powered sub), there is no problem using a 15A conditioner on a 20A line.

Best regards,
-- Al
Hi Bryon,

Assuming that "noise" is interpreted broadly, to include distortion and perhaps also DC offset, the only other things I can think of would be:

1)Dynamic fluctuations in voltage resulting from resistance in the AC wiring, as current demands fluctuate with the music.

2)Wiring inductance possibly limiting how quickly the supply of current can change, in response to abrupt changes in demand.

I don't have any particular quantitative feel for how much significance those factors may have, if any, and I suspect that in many cases they will be of no significance. Particularly if the power amp is Class A, and hence draws essentially constant current all the time. But those are the only other possible factors that occur to me.

Best,
-- Al
Interesting post, Cjl; thanks!

The one issue I have with your analysis, though, is that it seems to imply that if the impedance looking back into the wall outlet matched the impedance looking into the AC input of the component, the high frequency noise you measured would not cause a problem, and (if I interpreted correctly) might not have been generated in the first place.

As you probably realize, RF reflections occur at each end of a transmission line, in response to arriving energy, as a result of a mismatch between the characteristic impedance of the transmission line and the impedance of what is connected at that end of the line. In the case of most or all power cords, characteristic impedance is undefined, poorly controlled, and very unlikely to match the impedance of either of the things it is connected to, at any given frequency. So RF reflections will occur regardless of the relation between the source and load impedances.

Also, it seems to me that reflections of noise frequencies are not necessarily bad in this situation. If a gross impedance mismatch at the AC input of the component causes incoming noise energy to reflect back toward the outlet, and re-reflections from the outlet to re-re-reflect from the component back toward the outlet, back and forth essentially ad infinitum (the energy gradually being dissipated in the power cord), that would seem preferable to the situation in which a perfect impedance match between the power cord and the component results in that energy being absorbed into the component.

Regards,
-- Al
Hi Bryon,

Good comments by everyone above. My take is as follows, the bottom line of which is essentially what Elizabeth said:

"Dirty power" will consist of some combination of harmonic distortion (i.e., frequency components that are at multiples of 60 Hz, sometimes including multiples that reach into the upper treble and beyond), broadband noise (a mix of essentially all frequencies across a wide range), and voltage spikes that occur periodically or intermittently, which in turn will contain a considerable number of spectral components at various frequencies. Inevitably some extremely small but non-zero fraction of all of that spurious frequency content will find its way through or around the power supply of each component and into the signal path.

In addition to perhaps causing a directly perceivable reduction in background blackness, any and all of those numerous frequency components could, to some small extent, intermodulate with the audio signal, resulting in new spectral components at frequencies equal to both the sum of and the difference between the frequencies of any or all of the spectral components of the music and the frequencies of any or all of the spectral components of the noise or distortion. That will occur as a result of non-linearities in the components in the system, and no system is perfectly linear.

Given that the power line spurii and any musical note will BOTH be comprised of a great many different spectral components, the resulting sonic effects as perceived by the listener can be pretty much anything, conceivably including reduced resolution, less accurate reproduction of timbre, and imprecise imaging.

Obviously those kinds of effects will have little if any predictability with respect to either their magnitude or their character, and will be highly dependent on both the design of the system components and the characteristics of the power source.

In principle a power regenerator should eliminate or at least greatly reduce these effects, but as you've probably seen some people report reduced dynamics or other adverse effects when a regenerator is used, especially in the case of power amplifiers.

Best,
-- Al
Hi Bruce,

If the lights that are dimming are not on the same 20A line as the system, then it suggests that a lot of the voltage drop is occurring in wiring that is upstream of that line, presumably on the input side of the panel. In which case adding additional lines between the panel and the system would not resolve a lot of the problem.

It was some of the others who mentioned the Furmans. I'm using a $259 Brickwall surge suppressor, that also provides a modest amount of EMI/RFI filtering and some degree of isolation between its four dual outlets. I've been happy with the results, although I suspect that I have better than average power quality, as I'm in a fairly small low density town that has no industry and almost no commercial development.

