reverberant echoes, so the McCormack with its more tonally accurate upper midrange reveals more of this natural imaging information, while the Adcom with its tonally subdued upper midrange will of course reveal less of it. In sum, these are two excellent amplifiers, with distinct sonic personalities, and only your personal taste can determine which you would prefer.
      The Adcom GFA-7807 achieves its "just right" euphonic transformation by making the bass and warmth regions richer, by subduing the upper midrange (thereby making the sound darker as well as warmer), and by sweetening and softening the trebles. But, for this transformation to be just right for euphonically offsetting the typically lean, bright, hard sound of closely miked recordings, you should rely on only this GFA-7807 to be doing your sonic transformations, so the rest of your system should be accurate. There are a number of loudspeakers and cables on the market which also sound warm and dark and soft. And there are some electronic components such as preamps and disc players which have these same sonic personalities and similarly transform the sound (q.v. Adcom's own DVD player, below). If you use any of these in your system together with the GFA-7807, then their sonic transformations will be in series and will add, giving you too much of a good thing, and producing a total system sound that in our professional judgment is too warm, too dark, too soft. For this sonic transformation to be truly euphonic in your total system, it has to accurately re-create the sound you hear from a distance at a real concert hall, not go over the top into caricature.

Getting the Most from Your GFA-7807

      The Adcom GFA-7807 generously includes balanced XLR as well as the usual unbalanced RCA inputs, which is a welcome surprise in an amplifier at such a bargain price. Normally, we find the balanced XLR connection between electronic components to be sonically superior and far preferable. It usually affords a more airy, open sound and better dynamic punch, as well as the expected benefit of lower noise. We were disappointed to discover that this was not the case with the GFA-7807. Using our normal system test setup for evaluating balanced vs. unbalanced inputs on power amplifiers, we found that, with the GFA-7807, the unbalanced RCA input yielded the wonderful sound that we have praised in this review, but that the balanced XLR input sounded artificially hard, strident, and veiled, making the GFA-7807 sound much more like a typical bipolar solid state amplifier.
      Why could this be so? On the GFA-7807, the input impedance of 100K ohms for the unbalanced RCA input drastically drops to a mere 10K ohms for the balanced XLR input, due to an unusual resistor dividing network that Adcom has inserted across the input for the balanced XLR input, so it is possible that at least some of the sonic degradation we heard was caused by this low 10K ohm input impedance stressing the audio component driving the GFA-7807. In any case, this means that we would not recommend using the balanced inputs of the GFA-7807 for domestic audio systems, since many domestic audio components are not intended for driving low impedances like 10K ohms with optimum fidelity, and since the GFA-7807 sounds much worse through its balanced XLR inputs with such domestic components, whatever the cause. On the other hand, in professional studio environments, even lower impedances such as 600 ohms are the norm, and so the GFA-7807's low 10K ohm input impedance would not be problematic here (but the jury is still out on whether the GFA-7807's balanced XLR input sounds intrinsically worse than its unbalanced RCA input, for some other reason).
      Fortunately for you, the unbalanced RCA connection is of course the common connection readily available on all domestic electronic components, so you can simply use the GFA-7807's unbalanced RCA inputs, secure in the knowledge that you are hearing the Adcom GFA-7807 at its wonderful best.
      Now we come to our most important discovery for you, to get the best sound from the GFA-7807. It's a very simple setup tactic, but it is very surprising. The sound of the GFA-7807 crucially depends on where you plug it in.
      To hear the wonderful magic we have described above, you have to plug in the power cord of the GFA-7807 into a power socket immediately next to the socket where you plug in the power cord of your source electronic component that feeds the audio signal into the GFA-7807. The purpose of this mandatory tactic is to provide the shortest possible (yet practical and safe) direct ground bridge, from the ground prong of the GFA-7807's captive power cord to the ground prong of the power cord for your source component feeding the audio signal to the GFA-7807.
