Interviews

Virtually all of the active components in your system—DACs, preamplifiers, power amplifiers—work by modulating the DC output of their power supplies with an AC music signal. Surely, then, the more perfect your household AC is, the more perfect your audio system's output will be. Analogies abound—to dirty water used in distilling good whiskey, to inferior thread used to weave fine fabrics—and all amount to the same thing: you can't make a silk purse out of a sow's ear.

The problem with that thinking, like so much of what passes for thought in high-end audio, is that it's qualitative. Circuit designers work hard to design power supplies that stamp out grunge and circuits that ignore it. How successful are they? How much grunge is left at a power supply's DC output? How much of it is left in the output of a DAC or preamp or amp? If those circuit designers succeed, you may not be able to hear the grunge, even if those analogies are fundamentally sound.

I decided to talk about this with someone who knows about power supplies and circuits but isn't in the business of selling power-treatment devices. I turned to Nelson Pass, the founder of Pass Laboratories and First Watt. Experience tells me that Pass is great at explaining complex ideas—and he's been known to design a circuit or two. I asked him if well-designed power-line treatment makes sense, and if so, why.

"In the end," Pass replied, "the criterion that's important is that the circuitry sees a 'quiet and constant companion' in a power supply—quiet in the sense that it is clean, unvarying DC without AC content at any frequency above DC, and reaching into the radar band."

Most power supplies see, at their input, a noisy AC line. Conventional, linear power supplies run the AC through a power transformer, which offers some isolation from noise. The transformer's output goes to rectifiers, which flip the negative-going portion of the sinewave to positive. Capacitors placed across the circuit store charge and smooth the waveform to a roughly constant (DC) value.

Quantitatively, the 120V AC input could be transformed to, say, 50V, then rectified and smoothed to the point where it has perhaps 1V of "ripple." "That's what a typical unregulated linear supply looks like," Pass told me. "If you have a quiet mechanism to regulate the AC line voltage"—ie, some sort of device that improves the quality of your AC power—"all the better." That's because "the amplifier circuit will be affected by the variations in DC and the noise." It's not just amplifiers, though. Other audio devices can also be affected.

The ability of a circuit to ignore or reject noise is characterized by a quantity called the power supply rejection ratio (PSRR), typically expressed in decibels (dB). "If we say that the PSRR of an analog circuit is 60dB, that would mean that the signal the amplifier is handling also will contain variation/noise by a factor of 0.1% of the variation/noise on the power-supply DC line," Pass said. "For the 60dB example, 1V of ripple on the supply becomes 1mV of noise at the output of the amplifier. It's not just noise—if the amp draws down the supply line by putting current through a speaker, the momentary sag shows up in the output as distortion harmonics."

Such issues can be avoided, Pass said, "generally [by] a combination of generous power-supply hardware, additional filtering, and designing circuits which better ignore the variations and noise. The additional filtering can be passive, as in coils and more capacitors, or it can be active, as in linear regulators with transistors."

In designing amplifiers, Pass continued, "the first line of defense is to place the power transformer and noisy diode/capacitor rectification at a distance, preferably in another chassis, and then filter it passively and/or actively before it gets into the amplifier. Wire routing and grounding and shielding are all a big part of this, and then in the amplifier circuit itself, there are numerous techniques to make the amplifier immune to noise and variation. Much of the time we provide separate filtering for the low-power front-end circuits, which works great, and we can design those circuits with semiconductors that don't get upset by a little high-frequency noise. The output stage usually needs the raw supply, though, and some are better at PSRR than others, but what's left is usually taken out with the same feedback that also lowers the general distortion. In balanced output circuits, we catch a bit of a break in that two matched balanced circuits on the same supply tend to cancel supply noise."

Not all noise comes from the outside. Music-producing devices create their own noise—especially power amps. "The amplifier's power transformers, diode bridges, and capacitors will create plenty of noise both at the rail voltages and as noisy magnetic fields from the transformer and wiring," Pass said.

Are AC line-treatment devices necessary? Define necessary. "We try to build equipment that puts up with dirty AC lines," Pass told me. "Our customers expect it, and if they have a problem, we hear about it. We don't design around the assumption that there is AC line conditioning in the system. That said, if the line conditioning has adequate current capacity, we expect some benefit."

That's a big "if," since some types of line conditioner can choke off current and squash dynamics. The same is true of active devices with inadequate capacity, and even a power supply's own internal regulation. "The current peaks drawn are much larger than the continuous amperage," Pass said, "so for power amplifiers you need a rating considerably higher than the nominal current use. If you don't have that, you might get some losses."

Efficiency, then, is important in active AC treatment devices—but the most efficient devices use class-D technology, which is intrinsically noisy. You can make them quiet, Pass said, but it's expensive. The alternative is iron—lots of it. "I once designed a power amplifier that only put out a low-distortion 120V AC," Nelson told me. "It was larger than the power amplifier it was to supply."

Where does high-quality AC power treatment make the most difference: in power amps, or in lower-power devices such as preamps and digital sources? "Given that power amplifiers are going to experience this stuff internally no matter what you do, it makes some sense to put [in] the small effort required to clean up the line for source equipment and preamps and such. They don't demand a lot of AC power and they usually handle smaller signals, making them more susceptible to noise. It's relatively economical to filter their AC line, so this approach is quite practical."

As for power amps, the more expensive and overbuilt an amplifier is, the greater the likelihood that it's already dealing well with AC issues, Pass said: "In the high end, quite a bit of money is spent on this issue. It looks to me like it usually does a good job." Which is not to say that a good regenerator can't further improve the sound; better amplifiers just have less room for improvement.

"I should mention," Pass told me, "that grounding in a system is every bit as important as cleaning up the AC line—but that is a big subject by itself."