A minor correction: It's the 4th movement that you are referring to, not the 3rd, although the two movements are played with no pause in between. One of my favorite symphonies!

Best regards,
-- Al
Hi Bryon,

The effects I was describing do not necessarily relate to the DC outputs of the power supply at all. There are many possible pathways by which spurious frequency components riding on the incoming AC may couple into the signal path. To the extent that the filtering in the power supply is not perfect at all frequencies, the spurious frequency components may couple into the signal path by "riding" on the DC outputs of the supply. Some of those spurious frequency components may instead completely bypass the power supply, and couple into the signal path through stray capacitances that will inevitably exist in a great many places in the circuitry, or they may couple into the signal path via EMI effects, or they may radiate into the signal path as RFI.

Once the signal, at any given point in the signal path, combines with spurious frequency components that may be picked up at that point via any of those pathways, non-linearities in circuitry that is downstream of that point will result in the spurious sum and difference frequencies I referred to. See this Wikipedia writeup on Intermodulation Distortion.

To answer your specific questions:
Does "frequency intermodulation" basically mean that there are FLUCTUATIONS to DC voltage/current that are UNRELATED to the signal?
If the pathway by which AC line spurii enter the signal path is via the DC outputs of the power supply, then yes, there would be fluctuations in those DC voltages, unrelated to the signal. The value of those voltages at any instant of time would be equal to the numerical sum of the DC voltage and the value at that same instant of time of the fluctuating noise voltage that is riding on it.

However, that is not what is significant. What is significant is the noise coupling onto the signal, and subsequently intermodulating with it at downstream circuit points.
Why are DC fluctuations described in terms of "frequencies" at all? Is it simply because the fluctuations occur at a certain rate per second? Or does the use of "frequency" to describe fluctuations in DC voltage/current also imply that DC can be understood as a WAVE, just like AC?
When DC is fluctuating as a result of noise that is riding on it, it can be viewed as having multiple frequency components that are added together. One component, the DC itself, is at a frequency of zero Hertz and has an amplitude equal to the particular DC voltage. Other components will be present at each of the many frequencies that comprise the noise, with the frequency components of the noise having far smaller amplitudes than the amplitude of the DC. The net voltage at any instant of time will be the numerical sum of the individual amplitudes (voltages) of each frequency component at that instant of time.

The DC (zero Hertz) component of the combination of DC + noise has no relevance to the sonic effects we are discussing; it is just a possible pathway by which the noise may combine with the signal.

Best,
-- Al
Hi Bryon,

You're not there yet, but it is by no means hopeless :-)

My statement that:
The net voltage at any instant of time will be the NUMERICAL SUM of the individual amplitudes (voltages) of each frequency component at that instant of time
had nothing to do with intermodulation distortion, or with the sum and difference frequencies that intermodulation distortion results in. It was simply a description of the voltage at any instant of time of DC that is noisy.

As long as they are reasonably small, fluctuations in the DC voltages will IN THEMSELVES have no effect on anything. It is only when the frequency components corresponding to those fluctuations combine with the signal that a problem arises. The DC will not combine with the signal (in a properly functioning circuit). But the noise frequencies might, because they can couple from one circuit point to another via the various means I described.

Envision a musical note consisting of a 1 kHz fundamental frequency, and a harmonic of lesser amplitude at 9 kHz. (A real musical note would contain many other harmonics as well, such as 2 kHz, 3 kHz, etc. but I'm simplifying). And imagine that a noise frequency of 11 kHz, having some small but significant amplitude, couples onto that signal.

So far all that has happened is that a small 11 kHz signal has been added on to the signal having 1 and 9 kHz frequency components. That is probably not a major issue, assuming that the amplitude of the 11 kHz is not too great.