      If you plug both units into standard wall duplex power sockets, you should simply plug them both into the same duplex socket. If you use a power strip with a row of power sockets, plug them into immediately adjacent sockets. If you use one of those two-into-six, square shaped wall socket multiplier adapters, plug both units into immediately adjacent, side-by-side power sockets (this gives a shorter and better sounding ground path connection on these adapters than vertically adjacent sockets would).
      Of course, one implication of this setup dictum is that the GFA-7807 will be physically located near your source and control components. And your system will be running short interconnects and long loudspeaker cables, not long interconnects and shorter loudspeaker cables.
      If you don't obey this simple but crucial setup tactic, the sound you hear from the GFA-7807 will probably be worse than described above. Much worse. When we tried plugging the GFA-7807 into a nearby wall socket, but not the same wall socket, as the source component feeding the signal into the GFA-7807, the GFA-7807 suddenly became, to be frank, a distinctly mediocre amplifier. Its sound became very veiled, distorted in a grundgy way, and quite smeared. Also, the very slight slurring of treble transients, which accompanies the GFA-7807's euphonically softening treble defocus when this amplifier is heard at its best, became instead a woefully stretched out and pronounced slurring of what should be individually articulated treble transients. The point of our candor here is not to criticize this product, but rather to warn you to take seriously our admonition about where you plug in its power cord. If you were to blithely ignore our advice and plug the GFA-7807 into any old wall socket, then you might be very disappointed with the sound you heard from it, and you might think you had a defective unit, or you might wonder about the perspicacity of this reviewer's ears and analytical judgment.
      Now, why should this setup tactic make any difference, and why is it so crucial for hearing the GFA-7807 at its sonic best? To understand this, we have to briefly investigate three questions. First, what is the role of an ideal ground in audio electronics? Second, how does a real ground differ from an ideal ground? Third, why are there plural real grounds in audio circuits instead of just one real ground, and how can this mess things up?
      In the colloquial view of audio electronics, an audio signal inside an amplifier is carried along a single wire from solder lug A to solder lug B, and it obviously exists at solder lug A, and it also exists at solder lug B. Meanwhile, the function of a ground is merely to provide shielding, to protect the audio signal traveling in that single wire from external interference. But the scientific truth is quite different. In truth, an audio signal does not even exist in any single wire, nor at any single point like solder lug A. Rather, an audio signal is defined by, and exists only as, a potential difference (or voltage) between two points. Because its very definition and essence requires two points, an audio signal cannot exist at any one solder lug or in any one wire. An audio signal can only be defined with reference to these two points. It follows that the nature of both defining reference points must be specified and known before we can know the true nature of the audio signal (and, for those of us concerned with high fidelity, the true nature of the audio signal is what it's all about).
      Now, for the sake of convenience and simplicity, in most circuits one of these two defining reference points is held at a constant value throughout the whole circuit, while the other defining reference point of the signal (which we'll call the high point) is allowed to vary, as this high reference point endures many adventures and manipulations in its journey through the amplifier circuit, e.g. getting amplified, going from solder lug A to solder lug B along a single wire, etc. The first defining reference point, the one that is supposed to be held constant throughout the circuit, is called the ground reference. What we colloquially call the audio signal, as it journeys and varies through the amplifier circuit, has no meaning and no definition in itself. Its actual nature, indeed its very existence, can only be known and defined with reference to that other hopefully constant point, the ground reference (since, as noted, the audio signal is defined as the potential difference between two defining reference points). When we enquire into the nature, accuracy, and fidelity of an audio signal, we must look at two points, not just one, and we must exactly evaluate the difference or voltage between these two points, the ground reference and the chosen audio signal measuring point (such as the input or output jack of the amplifier). It's almost as if the audio signal actually had two portions, its high side that is processed by the regular audio circuitry we know, and the ground reference side that should ideally be constantly the same throughout the amplifier circuit.
