.

Die Übersicht über die anderen Artikel und vor allem über die Anzeigen kommt bei den US Zeitschriften. Wir beginnen auf Seite 43 von etwa 110 Seiten.

When you walk into any high-fidelity store today (1962) you can buy a standard, production-line amplifier that is actually superior to the most advanced laboratory amplifiers of only a few years ago.

Distortion is measured in fractions of a per cent; frequency-response curves are so flat they can be drawn with a ruler. Because the science of amplifier design has advanced to the point where any competent company can now build a good amplifier, it would seem that the criteria for excellence would be pretty well agreed upon. But this is not the case.

Manufacturers make claims and counterclaims concerning a fraction of a db, a few watts, or a tenth of a per cent (1/10%) of distortion. The whole matter, in fact, sounds at times like high-fidelity hairsplitting (Haarspalterei) and overeager sales promotion. But at the core of the controversy over amplifier quality is an area of problems about which there is much honest difference of opinion.

These differences grow out of the fact that amplifier design is as much of an art as it is a science (also mehr Kunst als Wissenschaft). This is largely because every amplifier must ultimately be judged by the sound it produces. And this is a subjective, not objective, evaluation, and thus is virtually unmeasurable (nicht meßbar). The final determination of an amplifier's quality must be made by a human ear, not by a meter.

What makes the difference between an amplifier that is merely good and one that is demonstrably better? There is no easy answer to this question, but it is possible to draw some guidelines from which useful conclusions can be inferred.
.

An amplifier has been defined as being a wire with gain. It should amplify any electrical signal that is fed into it without changing it in any respect except size. Of course, there is no such thing as a perfect amplifier. Tubes, transformers, and other components invariably add or subtract something. These changes imposed on the original signal are called distortion, and all amplifiers distort to a greater or lesser degree.
.

An amplifier is subject to many types of distortion. One of the most common is harmonic distortion. A note on the violin, for example, is composed of a fundamental tone and a number of overtones, or harmonics. The frequency of the fundamental tone determines the basic pitch, and is the number of times the string vibrates each second.

An "A" has a fundamental frequency of 440 cycles, or vibrations, per second. But while the string as a whole is vibrating at 440 cps, parts of it vibrate independently at twice that rate, at 880 cps. This is the second harmonic. Still shorter segments vibrate at 1320 cps (three times 440), thus forming the third harmonic, and so on through a dozen or more harmonics. The relative intensities of the harmonics vary. Some are strong, others are weak. But all of the harmonics blend together to form a composite waveform, and the pattern of this waveform is what gives the violin its special tone quality, or timbre.
.

A clarinet's "A" has a fundamental and harmonics of the same frequencies as does a violin's, but its harmonics have different relative intensities. These differences account for the tonal differences between the two instruments.

If a violin's waveform were to be reproduced perfectly by an amplifier, its fundamental and all of its harmonics would come through exactly as they went in. But an amplifier produces a certain amount of harmonic distortion; that is, it generates harmonic frequencies of its own during the process of amplification. These harmonics add to or subtract from those that are already present. In either case, the waveform comes out altered. If it is changed enough, it is conceivable that a violin could come out sounding like a clarinet, or perhaps vice versa.

In general, second-harmonic distortion is less noticeable than third-harmionic distortion. This is because the second harmonic is precisely one octave higher than the fundamental and thus is musically related to its fundamental. The third harmonic bears no such octave relationship.

All high-fidelity vacuum-tube amplifiers, incidentally, have push-pull output circuits that tend to cancel amplifier-produced even-numbered harmonics. Unfortunately, third-harmonic distortion, which is more objectionable to the ear, is not affected.

Intermodulation distortion, another type of distortion, is also caused by imperfections (Unzulänglichkeit) of the amplifier. In various parts of the circuit, one tone tends to interact with, or modulate, the tone of another instrument.

The cello interacts with the flute, the drum with the oboe, and so on in an increasingly complex pattern. This kind of distortion is usually more obvious and annoying than is a like amount of harmonic distortion.

This is because intermodulation signals have no relationship to the musical frequencies being amplified. IM distortion, therefore, may sound like a raucous buzzing, or it may add an unpleasant fuzziness to the music.

FREQENCY distortion, still another shortcoming, is caused by an amplifier's inability to respond to all frequencies equally. For example, if a violin is played in a barrel, some of its tones and overtones resonate at the barrel's natural resonant frequency and come out louder than others.

