from PAUL W. KLIPSCH - 1951

Unlike other equipment used in high-fidelity installations, the development of the corner speaker has been virtually a one-man job, carried on over a period of years by Paul W. Klipsch at his laboratory in Hope, Arkansas. We asked him for his ideas of equipment to use with the Klipschorn, so that optimum results could be achieved. Here is his full and very interesting reply.

ORIGINAL generation of sound requires large generating sources for bass tones, and relatively smaller sources for treble tones. Compare, for example, the piccolo with the tuba. The piccolo is an instrument only a few inches long, of less than 1 inch bore, playing a tonal range corresponding approximately to the top three octaves of the piano. The tuba plays in the bottom octaves of the tonal range, and is made up of nearly 20 ft. of pipe with a bell or mouth nearly 3 ft. in diameter. Compare the short and long strings of the piano; compare the small and large pipes of the organ.

The bottom note of most pipe organs is the third octave below middle C, and is referred to as a 16-ft. tone. This means it is a tone which would be generated by a resonating open pipe 16 ft. long. As we go up one octave, we find the corresponding pipe only 8 ft. long, another octave 4 ft., and so forth.

This leads to the concept that the lower the pitch of a tone, the longer its wavelength. Actually, physicists have known for a long time that the wavelength produced by an open organ pipe is about twice the length of the pipe (exactly twice the effective length of the pipe). The fact that it requires a certain length of pipe to produce a certain pitch leads naturally to the thought that producing deep bass tones requires a certain amount of size of tone-generating medium.

The same principle applies to a loudspeaker. Trying to produce any appreciable amount of sound power in the deep bass register with a small speaker is like trying to produce organ pedal tones with a saxophone. To carry it to extreme, try feeding a 6-in. cone speaker with electrical power derived from a big organ pipe, picked up by a good microphone and fed through a wide-range amplifier. The result will be merely a rattle. To produce a loudness at the ear corresponding with what one would hear at an organ recital, a 6-in. speaker cone would have to travel a foot or more to pump enough air to generate the deep bass tones. It is apparent that the 6-in. speaker is not an adequate device to reproduce the bass range in music.

But merely increasing the size of a cone speaker is not the answer to producing the bass range. True, one might build a cone about 3 ft. in diameter and get some bass, but such a large cone would fail at the treble end of the scale. Hence the concept arises of using several cones of different sizes to radiate appropriate parts of the complete tonal range.

But the cone itself presents numerous peculiarities. It is impossible to make a large cone stiff enough to act as a rigid piston over a wide tonal range. Cones have some weight and inertia to motion. They must be suspended in such manner that they can vibrate, and the suspension therefore entails compliance or springiness.

The combination of weight and springiness gives rise to the phenomenon called resonance, which makes the loudness of some tones greater than others. Furthermore, second-order resonances, generation of tones not present in the electrical power fed to the cones (distortion), tendency toward directivity so the loudness is different at different directions from the speaker - all these put the cone at a disadvantage in reproducing sound. Further, the cone speaker is inefficient; that is, it transforms only a small part of the electrical power received into sound power, the rest being lost as heat.

The horn speaker overcomes many of the limitations of the open cone or direct radiator speaker. *1)

*1) Note the distinction between an open-cone speaker - one in which the cone radiates directly into the room from a simple baffle board, and the horn type, in which the cone (or diaphragm) feeds through a horn or other enclosure.

By loading a diaphragm (usually a more or less conventional cone speaker) by means of a horn, the diaphragm can be made small enough to retain rigidity, while the horn performs the function of a ratio gear to transform low velocity cone motion at high pressure to high velocity at low pressure, with the result that a 15-in. cone speaker, when horn-loaded, becomes the equivalent of a 6-ft. cone for bass radiation, and at the same time makes efficient radiation possible over a range of several octaves.

And note the expression efficient radiation. The horn may permit an increase in efficiency from the 5% typical of an open cone speaker to 50%. This is a fact not to be overlooked, because it can be demonstrated that as efficiency is increased distortion is decreased.

Doubling efficiency reduces distortion by a factor of 1/4, increasing efficiency by a factor of 10 decreases distortion by a factor of 1/100. It also reduces the necessary power rating of the amplifier associated with the sound-reproducing system. Since as much as 800% (Anmerkung : 800% ??? Wie haben die das damals gemessen ?) distortion has been measured in a good speaker, it can be appreciated how important reduction by a factor of 1/100 would be.

