[BoyElroy]

I'm not sure if this belongs in a separate thread or not, but I've been trying out different psu's to use with my dc coupled szekeres and as Tomo says in the addendum to the original szekeres article, power supply designs do seem to make a big difference in sound quality.

So, my question is, and pls. forgive me if this seems too broad, what are the critical factors that account for these perceived differences? Some power supplies seem to open up a wide soundstage and others seem to flatten the various layers of sound. What accounts for these differences? The amount of regulation, the type of regulation, quality of caps, types of transformers, etc...?

I know that in audio circuit design, as per Nelson Pass' theories on minimized gain stages, simpler does tend to sound better than more complex. I can understand this in my layman's view of the audio signal as simply the fewer components you run an analog signal through, the more of the original signal integrity you retain.

However, with power supplies, the opposite seems to hold true. A very simple power supply will generally sound quite poor and ultra regulated power supplies sound much better. My question is, what is being gained (no pun inteneded!) by regulating a power supply beyond one or two stages? I noticed that one v. expensive psu that I bought recently has a 317 IC followed by several other transistors and many, many other unidentified caps/resistors.

What is the purpose of all this regulating beyond the initial 317 stage?


I hope my question makes sense and I apologize for my technically imprecise language, but I am pretty flummoxed by this psu black box mystery.


Thanks all--

 

[tophu]

I think the most critical factors in a power supply are:

1) how much peak current can it provide
2) how long can it sustain this amount of current
3) how much sustained current can it provide
4) how much noise does it carry
5) how does it react to loads

Some comments on these:

1 is often addressed by large capacitors. However, these are not always ideal from the point of view of 2.

If it has large values of 3, then it tends to not be very affected by differing loads, and this makes it behave well in question 5.

However, it is difficult to have something that gives large amounts of current and also have very low noise.

Thus, power supplies tend to be excercises in tradeoffs.

 

[Possum]

My simple explanation is to filter out noise (stray AC on the rails), with the goal of eliminating ripple, so the output DC is one constant voltage without any variation. However, completely eliminating ripple and noise might not be possible or easy. I've seen initial stages in a PSU work on high frequency filtering. Right out of a full-wave or bridge rectifier, the voltages you're getting are like half sine waves, one half after another. Once you add capacitors and regulators, the half sine shape begins to disappear leaving you with something that looks like the peaks of the half sines trailing off to under the peak of the next half sine. The difference between bottom and top part of the resulting waveform is your ripple voltage. The ripple voltage can sometimes still be noticed as noise in your equipment. The complex regulators and extra stages are to reduce the ripple voltage as much as possible. (Try doing a search on the web for a picture of what ripple voltage looks like in a Voltage vs. time plot and compare it to a pure DC voltage, which is a completely straight/horizontal line.)

As you've read, the simpler designs in an amplifier circuit can retain the original signal integrity. However, this doesn't mean the power supply has to be simple. Your sound signal runs through the amplifier circuit (simple), but not through your power supply circuit. But if the power supply is pretty noisy, than it can negatively affect the performance of the amplifier portion, which by being simple can also mean that it's more sensitive to the power supply. By isolating the power supply from outside noise on your power lines, and achieving the clean DC voltage through regulation, the simplicity of the amplifier can do exactly what it was designed to do, without passing on power supply noise/ripple as abnormalties in your audio signal.

 

[tangent]

Something I've discovered recently: power supply ripple doesn't only come from AC->DC conversion. You can get power supply ripple with a battery power source, if you don't have enough current capability from the batteries and/or enough rail capacitance to make up for that lack. I've watched power supplies ripple on the order of hundreds of millivolts in time with the music on an oscilloscope.

It's like a WinAMP visualization for electronics geeks.

In really bad cases with artificial torture testing, I've seen 1.2V of ripple on an 18V power supply that has only a few mV of ripple at idle. Adding rail capacitance fixes this.

 

[Joe Bloggs]

I think the difference is that your sound signal is pure goodness coming out of the first stage and more stages mess up the signal whereas your power supply voltage is pure crap to begin with and needs as much cleaning up as it can get