minus high side similarly relies on the local ground reference baseline for defining its signal. Thus, both top and bottom halves of this so-called balanced circuit are individually acting as a single ended circuit, and each is susceptible to imperfections in the ground circuit reference. There is potentially some partial salvation if both single ended halves of this balanced circuit are truly perfectly symmetrical with respect to each other in handling their respective high sides of the audio signal (which they almost never are in practice), and if the local ground reference baseline is exactly the same for both single ended halves, at each intermediate stage of the amplifier. Then, both single ended halves will still be susceptible to imperfections in the ground reference, but hopefully they will both be susceptible in exactly the same (but complementary) way and to exactly the same degree (again, unlikely in practice), so that the imperfections they picked up from the errant local reference baseline will cancel each other out at the final differential output of the amplifier.
      Now that we all understand how important it is to have a single ground reference circuit, which as accurately as possible carries and distributes a single constant ground reference baseline, to be exactly one and the same throughout the amplifier, it's time to throw a few new monkey wrenches into the pie.
      It turns out that a real amplifier doesn't have just one ground reference circuit, which the designer must worry about making as ideal as he can. A real amplifier, even a single channel monoblock, has at least three different ground reference circuits. There's the ground reference for the active audio circuitry, which we discussed above. But then there's also a ground reference for the active power supply circuitry, which is a whole separate circuit from the audio signal circuit, and has its own separate ground circuit as a ground reference baseline for all the various power supply voltages. And then there's a third ground reference circuit, comprising the amplifier's chassis and enclosure. These shield the amplifier's active circuitry from external (and internal) fields, and they must be able to sink induced currents from all these fields without contaminating the other two ground reference circuits. Thus, the amplifier design engineer has to worry about optimizing the performance of three distinct ground reference circuits, not just one.
      These three types of ground reference circuits differ from one another in nature and in purpose, so the best engineering solution for each of these three ground reference circuits must be addressed differently. These three types of ground reference circuits also differ in how they communicate their baseline reference with the external world. The power supply ground reference circuit is usually tied to the center ground prong of the amplifier's power cord, so it derives its external reference baseline via the power cord from the power socket in your wall. In contrast, the ground reference circuit for the amplifier's audio signal circuit is usually tied to the ground side of the audio signal input jack, so it derives its external reference baseline via the input audio interconnect cable from the previous source component in your system, specifically from its ground reference circuit for its audio signal circuit.
      After the design engineer has optimized the design of each of the three types of ground reference circuits in his amplifier, there are still further complications. These three different ground circuits must usually be brought together at some physical solder lug, and must usually communicate with each other. In large part, that's because the high sides of the power supply circuit's voltages are deliberately brought in contact with the high sides of the audio signal circuit's voltages, at every audio signal amplifying stage, in order to serve their function of energizing these audio signal amplifying stages in various ways. When the high sides of these two types of voltages, audio signal and power supply, are brought into contact in order to service each amplifying stage, then their respective ground reference circuits also have to be brought into contact, since each of these two types of voltages can only be defined, at each amplifying stage, with reference to its own ground reference baseline.
      Additionally, at the output stage of a power amplifier the amplifier's actual output signal comes from the power supply, not from the audio signal amplifying circuitry (which acts as a mere proportioning valve). This means that the accuracy and fidelity of a power amplifier's output might well be defined with respect to the ground reference circuit of the power supply, not with respect to the ground reference circuit of the audio signal amplifying circuitry. On the other hand, the input signal to a power amplifier is defined with respect to the ground reference circuit of the audio signal amplifying circuitry. But, if we want the power amplifier to have high fidelity, we of course want the total actual output signal to be an accurate facsimile of the total actual input signal, so we had better get these two these two different ground reference circuits to communicate with each other very intimately, so intimately that the two ground reference circuits not only are constant within themselves (regardless of time, frequency, and physical location within the amplifier), but also are at exactly the same reference baseline as each other.
      Only if these two ground reference baselines are exactly the same can the total actual audio signal at the output be an accurate facsimile of the total actual audio signal at the input (even assuming hypothetically perfect amplifier processing of the high side of the audio signal). The design engineer has to optimize each type of ground circuit within itself, and then also optimize their mutual communication with one another. The best sounding amplifier designs tend to employ specialized tactics for achieving these optimizations, such as star ground circuit topology, and placing the amplifier's output jack within an inch of the input jack (which may be a bit inconvenient for you the user changing cables, but which yields real sonic benefits by helping to ensure that the output signal is referenced to exactly the same ground reference baseline as the input signal).
      Bringing the ground reference circuit of the shielding chassis into the picture adds further complications, including ground loops and circulating ground eddy currents. But the details of this aspect go beyond what we need for this product review, so we'll skip detailed discussion of this aspect for now.