Now envision that the signal containing those three frequency components is passed through an amplification stage that has some degree of non-linearity, meaning that its output is not perfectly proportional to its input. Intermodulation caused by that non-linearity will result in many new frequencies, corresponding to the sums and differences between those three frequencies, and various other multiples of them. Perhaps most significantly, a 2 kHz frequency will be created as a result of intermodulation of the 9 and 11 kHz components. (20 kHz will also be created, among other new high frequencies, but that is obviously less significant). To the extent that the amplitude of that 2 kHz frequency is significant, it will alter the perceived timbre.

Now envision that a real-world musical note is present, consisting of a great many frequency components, and that noise also present, consisting of vastly more frequency components. Put all of that through a significant non-linearity and what you have is an unpredictable mess, having sonic attributes that can probably differ from those of the original signal in just about any way that is imaginable, including resolution, timbre, and imaging.

Best,
-- Al
Is intermodulation better understood as a fluctuation of voltage or a fluctuation of current? I understand that you can't change one without changing the other (Ohm's law), so maybe that question is meaningless. I would just like to have a better mental picture of what those electrons are doing! :-)
It's best to think of everything I have described in terms of voltage. The current at any given point in the signal path is what it is, based on Ohm's Law.

One reason that all of this tends to be confusing is that the statement which is sometimes seen that the power supply is in the signal path, since it supplies the current that drives the next component in the chain, is a bit misleading.

Yes, in addition to powering the internal circuitry of an amplifier the power supply supplies the current that drives the speakers (supplying it via the transistors or tubes + transformer in the output stage of the amplifier). But a better way to think of it, IMO, is that the output stage of the amplifier will generate a voltage that at any instant of time is (to a close approximation) proportional to the voltage at the input of the amplifier at that instant (actually, slightly before that instant, because it takes a very small but non-zero amount of time for the signal to propagate through the amplifier). The power supply will then provide to the output, via the output transistors or tubes + transformer, whatever amount of current is necessary to establish that output voltage across the load resistance or impedance, consistent with Ohm's Law.

Best,
-- Al
Dover, thanks for your comment.

I don't mean to be picky, but I think it would be better to refer to "reflected noise," rather than "reflection noise." The reason I make a point of saying that is that Cjl's post seemed to say that reflections caused by impedance mismatches PRODUCE noise, which is simply untrue. Reflections of noise frequencies that may be present can ALTER the effects of that noise on the system, although with little or no predictability. But impedance mismatches do not produce noise, and reflections do not occur unless there is something to reflect.
Now unless you are running a floating balanced output stage in your power amp your speakers are connected directly to the main grid on the negative side. The negative speaker terminal in your power amp is on the wall side of the mains transformer.
The negative speaker terminal will usually be connected to AC safety ground (and the amplifier chassis) through a low impedance, such as a low value resistor, or in some cases directly. The paths between AC safety ground and the primary side of the power transformer, in turn, will be via stray capacitance, especially within the transformer, and via the connection that exists between AC safety ground and AC neutral back at the electrical service panel.

In any event, the speaker will only respond to the instantaneous voltage DIFFERENCE that exists between the + and - output terminals of the amplifier. The noise levels that may be present in that difference, as a result of noise on the incoming AC, will be VASTLY lower than the noise on the incoming AC, assuming that the amplifier design is half-way decent. And the speaker, as well as our ears, will not be capable of responding to RF and digital noise frequencies anyway. The possible sonic consequences of high frequency noise on the AC line would result from intermodulation and other effects occurring in internal amplifier circuitry, that may end up affecting audible frequencies.
My understanding is that if a power cable is too short then the pc will behave as if it is part of the loop, whereas if it is say 2m or more it will behave as a spur. There is a difference.
There is no hard and fast dividing line. It depends on the frequency of each particular frequency component of the particular noise that is present, and (to a lesser degree) on the propagation velocity of the particular wiring. For the very high 8 to 13 MHz noise frequencies that were among those Cjl referred to, 2 meters is a reasonable rough rule-of-thumb approximation. Noise frequencies that are much lower would require a much longer cable length to behave as a spur.

Again, thanks for your comment, and my intention in responding is not to be picayune, and (as my earlier posts make clear) is not to deny that power cords can make a difference.

Best regards,
-- Al