Similarly, if an amplifier fails to amplify all frequencies exactly the same amount - if it doesn't have a so-called flat frequency response - it in effect can turn into an electronic barrel.

A signal being amplified can also be distorted by noise that is generated within the amplifier. A small amount of hum, sizzling, and crackling is present in any electronic circuit, so some amount of noise distortion is inevitable. Happily, this form of distortion is not serious in quality amplifiers. Current units frequently have noise levels that are completely inaudible even while listening at low levels in a quiet room.

A more subtle form of distortion is brought about by a peculiar characteristic of the human ear. Many years ago it was discovered that at low volume our ability to hear low-frequency sounds falls of substantially. A balanced orchestral recording appears to have less bass when it is played back at a low level. The Fletcher-Munson curves, named after the audio researchers who measured this effect, illustrate this relationship graphically. These curves have great importance for the music listener, since records are rarely played at full orchestral level. Therefore, the bass instruments will almost always seem softer than they actually are.
.
This loudness distortion, which is in our ears rather than in the amplifier, is quite easy to correct. Simply turning up the bass control will usually make the sound seem balanced. Or the so-called loudness control that is provided on many amplifiers automatically compensates for the Fletchcr-Munson effect by increasing bass response in proportion to the decrease in volume.

Although modern-day amplifiers all contain distortion of many kinds, the best ones reduce distortion to negligible, or nearly negligible, proportions. This raises the question, of course, of what is to be considered negligible.

Twenty years ago (1942) it was generally agreed that any amplifier with less than 5% harmonic- or intermodulation- distortion was, for all practical purposes, distortionless.

5% distortion means that spurious signals generated by the amplifier can contain up to one-twentieth as much power as the original signal. There was a great deal of controversy over this figure until the 1940's, when Dr. Harry F. Olson of RCA decided to settle the question.

Olson got the best amplifier available at the time - one that produced about three-tenths of one per cent distortion, a very acceptable figure even today - and set it up as a reference system. He also installed a variable-distortion amplifier and a set of high-frequency filters that allowed him to eliminate any signal above 3.000, 5.000, 7.500, 10.000, or 15.000 cps, depending on the setting of the filter. Then, to a test audience, he played the systems and introduced varying amounts of distortion. Each time he let the audience compare the sound with his low-distortion reference signal.

His results led to new standards. In general, the higher the frequencies that were reproduced, the less distortion the subjects could tolerate. With the full-range system seven-tenths of one per cent distortion was discernible, while with the restricted-range system more than twice that much could not be detected.

Olson shattered the five-per-cent-distortion standard, but the matter didn't end there. Today, with even wider-range equipment common, many experts feel that more than a few tenths of one per cent distortion is noticeable. Manufacturers of the best equipment try to limit both harmonic and IM distortion to this low level, and even lower, if possible.

The first hurricane of disagreement in this area was set off in 1945, when Howard A. Chinn and Philip Eisenberg of CBS decided to find out how much frequency range people preferred. They set up one wide-range system (40 to 10.000 cps), one medium-range system (80 to 7.000 cps), and one restricted-range system (180 to 4.000 cps). Then they played the same recorded music for a group of subjects over each of the three systems.

The results have been upsetting to the high-fidelity industry ever since. An overwhelming majority of the listeners liked the restricted-range system best, the medium-range system next, and the wide-range system least of all. People, the results seemed to say, are anti-high fidelity.

Why? No one knows for sure, although there have been many explanations. Chinn and Eisenberg's equipment may have had distortion, so the tests may have actually only confirmed Olson's finding that distortion is more objectionable on a wide-range system. Or the loudspeakers used may have been inadequate.

The matter has never been fully settled, but more light was thrown on the subject a few years later when Roger E. Kirk of Ohio State University duplicated the Chinn-Eisenbcrg experiment but added some new twists of bis own.

At first Kirk's results were essentially the same as those of the two CBS experimenters. Then he divided his subjects into three groups. For six weeks he had one group spend considerable time listening to the wide-range system, another to the medium-range system. The third group did no test-listening at all. Then the groups were given the original tests over again.

This time the results differed sharply. The group that had been listening to the wide-range system now liked it best; the group that had grown accustomed to the medium-range system preferred it; and the third group had not changed its opinion. From this Kirk concluded that people prefer the familiar, and that it apparently took an educated ear to appreciate high-fidelity sound.

Can we then conclude that most people would like live concerts better if somehow orchestras could eliminate the high-frequency components of the music ? The answer to this was supplied by another ingenious experiment, this one by Olson.