      A quick analogy may help. Suppose you're judging a contest for the world's tallest skyscraper. Do you just look at the tops of the buildings, to see which stretches the farthest into the sky? No. The height of a skyscraper, the nature and measure of its construction height achievement, is defined by two points, not just one point (the top), and, to be more explicit, is defined as the difference between its top point and its own local reference ground level. If one skyscraper is built on ground at sea level in Manhattan, but the second competing skyscraper is built on ground that's atop cliffs right across the Hudson River (say 1000 feet above sea level), you can't just eyeball the tops of the two skyscrapers as seen side by side from a distance. Instead, you have to take into account the differing ground baselines, since the construction height of each skyscraper's top is actually its height with reference to its own base at its local ground baseline, and in this case the local ground reference baseline differs for the two skyscrapers.
      Ideally, a ground reference baseline is exactly the same throughout the complete circuit, so that all the various audio signal high side points, as the audio signal is manipulated through the various stages of the complete amplifier, are defined with reference to exactly this one and the same ground reference baseline. In this way, when a high fidelity amplifier circuit seems to be doing a good job because its output signal high side point appears to be a nearly exact facsimile of its input signal high side point (for all moments of time), it truly will be doing a good job, because both the output signal high side and the input signal high side will be referenced to exactly the same ground reference baseline.
      But what if the local ground reference baselines were subtly different at different physical locations, say at the output jack vs. the input jack of the amplifier? Then, even if the main audio circuit were doing an absolutely perfect job of reproducing an exact facsimile of the input signal high side point at the jack where the output signal high side point appeared, the actual true output signal would not be an accurate high fidelity facsimile of the true input signal, because the output signal would be referenced to a different ground level baseline than the input signal was.
      Go back to our analogy of the two skyscrapers across the Hudson from each other. Imagine that the job foreman atop the competing skyscraper being constructed atop the cliff on the west side looks across at the already completed skyscraper based at sea level in Manhattan, and sees that the altitude into the sky of his own high point is now an exact facsimile of the altitude into the sky of the high point of the already completed skyscraper in Manhattan. So this job foreman yells down to his boss, telling him that they don't need to build any higher, since they have now exactly matched the height of the competing skyscraper in Manhattan. That's like the output jack of an amplifier looking across at the input jack, seeing that the high side points of the audio signals at output and input exactly match in altitude (for all moments in time), so he says to the amplifier reviewer, hey, I'm a perfect amplifier, because the high side points of my output signal exactly match (i.e. are an accurate facsimile of) the high side points of my input signal. Of course, what the job foreman forgot was that the two buildings have different ground reference baselines, so they actually have different heights above the ground. Likewise, in the amplifier, the actual output signal would not in fact accurately match the true height of the actual input signal, because their ground reference baselines were not exactly the same.
      Why does a real ground reference baseline fail to live up to the ideal of being exactly the same throughout the circuit? Why in reality are there differing local ground reference baselines for various parts of an amplifier circuit? As you know, the circuits that process an audio signal (actually, as we now know, just the high side of that audio signal) are never ideally perfect. The reality is that it's impossible to get an amplifying stage to reproduce at its output a perfect facsimile of its input. Heck, it's even impossible to get even a simple wire to perfectly carry an audio signal from one amplifying stage to another (that's why different wires sound different). It takes science, engineering, and artful skill to work around the imperfections imposed by the realities of nature, and execute a passably decent amplifying circuit.
