Why a Preamp?
Every field of endeavor, including audio, has its conventions: habits of thought so ingrained that they are taken for granted, sometimes in the face of all logic. Most experts would agree that the amplifier/loudspeaker should be designed as a package, yet such integration has fallen flat in the high-end market. Ditto, with a few notable exceptions, the system approach to the turntable, arm, and cartridge. In audio, indeed, the tendency is to more and more separation, rather than greater integration, of functions—witness today's separate CD transports and D/A converters (footnote 1). This inclination may have as much to do with the history and customary use of the preamplifier as with present practical design considerations.
In high fidelity's early days, preamplifiers were often separated from what we now call power amplifiers for very practical reasons: to avoid the heat and transformer-generated hum of the tube amplifiers then in universal use. This was especially prevalent in high-power designs—in those days, anything over 30W (mono, of course). The heat reduced the longevity of not only the preamp stages, but of the power amplifier itself, though the latter's gradual deterioration was unavoidable. The hum problem was even more pressing, especially in phono stages. Up until just recently, remember, for most listeners the phono stage—which is very sensitive to induced hum—was the input most frequently used for critical listening.
Smaller integrated amplifiers continued to dominate the affordable end of the market, later to be largely superseded by the receiver, especially after the advent of FM stereo and the transistor. But the separate preamp soon took over the High End—though no one called it the High End in those innocent times. Putting the preamplifier circuitry in a separate chassis—at first powered most often by the power amp via an umbilical, later by its own internal dedicated power supply, permitting more universal use—eliminated both the heat and hum problems, at least as much as they may have been caused by proximity to the power amp. The preamplifier thus became a fixture in the highest-end equipment, an arrangement which continues to this day.
While the preamp persists as much from custom as from practicality, there is still a method in this madness. While it is certainly possible today to assemble a system without a preamp, there is still a need for one in many, if not most, applications. Ignoring for now the question of integrated amplifiers and receivers, which combine in a single chassis a preamp, power amp, and, in the case of the receiver, a tuner (a more practical arrangement with today's solid-state circuits than with yesterday's tubes), the justification for a separate preamp is today based on issues of gain and impedance incompatibility.
No one will argue that passive attenuation and switching avoids the perhaps unnecessary complication of a preamp-with-gain. The simpler the better, the philosophy goes. Devices for performing this function, known as passive attenuators, or the more appropriately descriptive but nonsensical "passive preamps," have carved out a small niche in the marketplace. But they are not appropriate in all applications. There is no widely adhered-to standard for line-level output voltage or line-level input sensitivity. While many line sources—CD players especially—will easily drive many power amplifiers to their maximum outputs, there are situations in which you won't quite be able to get the level you desire from the system.
I have encountered this with two different "passive" preamps. A preamp with at least a moderate amount of gain will solve this problem, assuming that the amps and loudspeakers are up to the user's demands.
Impedance compatibility is the relationship between the output impedance of the source and the input impedance of the amplifier it is driving. This more complex problem is not unrelated to the gain dilemma described above. It is generally not desirable for the two impedances to be identical, though such a matching might be intuitively appropriate. In fact, if the output impedance of the source is too close in value to the input impedance of the amplifier it is driving, demands may be made on the source for more current than it was designed to deliver, with a resulting increase in distortion.
There will also be a decrease in effective gain as the voltage-divider effect takes place and more of the available source voltage is wasted across the source or output impedance instead of across the input of the power amp, where it is needed. (The situation is actually even more complicated than this; increased current demands may cause the source voltage output to sag.) The system may also become more sensitive to the cable used to connect the source and amp in this situation. An input impedance ten times that of the output impedance of the preceding component is commonly considered a reasonable figure to aim for. Though this is not universally accepted, it will generally minimize such impedance interactions (footnote 2). The impedances of many, perhaps most, of today's sources do match up reasonably well with those of most modern amplifiers. But not all.
If this isn't enough to be concerned with, the varying resistance of the system's level control, whether incorporated in the source or in a passive attenuator (you'll need some way of setting gain), also enters the picture. This level control may be buffered (footnote 3), as in the variable outputs of some CD players and such devices as Corey Greenberg's buffered passive preamp, but in truly "passive" preamps it is not. In addition to gain, an active preamp generally provides a consistent, and low, output impedance independent of the level control setting, relieving these concerns. (In the real world, of course, not all preamps are equally successful at all of these tasks.)
It simply isn't possible to completely ignore, as I was asked to do for the sake of this discussion, such convenience functions as switching. If you have more than one input, you do require switching of some sort—unless you actually enjoy plugging and unplugging inputs. And for those audiophiles who still value vinyl, how can I ignore the LP preamplification stage, which is still most conveniently incorporated into a "preamp"? The switching, of course, can be done by a "passive" preamp, with the above concerns still applicable. But a phono stage requires a more active approach. Here is yet another example of the division of functions. The rise of CD has forced many preamp manufacturers to delete phono stages and, in many cases, market the latter as separate components to those who require them.
Again, in the right system, you may be able to get along without a "preamp." If you have the gain and some means of controlling it, know that your impedances are compatible, and don't mind the potential inconvenience (in a situation where no switching is provided), this system will not be compromised by the lack of a preamp. It may even be the better for it. But most audiophiles still desire a preamp which gives them flexibility and predictable system interactions.
Footnote 1: Cynics have said that this simply cuts the sales pie into more pieces, and perhaps even bakes a bigger pie. But that's another article.
Footnote 2: In the clearest exception known to me, the Jeff Rowland Design Group argues for maximum power transfer by matching input and output impedances. This requires, however, that the source and amplifier be carefully designed to operate in this fashion, something which Rowland preamps and power amps definitely are. Most random source (or preamp) combinations cannot be assumed to be so compatible, making the 10:1 ratio a more universally applicable recommendation.
Footnote 3: Buffering is a design technique used to isolate one part of a circuit and keep it from adversely affecting the stage which follows. It is generally realized by using an active zero-gain stage, the output of which presents a constant low impedance to the next gain stage. A buffer may itself have gain, though we usually don't think of it in this way. An entire active preamp with a low and constant output impedance does, technically speaking, perform a buffering function, with gain.