Unfortunately, the complete range of tones audible to the human ear encompasses about 9 octaves, and so far, horn design has not achieved better than about a 5-octave range without loss of the other desirable features of a horn, such as low distortion and high conversion efficiency. Hence, two or more horns are necessary to cover the entire audible tonal range.

Theater speakers for sound movies are horn speakers. These are bulky systems occupying from 100 to 250 cubic feet or more. Obviously, they are not something for the home! But as early as 1930, work was being done with corner location of speakers, whereby the physical size could be reduced without sacrificing bass range.

Dr. Kellogg described a horn speaker for corner use in the technical literature of 1930, and Mr. Sandeman was issued a patent in 1934 on a corner horn speaker. These ideas offered a vast increase in bass range per cubic foot of speaker size, but they were not exploited for home use; indeed it is doubtful if they would have been practical for home use in view of the peculiar shapes involved.

In 1940, the writer (Anmerkung : das war Paul Klipsch) undertook a new approach to the corner horn and eventually developed a speaker system to cover the tonal range down to the 16-ft. organ pitch, small enough to fit in a living room. During the last few years, this bass range has been extended below the limits of hearing and available measuring devices without requiring more than 16 cubic feet of occupied space for the main horn.

A second small horn is added to cover the upper tonal range to a point beyond the limit of human hearing. Refinements have reduced distortion even at loud listening levels and have resulted in a uniform response without resonances. There being no resonances in the speaker system, the resonances of the original sound are preserved. This system was facetiously dubbed the "Klipschorn" *2) by some experimenters in New York. The name stuck and was adopted as a trademark.

*2) For those interested in the technical aspects of the Klipschorn and the theory relating to its design, an extensive bibliography is available in a paper describing it, "A High Quality Loudspeaker of Small Dimensions", Journal of the Accoustical Society, January, 1946.

With the development of a speaker system of wide range and low distortion, it was realized that the better the reproducer, the more it brought out the sources of noise and distortion fed to it by the driving amplifier. At each stage of development of the speaker, it was found necessary to work on amplifiers, phonograph pickups, radio tuners, and the like before advantage could be taken of the extended tonal range of the speakers.

Whereas 5% distortion was the accepted standard back in 1930, amplifiers are now being built with a fraction of one percent distortion, and the listening pleasure derived from such a modern amplifier is vastly greater than from amplifiers that were regarded as good a dozen years ago.

Today, the art of sound reproduction is entering on a new era. The means of sound reproduction are approaching perfection. But just as much of the equipment in use today falls below this standard, so too, many recordings fall below present day potentialities of the recording art.

To digress into the quantities again for a moment; the 16-ft. pitch corresponds to 32.7 vibrations per second, or 32.7 cycles. This is the third C below middle C. The second C below middle C would be 65.4 cycles.

When it is pointed out that recording practice in this country is largely standardized on response down to 50 cycles, it is realized that a sizable chunk is omitted from the bottom octave. A few records, probably recorded under what were considered substandard conditions, retain the original depth of the pipe organ.

Two notable examples are Columbia 7356-M (recorded in England) "Gypsy Princess" and "The Merry Widow" played by Sidney Torch on the organ in the Gaumont State Cinema, Kolburn, England, and Columbia Gramaphone ROX 149 "Prelude and Fugue in G Major" (Bach) played by Albert Schweitzer, recorded at All Gallows, Barking by the Tower.

These releases, by accident or design, appear to have retained the powerful sub-bass bourdon tonal depth which one associates with a pipe organ, and which one senses as much by feeling as by hearing. As one listener inelegantly, but effectively, expressed it: "It shakes your stomach."

Radio transmission (Anmerkung : Damals 1951 in USA war das immer noch AM, also Mittelwelle) suffers the same lacks as do the records. Network lines appear still to be limited at the low end to about 100 cycles even though the upper end has been extended in some instances to 8,000. This represents about one octave short of the limit of hearing at the upper end, and nearly 1 octaves short of what would be considered a truly full-range bass. Even live programs of local origin are most likely to be deficient in bass range due to inadequate microphones and associated apparatus.