      All the considerations and complications discussed above pertain even to a single channel monoblock amplifier. Then, as soon as one put two or more channels on a single chassis, there are further new complications that the design engineer has to battle. In a seven channel amplifier (as an example) there are seven distinct audio signals, which should be kept as separate as possible, lest they cause crosstalk or intermodulation distortion with one another. That would imply that it might be better to also keep separate the seven ground reference circuits for these seven channels. But there is likely only one power cord for this seven channel amplifier. Thus, even if there were seven distinct power supplies, one for each channel, they would still be tied together at the power cord ground, so there would be only one common ground reference baseline for the seven power supplies.
      Of course, in most plural channel amplifiers, some or all of the power supply circuitry will also be common to all channels, including power transformer, rectifiers, regulators, and/or capacitor filter banks. Thus, in most plural channel amplifiers, there will be a single common power supply ground reference baseline for all channels. However, this can cause complications. As we saw above, the audio signal ground reference for each channel needs to be intimately tied to the power supply ground reference. We want to keep the seven audio signal ground references separate from one another (to avoid crosstalk and intermodulation distortion), yet at the same time we also want each of the seven audio signal ground references to be intimately tied to its own power supply ground reference, but there is only one of the latter for all seven channels to share. The unfortunate result is that the seven channels can cause crosstalk and intermodulation distortion with one another, indirectly via their common intimate connection to the single power supply ground reference that they all share.
      As you may know, if the drain from a strong signal transient in one of the seven channels causes the high side of its power supply voltage to dip, then that dip can cause automodulation distortion of the audio signal in that same channel, and can also cause intermodulation distortion of the audio signal in the other six channels, if the other six channels depend on the same power supply circuit for the high side of their operating voltages. Well, a similar problem arises for the ground reference circuits. A strong signal transient in one channel could cause its own ground reference to vary, if that ground reference circuit is not optimally designed, thereby causing automodulation distortion within that one channel. And this strong transient could also cause the power supply ground reference to vary, if that circuit is not optimally designed. If the power supply ground reference varied, then it might cause the audio signal ground references of the other six channels to vary, since they are all intimately tied to the common power supply ground reference. Thus, the audio signal in one channel could cause intermodulation distortion in the other six channels, not by the usual route of affecting the high sides of the power supply voltages, but rather by affecting the ground reference baseline of the power supply voltages, and thence the ground reference baseline of the audio signals in the other six channels.
      Again, the audio signal, and its nature and accuracy and fidelity, is defined (like the constructed height of a skyscraper) as the difference between the high point and the ground reference baseline. So, if the ground reference baselines changes (from the absolute constant it's supposed to be), it will distort the total actual signal, as surely as if the high point changes its altitude from what it's supposed to be.
      In our seven channel amplifier example, all seven channels share the same chassis, so there is only one chassis ground reference, and the seven separate audio signal ground references will probably be tied in common to this single chassis ground reference. Thus, this single common connection via the chassis ground also has the potential for causing various sonic degradations, including crosstalk and intermodulation as we saw with the power supply ground, and also including further degradations due to ground loops and circulating ground currents.
      Careful design of all ground circuits can minimize all the above problems in plural channel amplifiers, as well as in single channel amplifiers. There's no such thing as a perfect ground circuit design, any more than there is such a thing as a perfect amplifying stage. All power amplifiers exhibit sonic traits that are attributable to their engineer's attempt (either lax or assiduous) to minimize ground circuit problems.  It's important for you to understand all the above problems so you can appreciate how tough a job the amplifier design engineer has in properly dealing with ground circuit design. Ground circuit design, oft neglected as an afterthought, can be just as sonically significant as the design of the circuitry handling the high side of the audio signal that usually gets all the popular attention. And now you can better understand that, when one amplifier sounds better or worse than another, in certain ways or under certain conditions, it may be due to its ground circuit design rather than the design of its circuitry that handles the high side of the audio signal.
      Which brings us back to the subject of this review, whose seemingly weird behavior you are now ready to understand.
      Evidently, the ground reference circuits of the Adcom GFA-7807 are designed in a somewhat unusual way. Evidently, the ground reference for its output signal (which is of course where you actually hear the amplifier's sound) is effectively intimately tied to the power supply ground reference, but is not so intimately tied to the input signal ground reference. It's easy and natural for this misrouting to happen in a power amplifier design, since, as discussed above, the high side of the output signal of a power amplifier actually comes from the power supply, not the audio signal circuitry, and the power supply ground reference is the reference baseline for this high side of the power supply voltage.
      Thus, the GFA-7807 evidently gets its ground reference for its output signal (the very ground reference that defines the reference baseline for the altitude of its output signal, and hence in part defines the height of its true total output signal) at least in part via the ground lug of the amplifier's power cord, rather than exclusively via the ground side of the interconnect cable and signal input jack. Therefore, if you want the output signal from the GFA-7807 to be an accurate facsimile of the signal coming into the amplifier from your program source via your interconnect cable and through the GFA-7807's signal input jack - in other words, if you want to hear the GFA-7807 sounding its best - then you have to take what reasonable steps you can to make sure that the power supply ground reference of the GFA-7807 is as intimately tied as possible to the signal ground reference of the audio component source feeding signal into the GFA-7807.