Olson put a test audience in a room with a live orchestra. He separated the listeners from the musicians with a screen so the audience would have no way of knowing whether the music they were hearing was live or recorded. The screen was a frequency-selectivc acoustical device that could be adjusted either to let all sounds through or to remove frequencies above 5.000 cps. By a ratio of two to one, Olson's audience preferred full-frequency sound to restricted-range sound.

This experiment lends weight to the theory that amplifier designers should attempt to build units that will reproduce everything the ear can hear. Some designers contend that even this is not enough. For example, A. Stewart Hegeman, the designer of Harman-Kardon's Citation line of amplifiers, feels that an amplifier should be able to reproduce frequencies far beyond the audible range. Hegeman says that transient response - the reaction of the amplifier to sharp percussive effects such as drum beats - depends on extra-wide-range performance. In order to reproduce transients at a fundamental frequency of 20,000 cps, says Hegeman, an amplifier must be capable of handling at least the tenth harmonic of 20.000 cps. or 200,000 cps.

Other designers differ sharply. Fred Mergner, chief engineer of Fisher Radio, believes that the upper frequency response of the amplifying system should be limited to about 20,000 cps. He is careful to point out, though, that "the response of the preamplifier alone should be restricted to eliminate extraneous noise signals above and below the audible range. On the other hand, the power amplifier should be designed for the widest possible frequency response, consistent with other design considerations, to provide the best possible transient response and highest stability."

Mergner goes on to say that "although wide-hand design of both preamplifier and power amplifier would theoretically be the ideal, a realistic designer must start with the basic fact that present-day program sources, such as records and tapes, contain a significant amount of high- and low-frequency noise, which, if allowed in pass through the amplifying system without attenuation, would result in needless distortion, overloading, and reduced signal-to-noise ratio.

"Further, as is well known, the transmission of stereo FM programs covers an audio bandwidth from 50 to 15.000 cps. Both the pilot carrier of 19 kc and its second harmonic of 38 kc plus the L-R sideband modulation, must be prevented from reaching the audio sections of tuners or amplifiers. This will improve the signal-to-noisc ratio and eliminate the possibility of audible whistling tones, especially during tape-recording.

Considering that the necessary attenuation should be approximately 30db between the highest audio frequencies of 15.000 cps and the 19,000-cps pilot carrier, it can easily be seen that a very steep filter is required. Even if the filters affect the transient response of high audio frequencies to a degree," Mcrgner concludes, "their insertion is still a better solution to these problems than is a wider frequency response, which can create whistle tones and a lower signal-to noise ratio."

Saul Marantz, the president of Marantz, Inc., says his amplifiers are designed to cut off slightly below 100.000 cps. His reason is that amplifier instabilities - parasitic oscillation, blocking, and other ills - can occur above that frequency and spoil the high-frequency signal. These troubles, says Marantz, "are caused by the multitude of phase shifts present in all output transformers. Every winding has its own phase characteristics and causes problems, even up in the megacycle range. Maximum stability is vital to clean sound quality, and we are willing to sacrifice some frequency response for the benefits of absolute stability. High-frequency instability may not show up when an amplifier is tested with a resistive load (ohmsche Last), but with the reactive load furnished by a loudspeaker, it becomes a real problem."


Hegeman's position on high-frequency distortion of program material is that preset bandwidth limitations should not be built into either amplifiers or preamplifiers. "I agree that poor program material will sound bad, perhaps worse, on a wide-band system," he says. "But that's why low- and high-frequency filter controls are put on preamplifiers.

I can't accept an equipment designers estimate of how poor programs are going to be. This belongs under the listener's control."

There is also disagreement on how far an amplifier's low-frequency response should extend. Some manufacturers feel the bass response should be cut off at about 10 or 15 cycles. "We can't hear anything down there," says Mergner. "Response in this range can't do any good, and it may do a lot of harm. Take an eccentric record, for example. Its off-center track can generate a strong signal at two or three cycles as the record goes around. You can't hear a sound that low, but it can take most of the amplifier's available power to reproduce it. There isn't much left for the music."

Again, Hegeman disagrees. "Subharmonics several octaves below 20 cycles are of great importance to musical perception, and they unquestionably contribute to more realistic and more transparent reproduction," he says. "Amplifiers that offer a frequency response extending at usable power levels to below five cycles have a tight and clearly defined low-frequency response in the audible spectrum.