Hence our new era has been reached only to the extent that the apparatus is now available for a full tonal range, and that some few - very few - sources of program material represent a near approach to the ideal. With the increasing number of high-quality home installations, it is fervently hoped that more program material, in recorded and broadcast form, will become available for full appreciation.

In the previous section it was pointed out that the inclusion of 16-ft. organ tones requires a lot more speaker than the simple direct radiator paper cone units. It was mentioned that the 16-ft. C3 (3 octaves below middle C is a frequency of 32.7 cycles per second. If the upper limit of human hearing is regarded as 16,000 cycles per second, there are approximately 9 octaves of tonal range to be covered by a speaker system.

Every time developments permitted adding an octave to the speaker reproduction at either end of the tonal spectrum, problems of noise and distortion had to be solved all over again. In the development of the Klipschorn, which covers the full 9 octaves, as much work was done on amplifiers and other associated apparatus as on the speakers. It is the purpose of this study to point out the requirements for apparatus associated with a speaker system of truly wide range and low distortion.

Let it be said there is no advantage in increasing the tonal range of the speaker, and decreasing its distortion, if remaining apparatus is of restricted range and high distortion. To repeat: The better the speaker, the more it emphasizes the deficiencies of the associated apparatus.

A dozen years ago, the typical advertising claims of better amplifiers were directed to the fact that from 30 to 60 watts of power were available. Distortion, if mentioned at all, was claimed to be some stipulated amount, without indicating at what frequency the distortion was measured, and sometimes without mentioning the power output at which such distortion would occur. Distortion measurements were often made at 400 cycles where any amplifier, however cheap, would probably give a good account of itself. But it happens that the lower the frequency was, the greater was the distortion in any amplifier. This is due to the all-important but necessary evil, the output transformer, where large amounts of distortion may be generated at low frequencies. Hence, to be valid and useful to the owner of a full-range speaker, the distortion and power rating at some low frequency like 30 cycles should be specified.

In the days of the 30- and 60-watt amplifiers, speakers of the direct-radiator class were typically 5% efficient; that is, 5% of the power fed to the speaker electrically would be converted to sound power. Hence a 30-watt amplifier driving a typical speaker of 5% efficiency would deliver about 1.5 acoustic watts maximum, and to hold distortion to tolerable levels would require backing down the gain control to keep the electrical power output of the amplifier down around 10 electrical watts, or 0.5 watt of acoustic power.

Now that wide-range low-distortion speakers exhibit around 50% efficiency, less electrical power is needed. An amplifier rated at 10 watts may deliver 5 watts acoustic power, and if one needs only 0.5 watt of acoustic power, the amplifier can be operated at only 1 watt output with extremely low distortion.

The favorite amplifier of 1940 consisted of a pair of 6L6 or 6V6 "beam power" tubes of high efficiency but unfortunately - high distortion even at low power output levels. These amplifiers put out tremendous power, and were cheap to buy. But they would not serve in a wide-range sound system because of the intolerable distortion. By adding feedback to such systems, distortion was reduced, but even with large amounts of feedback, resulting in measured distortion of very low level, it appears that there remained some almost immeasurably small high-order distortion which could not only be heard but proved irritating to listeners, whether of the golden ear or tin ear category.

This called for the expenditure of much effort on development of amplifiers of inherently low distortion. Triode tubes instead of pentode and beam tubes were chosen, and then, to bring distortion to vanishingly low levels, feedback was applied to them just as it was to the beam tube systems. Output transformers, previously mentioned as a necessary evil, were studied, and now have been developed to the point that they no longer constitute the weak link.

Very frequently, the author is asked: "What amplifier do you suggest as being able to give maximum performance from my Klipschorn?" That is the type of question an engineer answers with reluctance, not only because it is impossible to label any one make as best, but because most of the manufacturers are doing an admirable job of continuous improvement.

On the other hand, people sometimes attempt to drive Klipschoms with amplifiers that were never intended for that particular job. It can be said from experience that the Brook, Leak, and Williamson types perform admirably. These have correctly designed transformers and associated equipment. For the home constructor, there are output transformers available that hold distortion to tolerable levels.

If the above appears to be a diatribe against "beam tubes", reconsider. The "beam tube" in conventional amplifiers has never found wide acceptance by the users of top-quality wide-range speakers. But note the adjective: "beam tube" in conventional amplifiers. There is one "beam tube" amplifier, decidedly not conventional, in which the beam tubes are used in a peculiar combination utilizing the advantages of the cathode follower while at the same time preserving the high-efficiency principle and high output capability of the beam type tube.