      Put simply, the GFA-7807 may get the hot high side of its input audio signal from your source component via the usual expected path, the interconnect cable and signal input jack, but it gets the ground reference for its audio signal from your source component, at least in part, via an unusual and unexpected path, the power cords of both units and via the bridge that is the ground path in your wall between the two power sockets that these two power cords are plugged into. To hear the GFA-7807 at its best, you naturally want this unusual ground reference circuit via the power cords to be as direct and perfect as possible, so that the GFA-7807 is working from a ground reference as a defining baseline that is, as closely as feasible, exactly the same as the ground reference in your source component. That's why we observed such a dramatic sonic difference in the GFA-7807's sonic performance, dependent on where it was plugged in, and why the GFA-7807 sounded far better when the ground bridge between the two power cords was made as short and direct as possible, by plugging both cords into immediately adjacent sockets of the same wall outlet.
      You could improve this ground bridge still further by employing the best possible power cord for your source component feeding the GFA-7807. You cannot easily improve the power cord of the GFA-7807 itself, because it is a captive heavy gauge cord. In theory, you could shorten this ground bridge even more by placing the GFA-7807 next to your source component and running a custom short wire reaching into both components and connecting their two power supply grounds, but this is too dangerous for us to recommend, since dangerous power supply voltages are probably running around near the power supply grounds inside both products.
      As part of the thorough research into this whole grounding issue, we did try placing the two units side by side and running a custom short wire connecting the two chassis ground references. This also changed the sound, and quite dramatically again for the GFA-7807. The dramatic change in itself suggests that the GFA-7807's chassis ground circuit is not well tied into its audio signal ground circuit. In this case, however, the change to the sound of the GFA-7807 was for the worse, not for the better. Connecting the two chassis grounds more intimately, via this short wire, made the GFA-7807 sound like a typical bipolar solid state amplifier, with artificial upper midrange glare. So obviously we recommend that you don't do this.
      Incidentally, as we've discussed in other product reviews, the power cord strongly affects the sound of nearly all audio electronic components. The power cord is in series with the audio signal circuitry, and is in series with the power supply that both supplies the operating voltages for the audio signal circuitry and also directly supplies the output signal of the audio component (the output stage merely acting as a valve for this output signal from the power supply). Thus, the exact electrical parameters of the power cord can have a profound effect upon the sound of every audio component.
      This is doubly true for the GFA-7807, since this amplifier obtains not only its power but also its ground reference for its output signal via this power cord. Thus, the GFA-7807 should be even more strongly affected sonically by its power cord than other audio components are. We could not easily test this directly by substituting power cords (as we have with other amplifiers), since the GFA-7807's power cord is captive. But the fat gauge of the GFA-7807's power cord suggests that its inductance is high, which should make the upper frequencies sound soft, sweet, and defocused. Since that is precisely how the GFA-7807 does sound, we can infer that the GFA-7807 owes much of its wonderful euphony to its power cord.
      The GFA-7807 uses only bipolar transistors, and no FETs, yet manages to deliver a sound that is even softer than that of amplifiers employing FETs. The topology of the GFA-7807 is reportedly balanced, so it does not rely upon the single ended transfer function to achieve its gentle, sweet, tubelike sound as the Plinius Odeon deliberately does. Well then, how on earth does the GFA-7807 deliver such soft, sweet, tubelike sound from a balanced bipolar circuit? A good part of the answer may well be that, due its different grounding circuit schemes, the GFA-7807 is doubly susceptible to its sound being affected by its power cord, and thus the higher inductance (plus other electrical parameters) of its large gauge captive power cord might well be largely responsible for this amplifier's euphonic sound.
      Our suggested crucial tactic, of plugging in the power cord of the GFA-7807 directly next to the power cord of the source audio component feeding the GFA-7807, may vary in effectiveness for all of you, depending on a number of variable factors. For example, some wall duplex sockets will have a better ground bridge than others, and some source audio components will have a more intimate internal connection than others, between their power supply ground reference and their audio signal ground reference. We heard a huge difference in our research lab, and a huge improvement in the GFA-7807 when the two units were plugged into the same wall socket instead of nearby wall sockets, but your mileage may vary. Some surround processors nowadays come with only two prong power cords having no ground connection (including the Adcom processor), so in these cases it might sound better to plug the grounding power cord of say the DVD player (the true source of the audio signal) into the same socket that you plug the GFA-7807 power cord into.
      If your source audio component happens to have power outlets on its back (as some preamps used to), you could even try plugging the GFA-7807 into one of these outlets (try several of them), to get an even shorter connection between the power supply ground references of the two units. Note that this might give you better sound from the GFA-7807 for low to moderate listening levels, but possibly somewhat worse sound for loud levels, since the GFA-7807 would now be facing a higher source impedance (two power cords in series) for its needed current draw from the wall socket at high power levels.
      In practice, no physically real ground circuit scheme can achieve ideal perfection, so all

(Continued on page 126)