This is particularly noticeable in the region from 40 to 100 cps. This subsonic characteristic improves the amplifier's ability to damp [control] speakers, even those tending to sound muddy, and this improvement of the low frequencies is distinctly audible." Marantz's design goals are for amplifiers that deliver full power as low as 10 cps, and good frequency response down to about 2 cps.

Although there is general agreement that distortion must be low and frequency response wide, each designer has his own opinion as to "how low is low" and "how wide is wide". There is still a long way to go in establishing the correlation between the objective measurement of amplifier characteristics and the subjective results of these characteristics on the listener.

The fundamental measurable qualities discussed here only begin to tell the full story. There are other capabilities and characteristics that are less well understood and even more controversial. No standardized method of measuring or evaluating them has been devised, but they profoundly affect amplifier performance. Next month, some of these elusive qualities will be explored. (To be concluded next month)

.

Ken Gilmore designed and built his own amplifiers back in the early postwar period when the engineering concepts of high fidelity were just taking shape. He has kept abreast of the challenges of audio ever since. His last contribution to Hifi/Stereo Review was "A Star is Made" (August. 1962) describing current popular-recording techniques.
.

.

Among these hard-to-measurc qualities are transient response, power-supply recovery ability, clipping characteristics, phase shift, and other related phenomena. Transient response, for example, is the measure of an amplifier's ability to respond faithfully to a sharp, percussive sound such as a drum beat, a cymbal crash, or the click of wood blocks. These sounds have abrupt beginnings and endings and are a severe test of an amplifier's ability to respond instantly to high-intensity signals.

Like most of the other phenomena to be considered here, an amplifier's transient-response characteristics can probably be judged most accurately by viewing its output waveform on an oscilloscope. But while an engineer can get a good idea of an amplifier's transient capability by examining its waveforms, it is impossible to translate this into numerical terms.

The manner in which an amplifier overloads is another important measure of its quality. Every amplifier is occasionally called upon to amplify signals that are actually too large for it to handle, particularly if it is driving a low-efficiency loudspeaker. The way it performs during these momentary overloads has an important effect on the way it sounds.

A good amplifier in effect refuses to amplify whatever part of a waveform is beyond its capacity and reproduces everything else accurately.

Edgar Villchur, President of Acoustic Research, Inc., and designer of the AR loudspeakers, describes the difference between amplifiers that distort badly during overload and amplifiers that clip cleanly this way :

"Certain units stay clean as a whistle up to their rated power, and then go to pieces, splattering all over the place and making a horrible noise, while other units overload gracefully."

What, precisely, is clean clipping? The best clue is obtained by driving an amplifier with a sine-wave input until it goes beyond its overload point, then viewing the resultant waveform on an oscilloscope. If the amplifier clips cleanly, it will simply slice off the top and bottom of the sine wave (see Figure 1A). A less desirable type of clipping will distort the waveform in other ways, such as those shown in Figure 1B.

Oscilloscope tracings can give a fair impression of an amplifier's overload-handling capacity, but the point at which an amplifier stops being good and starts becoming bad is hard to determine, especially in view of the fact that there is no way to express clipping characteristics in terms of number.

Also, the audible differences between an amplifier, that clips cleanly and one that does not, is difficult to express in anything but generalities. There may be little or no apparent difference so long as both amplifiers are playing at low levels.

At higher volumes, however, the differences become audible. The unit that clips cleanly will produce a reasonably pleasant sound, while the one that does not will produce a rattling, buzzing, or generally fuzzy sound.

Thus, even though the two amplifiers may be rated at the same power, the clean unit will appear to have greater power-handling capability.
.

• Anmerkung : Mit dem QUAD 405 (2 x 110 Watt Sinus an 8 Ohm) hatte ich leider schlechte Erfahrung machen müssen, denn bei der kleinsten Übersteuerung gab es riesen Knackse und Schläge in den Boxen, daß die Tanzschüler dachten, die Boxen fallen jetzt von der Decke.

.

Phase shift is another characteristic that affects sound. It is brought about by the fact, that an amplifier is called upon to reproduce many different frequencies simultaneously. An amplifier with phase shift - and all have it to some extent - amplifies signals of some frequencies slower than others. For example, all of the different harmonics produced by a single instrument might be amplified in their proper loudness proportion, but, because some frequencies might be passed through the amplifier at a slightly slower speed than others, they might not end up in exactly the same phase relationship in which they started out. Phase shift is
usually negligible in the mid-range and greatest at the extreme low frequencies and high frequencies.