This amplifier is "the Mcintosh", which exhibits such low distortion even at high power outputs that it is doubtful if distortion measuring equipment with sufficient resolving power is available to detect the actual quantity of distortion produced. And this is not a case of eliminating the measured distortion while leaving high-order distortion to irritate. Listeners say it is as clean as the best amplifier against which they have compared it.

At the risk of getting commercial in an article of this nature, several amplifiers have been mentioned by name. It is not the author's purpose to rate one amplifier over another. However, for the benefit of those readers who still insist on knowing the author's personal preference for an amplifier to use with the Klipschorn, here is the answer: the Brook.

Not that this is a better amplifier than the others mentioned by name, but for the following reasons:

1) it is a good, low-distortion wide-range amplifier,
2) it has a rather definite upper power limit which prevents damage to the delicate high-frequency speaker driver used in current Klipschorn systems,
3) it has a very highly refined preamplifier with proper equalization provisions, and
4) it is amenable to a very slight revision for full bass extension applicable to Klipschorn (referred to as K-1 re-equalization).

The specific designation of this amplifier is the Brook HA3-K-1 and is rated at 10 watts output. Significant is the fact that it has been found to deliver 9.5 watts of clean output at 30 cycles. This is more than adequate power for driving high efficiency, horn-type corner speakers. Probably, for home use, peak amplifier output of 1 watt would suffice to feed a Klipschorn. At one gathering, a Chamber of Commerce dinner, nearly 1,000 guests were served dinner music at comfortable listening level with the Brook 10-watt amplifier; previous calibration of the gain control indicated that this was attained with 1.5 peak watts input to the speakers. An estimate of the size of the auditorium indicated it to be about 600,000 cubic feet, or about the size of a large aircraft hangar.

It should be understood that these remarks about specific amplifiers are to be considered in the light of the author's experience. Limitations of time and expense have made it impossible to test all the models available. There are certain basic features which should be checked in selecting an amplifier to drive a Klipschorn.

These are: triode output with feedback, or the beam tube arrangement exemplified by the Mcintosh design should be a criterion; the correct design of lower-level stages so that they will be as distortion-free as the output stage; an adequate output transformer; at least 35 decibels of available loss in the preamplifier to permit that much linear bass-boost to equalize velocity-type phono pickups; and equalization to a linear range down to 30 cycles, and preferably down to 15 cycles or below.

Since an audio system starts with the pickup, this device must be considered here, too. The original phonograph consisted of a stylus tracking a groove, the shape of which represented the wave form of the original sound.

The stylus forced a diaphragm to move, and air compressed on one side of the diaphragm was forced into a horn where the high-pressure, low-velocity air motion was transformed into the proper pressure-velocity ratio to be emitted into the listening area.

All the power eventually turned into sound was derived from the groove, and the stylus had to be driven with considerable force. Stylus pressures of the order of half a pound or more were required to drive the diaphragm to sufficient amplitude of motion to achieve a moderate degree of loudness.

With the development of the vacuum tube amplifier, and about 1915, the settlement of the question concerning tube bias requirements for stipulated distortion, it was recognized that the record groove would not be required to deliver so much power if a little power fed to a magnetic pickup could be amplified to drive a speaker.

The result was the reduction of stylus pressures to a matter of 4 ounces or so. Developments in crystal pickups brought this down to 3, then to 1 ounces. By 1940 crystal pickups were on the market with recommended pressures of 1 ounce.

Meanwhile, sound-on-disk motion pictures had served as a proving ground for other types of pickups, and reversion to the magnetic type offered possibility of lower distortion. During the past decade, pickups have been developed which utilize the following principles: strain-gage, magnetostriction, capacitance variation (frequency modulation), variable dissipation (amplitude modulation of an oscillator) and others. A rather important stride in reducing stylus pressures was exemplified in the photo-electric pickup.

The problems in designing a pickup are of a mechanical nature, regardless of what principle is used to transform the mechanical motion into electrical changes. This is a highly involved study, and there is as much engineering and mathematics in the stylus problem involving a few milligrams of weight as there is in a locomotive weighing thousands of kilograms.

