What It All Means: line-level preamplifier measurements Sidebar
When a writer gets deeply embroiled in a subject, what he takes to be simple concepts can still lead to confusion among his readers. Take the subject of component output impedances, which we routinely measure as part of Stereophile's equipment reports. Back in early October, veteran audio writer E. Brad Meyer posted a message on the Compuserve "Audio Hardware" Forum in which even he seemed muddled about this concept. Talking about Stereophile's published measurements of output impedance, he went on to say that "I'm led to wonder how Stereophile gets these figures. Ironically, they're more likely to be right if they're taken from the literature, as they would be when a non-technical musician like [Lewis] Lipnick reviews something, than when [Robert] Harley tries to measure for himself. Case in point: In the latest reviews [of the Theta DS Pro Generation III], Harley measures an output of just under 14V RMS at the balanced outputs, and lists the output impedance as about 6 ohms. It seems unlikely to me that the [Theta's] output stage is putting out 30 watts."
It seemed unlikely to us, too. Quad's Peter Walker once said that everything in audio can be figured out using Ohm's Law and common sense. Mr. Meyer's "30 watts" is derived from taking the Theta's maximum output level of 14V RMS and, by Ohm's Law, deriving the power delivered into a 6 ohm load. In this case, power = V x I where V is the applied voltage and I the current flowing in a resistor R. But as I = V/R, the power = VxV/R. 14V applied to a 6 ohm load would therefore give (14x14)/6 watts of power, or 32.67W.
But the Theta isn't driving a load impedance of 6 ohms with that 14V; rather, it has an output impedance of 6 ohms (footnote 1). This means that its output resembles a perfect, loss-less AC voltage generator coupled to the outside world through a 6 ohm resistor (as shown in fig.1). In his critical zeal, Mr. Meyer has confused the concept of output or source impedance with that of load impedance. In the common-sense case of the Theta hooked up to a preamplifier with a typical load or input impedance of 10,000 ohms (or 10k ohms) (fig.2), the Theta will quite happily source up to 14V RMS from its 6 ohm output impedance. By Ohm's Law, the output current will be 14/(10,000 + 6) amps, or 1.39 milliamps (mA), and the power dissipated in the Theta's output stage will be less than 0.01 milliwatts (mW), this rather lower than the supposed 30W and well within the output devices' capabilities.
What if, instead of hooking up the Theta to a preamp, we short-circuit its output to ground (fig.3)? Now the full 14V will be developed across the Theta's 6 ohm output impedance, leading to 32.67W of power being dissipated if the output stage can deliver the necessary 2.33A of current. Of course it can't, but it's possible that, in trying to do so, the Theta's output devices will melt (footnote 2). Which is why it is unwise to short the output of a component being driven at its maximum level---unless you want to break it!
If we publish measurements of output impedance, we must think them important to sound quality. What are "good" and "bad" measurements, therefore? Basically, the lower the output impedance, the farther away from the audio band will be the capacitive and inductive effects of cables, and the more easily the component will drive low impedances (up to its current limit, of course). With a high output impedance, the effects of cables will be much lower in frequency, even resulting in some HF rolloff, while the load impedance of the next component in the chain must be kept high, which might not always be feasible.
The California Audio Labs Sigma D/A processor reviewed by Robert Harley in October 1993, for example, has both a high output impedance of 2000 ohms and a limited current delivery. Not only should Sigma owners keep interconnects short to avoid HF rolloff, but, as RH noted in his review, they need to use a preamplifier with an input impedance of more than 14k ohms to avoid current starvation at high signal levels, which rules out many passive control units and solid-state designs. By contrast, the Melos SHA-1 headphone amplifier also reviewed in October '93 has an output impedance from its 'phone jacks of just 0.7 ohms, and can deliver its maximum output voltage of just under 4V into anything from a piece of damp string down to low-impedance headphones.
Footnote 1: We estimate the output impedance of a product by measuring its output voltage at around a 1V level into an open circuit (actually the 100k ohms maximum input load of our Audio Precision System One), then measuring the voltage drop as that load is reduced to 600 ohms. The component's output/source impedance can then be calculated using Ohm's Law.
Footnote 2: Actually, the BUF-03s used in the Theta DS Pro will protect themselves in the event of their outputs being short-circuited, but it is still unwise to do this unless no signal is being passed.