The most serious effects of phase shift are on the amplifier's negative-feedback circuitry. With negative feedback, a small amount of the output signal of the amplifier is constantly being routed back to the input.

Because the feedback signal is exactly 180 degrees out of phase, or, out of step, with the incoming signals, distortion introduced in the amplifier is largely cancelled out. Of course, the total gain of the amplifier is also reduced at the same time.

The trouble begins when excessive phase shift occurs. The signal, that is being fed back, begins to be in the wrong phase relationship to the input signal, and can begin to add to, rather than cancel, part of the input.

This causes an increase in distortion, and the amplifier can go into oscillation at certain frequencies - usually above the range of hearing. This disturbance, called ringing, can make the sound muddy and unclear.

Phase shift can cause other types of signal degradation that are harder to pinpoint. They stem from the fact that the various harmonics of, say, a clarinet, are shifted in phase by differing amounts. In a reproduced clarinet tone, then, the various harmonics can hear different phase relationships to the fundamental and to each other than they did in the original tone.

Amplifier designers differ in their evaluation of the importance of this kind of distortion. Some feel that the ear compensates for it, regardless of the distortion of the original phase relationships. Others feel that the change, while subtle, is clearly discernible and deleterious.

One describes the sound produced by phase shift as being "constricted, two-dimensional. A stereo record played through an amplifier with phase shift makes the music sound as though it were coming from a two-dimensional plane along the wall. Without phase shift, the sound opens up, has depth and realism."

Phase shift obviously must be kept to a minimum if an amplifier is to retain the advantages of negative feedback and avoid ringing and other troubles. Ringing can be spotted easily by sending square waves through the amplifier and observing the output on an oscilloscope. It shows up as a series of jagged spikes on the horizontal part of the wave (see Figure 1C).

How do you eliminate ringing ? - Some amplifier designers simply reduce the amplifier's frequency response. If ringing takes place at 40,000 cps, for example, one way to suppress it is, to limit the response to a frequency below that where the ringing occurs, 35,000 cps, perhaps. Designers who don't want to sacrifice frequency response go to great lengths to reduce phase shift and the subsequent ringing.

In recent years engineers have become more and more concerned with what are called an amplifier's recovery characteristics. In other words, can a unit take a tremendous cymbal crash, for example, which completely exceeds its output capabilities, and recover quickly enough so that the musical passage immediately following the overload point is not affected ?

The key circuit here is the amplifier's power supply. With the introduction of silicon-diode rectifiers and the use of larger filter capacitors, power supplies have improved considerably. Silicon diodes, now used in many amplifiers, contribute to improved operation because they tend to maintain a stable output voltage under all conditions. The reason for the larger capacitors is, that power-supply capacitors act as electrical storage tanks, and can provide power from their stored energy for momentary overloads. The more their capacity, the more they can help during periods of stress.

Any discussion of overload recovery must be related to the entire question of power output. And this brings up one of the oldest battles over amplifier unmeasurables. Power output itself, of course, is simple enough to measure. The trouble comes in deciding just how much power capability is necessary for good performance.

Good overload characteristics are certainly important, but how much power should an amplifier be capable of producing before it reaches an overload condition?

Although there is a diversity of opinion about this, a clear trend toward higher-power amplifiers has been emerging in the last decade or so.

Twenty years ago, it was commonly agreed that a good 10-watt amplifier produced adequate power for home listening under any conditions. Now, especially with today's low-efficiency loudspeakers, most designers say that 50 or 60 watts per channel is not too much.

Efficient speakers probably require amplifiers of less power, perhaps in the 20- to 30-watt range. Fred Mergner, chief engineer of Fisher Radio, and Saul Marantz, president of Marantz, Inc., for example, while emphasizing that power requirements vary with program material, room acoustics, speaker efficiency, and other factors, agree in general with these figures.

However, Marantz adds, "But there is no doubt that greater power improves the sound. Transients, particularly, benefit from greater power. They're cleaner, crisper, with a high-power amplifier."

On the other hand, A. Stewart Hegeman, designer of Harman-Kardon's Citation amplifiers, emphasizes the importance of an amplifier's ability to develop appreciable power below 20 cycles and above 20,000 cycles.