The experience of this writer dictates the following criteria for a pickup: the stylus should be short, but slightly compliant; the motion of the tip rather than the shank or some other part should be transformed to electrical motion; and the stylus should be horizontally disposed rather than vertically (this is important as it determines the ability to track the up-and-down pinch effect motion as well as the lateral motion of a laterally cut groove without distortion).

As the pickup head and arm on which it is mounted are functionally inseparable, and as the compliance of the stylus and the weight of the arm determine the lowest frequency that can be tracked, it is logical to conclude that the arm must be adequately heavy, yet counterbalanced to maintain proper stylus pressure. It appears from current knowledge that stylus pressure cannot be reduced below about 14 grams for standard-groove records, or 5 grams for microgroove records, without sacrificing one or more of the requirements of frequency range, tracking, wear, or distortion.

It was this writer's ill fortune (also Paul Klipsch) to be faced with speaker development at a time (1941) when there were no means of playing phonograph program material at sufficient quality level to permit the speakers to be evaluated on a practical basis. Therefore, over the past decade, experiments were made with several types of pickups including some of the types mentioned above.

The most successful of these consisted of a sapphire tip attached to a steel spring about 1/4 inch long, with the tip end vibrating between a pair of magnetic pole pieces. The vibrating spring was substantially parallel with the record, so that it was compliant in both a vertical and a lateral direction. It was noted that stylus chatter with this arrangement was vastly lower than with other stylus type. While planning a tentative production program for this pickup, Mr. W. S. Bachman of the "General Electric" Company described substantially the identical arrangement in Electrical Engineering of March 1946. As his paper had been submitted before the author's development work was well under way, it was decided to drop the idea of producing the pick-up, and to wait patiently for the commercial introduction of the General Electric Reluctance pickup, as it has come to be known! This pickup incorporates all the features found to be necessary and sufficient in a pickup, and has become the author's choice.

Specifically, it is the RPX-046 professional model GE head, with a .0015 tip of either sapphire or diamond for standard groove records (shellacs and transcriptions). For microgrooves, where the recorded level is lower, the RPX-Q4I head is preferred as this head has a higher output and also a higher inductance whereby the high-frequency or treble response is attenuated slightly, and the pre-emphasis of the microgroove recording is compensated for in the head itself.

Using this combination of heads as a matched pair, it is possible to change from one type of record to the other with little or no change in volume control setting, and generally no change in treble control setting at all, except in the case of old, noisy shellacs, where the treble control would be rolled back in any case.

After designing and building a large variety of tone arms, some good, mostly bad, the author settled on the Gray 103 and 106 arms, and these are still highly regarded. However, the fact that the tone arm weight and stylus compliance comprise a resonant system, and the resonance may cause mistracking on radially warped records, it has long been realized that some form of damping is needed. The early GE cartridge made a good approach to the damping problem with their jelly mounting. But in 1950 Gray introduced the 108B arm which has proved to be a very successful damped arm.

It is beyond the scope of this article to consider turntables and motors, but a brief mention will be made of their requirements. For wide-range speaker use, the rumble and wow must be minimized, and this appears to limit consideration to systems not containing gears or governors. Likewise, it eliminates changers of all known types except for applications where a certain amount of noise can be tolerated. Many Klipschorn owners report that they use changers for low-level listening (dinner music) and for applications where a little noise is not particularly objectionable (dancing).

The Rek-O-Kut T-11 is a very quiet table. By remounting the motor of a General Industries type DR table, it is possible to lower the hum and rumble level to a point below that which is present in most, if not all, records.

It should be noted that not all the rumble comes from the turntable. Floor vibration has proved very annoying, particularly so with microgroove record, and especially when there is dancing.

Unless one has a concrete floor, even normal walking may shake the stylus out of a microgroove. Even a concrete floor transmits traffic-induced seismic disturbances with resulting rumbles. The best remedy found so far has been a 4-stud oscillating support suspended from 8 shock mounts.

This has proved better than springs because of the damping (shock absorber) action of the rubber. It appears desirable to make the table suspension resonant at a point at least two octaves below the resonant frequency of the tone arm. In the case of the Gray 108B arm, the resonance frequency appears to be of the order of 2.0 cycles, so the table should have a natural frequency of 5 cycles or lower, two octaves lower being one-fourth the frequency.

.
Weitere Artikel von Paul Klipsch finden Sie hier bei Lautsprecher-Wissen.
.