      Similarly, the circuits that handle the ground reference side of the audio signal are also bedeviled by imperfections imposed by the realities of nature, and they too require considerable engineering attention and skill. Just as a simple wire is still notably imperfect at carrying the high side of an audio signal, so too are notably imperfect those simple wires and ground planes that carry and distribute the ground reference side of an audio signal, to and among all the various physically dispersed locales of an amplifier circuit. In other words, the ground reference side of the audio signal in reality fails to achieve the ideal of being exactly the same throughout the amplifier, because the circuits carrying this ground reference side of the audio signal are notably imperfect. The ground reference at the input jack of the amplifier is not carried and distributed perfectly to other parts of the amplifier, including the output jack. Thus, even if the amplifying circuitry were hypothetically to do a perfect job of reproducing the high side of the input signal at the high point output jack, the total true signal at the output jack, defined as the difference between the high point and the local ground reference baseline at the output jack, would in fact not be an accurate facsimile of the total true signal at the input jack, since the local ground reference baseline at the input jack would not be the same as the local ground reference baseline at the output jack.
      There are many technical reasons for these imperfections in the ground reference side of the audio signal. Here's a brief list, just so you can appreciate how tough a job the amplifier design engineer has in trying to come closer to the ideal of achieving a truly constant ground reference that is exactly the same throughout the amplifier circuit. The wires and planes carrying the ground reference side of the audio signal are prey to resistance drops, inductance drops, stray capacitive coupling, induced noise, circulating eddy currents, circulating loop currents, modulation over time by the changing fields and current draws from the high side of the audio signal as it is processed by various amplifying stages, etc., etc.
      Note that some of these imperfections vary with frequency (e.g. inductance), and some of them also vary over time (e.g. modulation over time by the changing high side of the audio signal). Thus, the accuracy and constancy of the ground reference side of the audio signal will be degraded and contaminated nonlinearly, varying with respect to both frequency and time. This means that a real ground reference baseline, instead of meeting the ideal of being constant with respect to time and constant with respect to frequency and constant with respect to physical location in the amplifier, actually fails to meet all three ideals.
      Whenever the accuracy and constancy of the ground reference baseline is degraded, the total true audio signal is degraded, since the total true audio signal is defined as the voltage difference between the high point and the local ground reference baseline. Therefore, the total true audio signal will be degraded nonlinearly, varying with respect to both frequency and time, as well as various physical locations in the amplifier, merely because the ground reference baseline was thus degraded. An amplifier could cause automodulation distortion of itself, if the varying audio signal caused the ground reference baseline to vary, thereby distorting the amplitude of the true total audio signal. Even an amplifier whose active circuitry is seemingly perfect might therefore nevertheless exhibit frequency aberrations and nonlinear distortion in its output, due simply to imperfect design of the ground reference circuit.
      Indeed, we recently witnessed a startling example proving this point. We heard a demonstration comparison of two power amplifiers, the second having exactly the same active circuit as the first, but transplanted to another chassis package. The first unit sounded like (and is) one of the very best, greatest amplifiers on the planet. The second unit sounded like a run of the mill, second rate solid state amplifier (with excessive bright energy from clearly audible distortion in the upper midrange and lower treble; plus artificially hard glare and clogging, obscuring veiling in those spectral regions; and then dull, amputated upper trebles). What on earth could make the second unit sound so radically inferior to the first, when they were identical active circuits? After exhaustive diagnosis, it was discovered that the internal ground reference circuit had been slightly changed when the same active circuit was transplanted to the different chassis package, and that this one minor change in the internal ground reference alone had caused all these drastic sonic degradations.
      Incidentally, many balanced audio signal circuits are still susceptible to all the above problems, arising from flaws in their ground reference circuits. Balanced audio signal circuits process two high sides for the audio signal simultaneously, a plus high side and a minus high side, one hoped-for goal being that imperfections in the ground circuit will not affect the amplifier's performance, since at the output (and input) of the amplifier the true total audio signal is sensed as the difference between the plus high point and the minus high point, with the ground reference level being irrelevant. That's a worthwhile ideal goal, but it is rarely achieved in practice. That's because, within most balanced amplifiers, especially at all the intermediate stages, the circuitry handling the plus high side senses and processes its signal the same way as a single ended circuit does, as defined as the difference between the plus high side and the local ground reference baseline, while the circuitry handling the

(Continued on page 125)