"These are the important areas for the preservation of transient information," he says, "Amplifiers of surprisingly modest ratings that have wide power bandwidths outlisten higher-power amplifiers that have restricted power bandwidths. Thus the mid-frequency power rating of an amplifier is not a true indication of its capabilities. As one would expect, higher-power amplifiers that have equivalently wide power bandwidths seem to listen somewhat better and cause less listening fatigue. However, power-handling capabilities of loudspeaker systems and acoustical requirements of listening rooms do not call for a 'horsepower' race,"

A few years ago, the Institute of High Fidelity Manufacturers agreed upon a new way of measuring output power called the music-power rating system. Theoretically, the music-power rating indicates an amplifier's performance under momentary loads. The unit's steady-state power output -that is, its maximum power with a sine-wave input signal - will be somewhat less than the power it can deliver in short bursts. The justification for the music-power rating is, that it more truly reflects an amplifier's capability for reproducing music - the job it is designed to do - than does the old system.

The battle over music power has gone on for some time now, and it will probably continue. But whether the music-power rating system is valid or not, it does tell something useful about the amplifier.

A comparison of a unit's music-power rating to its continuous-power rating reveals a great deal about its power-supply regulation. If the two ratings differ by a large amount - 10 per cent, perhaps - then the power-supply regulation is probably not as good as it should be. The closer together the two ratings, the more stable and better-regulated the power supply. This undoubtedly affects some of the hard-to-evaluate factors discussed earlier, though it would be difficult to say how much in any given case.

Even though much information concerning an amplifier's performance cannot be reduced to a table of specifications, a specification sheet can provide much information. Unfortunately, though, a list of statistics that seem impressive at first glance can turn out on closer examination to be virtually meaningless.

The specifications for a hypothetical amplifier, for example, might read something like this;

"Power output 30 watts at 0.5 per cent distortion. Less than 0.1 harmonic and intermodulation distortion at normal listening levels. Frequency response flat ±1db from 10 to 45,000 cps."

By common understanding, "normal listening level" has come to mean an output level of one watt. This one watt, when delivered to an efficient loudspeaker, will make quite a loud sound, and average listening levels probably do not exceed this.

The rub here is the word "average." Loud passages or transients such as drum beats can be many times that one watt. An amplifier's performance at one watt, therefore, does not tell what will happen when these louder signals are encountered. Also, the figures regarding the hypothetical amplifier's low distortion may be misleading, If the 0.1 per cent distortion refers to a measurement made at 1,000 cps, where the unit's performance will he best, its performance with a high-level signal at either frequency extreme could be very poor.

Suppose, for example, the amplifier will produce only a few watts at 50 cps and its distortion has risen to 5 or 10 per cent, a not unusually high figure for less-expensive amplifiers. This combination of circumstances might lead to a situation where a low-frequency signal might he heard hardly at all, while its harmonics, representing distortion, would muddy the amplifier's audible range. In spite of the fact that this amplifier's quoted specifications seem reasonably good, it could produce a very unpleasant sound.

Clearly, a fuller statement of the amplifier's operating characteristics is needed. To be of much meaning, the specifications should tell something of a unit's power response over a broad band ol frequencies.

That is, the figures should tell how much power the amplifier is capable of producing over the entire audible range and beyond. Distortion figures for the entire audible band should be given, too.

A good amplifier's specifications, then, might read something like this:

"Power response 35 watts flat to within one db from 20 to 25,000 cps. Harmonic and intcrmodulation distortion less than 0.5 per cent at full rated power output from 20 to 20,000 cps."

An instrument with these specifications would probably be a superior, clean-sounding amplifier.

In general, a good amplifier would have the following characteristics. Its harmonic and intermodulation distortion would be low, its frequency response flat, its internally generated noise inaudible, its transient response good, its overload recovery good, its phase shift small, and its power output over the audible range sufficient to drive the speakers connected to it. It should clip cleanly, and it should not ring or otherwise become unstable under any conditions.

Obviously, most of these qualifications are imprecise. At this time, there are virtually no agreed-upon standards governing just how these characteristics can enable one to differentiate an outstanding amplifier from a good one.

Once a unit meets general levels of good performance, the rest must be decided by the listener. Fortunately, he is equipped with an instrument superbly suited to make a qualitative judgment of this kind. His ear, a sound-analyzing device of exquisite refinement, is able to detect differences that are measurable in no other way. In the final analysis, then, once the measurable minimum requirements have been met, the listener must rely on his ear to determine when an amplifier sounds good.

• Ken Gilmore studied music in college and electronics in the U.S. Navy. He has been a symphony-orchestra player, a broadcast engineer, and a television announcer, and is currently a successful free-lance writer in New York City.

OCTOBER 1962
.