Many schemes for utilizing the higher acoustical impedance existing in a room corner have been devised.

Kellogg,1) Stone,2) Weil,3) Ephraim,4) Sandeman,5) and Horace-Hulme,6) are representative of the early art. Very recently there has been announced a bass-reflex enclosure designed to fit in a corner.

The present development offers a horn type reproducer of dimensions which are smaller compared to the longest wave-length to be reproduced than has heretofore been possible.

The advantages of the horn are well known. Compared to the direct radiator operating in a flat baffle, the efficiency of the horn is 10 to 50 times higher, and the acoustic loading due to the horn permits a given sound power to be generated with much smaller diaphragm excursions, a factor which greatly reduces harmonic distortion. Well-designed horns are capable of very uniform frequency response. The power handling capacity is high.

Referenzen:

• 1) E. W. Kellogg, J. Acous. 3, 54 (1931).
• 2) Stone, U. S. Patent 1,819,721 (1931).
• 3) Weil, U. S. Patent 1,820,996 (1931).
• 4) Ephraim, U. S. Patent 1,962,300 (1934).
• 5) Sandeman, U. S. Patent 1,984,550 (1934).
• 6) Horace-Hulme British patent, 342,369 (1931).
• 7) Electronics, p. 99 (May, 1941).

-

In the past, however, horns for reproducing low frequencies have been very bulky. A typical theater (gemeint ist "Kino") "woofer" occupies a space nearly 4 x 4 x 8ft.

Figure 1 shows a horn unit adapted to fit in a corner. With a frequency response down to 40 cycles, corresponding to a wave-length of 340 inches, comparable to the performance of the typical theater woofer, this unit measures only 39 inches high, about the same width across the "wings" and only 28 inches deep from front panel back to the corner. The cabinet size is comparable with radio consoles and is entirely suitable for use in homes. The power handling capacity is such as to make it adaptable for use in community centers, little theaters and medium size recreation halls. By suitable modification, power handling capacity for large theaters may be attained.

To reproduce this wave-length with the same degree of smoothness of throat impedance and the same efficiency, a conventional horn in free space would require a mouth area of about 4.500 square inches. If operated close to a floor or wall, 2.300 square inches would be required corresponding to a circular mouth 53 inches in diameter. The horn length required to match a 15-inch diaphragm would be over 80 inches. The horn of Fig. 1 has an actual mouth opening of 570 square inches and a horn length of approximately 40 inches representing a saving in volume of over 75% and a corresponding saving in material for its construction. A further saving in material results from using the room walls for part of the horn structure.

Figures 2, 3, and 4 show the construction of the unit. Figure 2 is a sectional view showing how the unit is nested into the room corner, and also the arrangement of side baffles, front panel, and driving unit. The side baffles are arranged at such an angle that they, together with the room walls, floor, and cover plate, form the final horn section. Radiation into pi/2 solid angle (due to the reflections from the wails and floor) make the required mouth area only 1/4 as great as would be required for radiation into 2pi solid angle where only the floor is used as a reflector.

Figure 3 is a vertical section through B-B of Fig. 2 and shows the air passages from the diaphragm to the rear of the horn cabinet. The driving unit and an air chamber behind the diaphragm are illustrated. Figure 2 is the section A-A of Fig. 3.

Figure 4 is a section through C-C of Fig. 3 and shows the throat opening from the diaphragm and the first diverging set of baffles.

The design illustrated is for a 12-inch driving unit (Anmerkung: das Bass-Chassis). The frequency range is from 40 to 400 cycles. For good diaphragm loading at 400 cycles a throat area of about 50 square inches is about right for a 10.5-inch diameter diaphragm and a moving system weighing between 14 and 18 grams. Such a small throat imposes an unnecessarily high load at the lower frequencies. Consequently, the initial taper is such that the horn area doubles in a length of about 8 inches, corresponding to a cut-off of 100 cycles.

The remainder of the horn flares at such a rate that the area doubles every 16 inches so that the nominal cut-off is 47 cycles. This results in a "rubber throat" the area of which is about 100 square inches up to 100 cycles and decreases with frequency until at 400 cycles the effective throat area is 50 square inches. Except that the initial taper is conical, this idea is roughly the same as that offered by Olson,8) and roughly the same results may be predicted.

The mouth size selected is 570 square inches. This corresponds to an equivalent mouth of 2280 square inches radiating into 2x solid angle. This condition exists when the horn mouth is fitted with a flange or is operated close to a wall or floor.

The horn performance may be computed on the basis of one having an actual mouth area equal to the 2280 square inch equivalent mouth and a length equal to the actual length. This results in the computed throat impedance being that of a throat of 4 times the actual throat area. The mechanical inpedance on the actual diaphragm may be found by multiplying this computed mechanical impedance value by 4. Calling the mouth area 570 square inches is conservative, as that is the horn area at the forward edges of the non-parallel side baffles. The additional taper due to the parallel side portions will undoubtedly reduce the magnitude of the first peak in the throat impedance curve below the computed value.

For an equivalent 100-cycle horn, the mouth area would be (47/100) 2 x 2280 = 500 square inches, the area of a 25-inch diameter circle. Olson (9.) has worked out throat impedance curves for various mouth sizes. This 25-inch mouth is intermediate the 20- and 30-inch mouth sizes, the performances of which are shown in his figures 5.5b and 5.5c, and the horn length corresponds roughly with his Fig. 5.6d.

Thus the impedance curve may'be expected to be somewhat smoother than that shown in his Fig. 5.6d due to the 25 percent larger diameter mouth. A few points were computed and this seems to be the case. The impedance was computed to 200 cycles, and the effect of the multiple taper beyond this frequency was estimated on a basis of Olson's work on multiple flare horns. (8) The results are shown in Fig. 5. The solid curve is the resistive component of the throat impedance in mechanical ohms on the diaphragm. The dotted curve is the reactance in the same units. The dashed curve is the combined mechanical reactance of the diaphragm mass and the behind-the-cone air chamber.

The effect of the air chamber between the throat and the cone can be computed from the equivalent circuit. The combined effect of these factors is plotted in-Fig. 6, in which the solid curve is the resistive component and the dotted curve is the total reactive component of the load on the diaphragm in mechanical ohms. The reactive component includes the horn throat reactance, the cone mass reactance and the effect of both front and back air chambers.

The motional impedance for a driving unit with a product of flux density square times voice coil mass of 550 X106 and a voice coil resistance of 6.5 ohms is plotted in Fig. 7. This impedance is simply

Figure 9 shows the results of motional impedance measurements which were made on an experimental model. This model was constructed along the general lines just described but differed somewhat in the more complicated and less efficient folding (from the standpoint of conservation of space), and in that the back-of-cone enclosure consisted of a box bolted to the front of the cabinet. The construction of this rather awkward affair is shown in Fig.10.

The experimental model is inferior in many respects to the design described earlier in the paper. Orders to report for active military duty preclude a thorough investigation of the new design. From the computed and experimental results on the experimental model about to be described and from the computed performance of the new design, it may be safely predicted that the actual performance of the new design will be better than for the experimental model.

The "blocked" impedance measurement was obtained by turning off the field supply. The differences between the pairs of curves give the motional resistance and reactance. There may be some question as to the accuracy of such a "blocked" impedance measurement, but, since the flux from the voice coil has to traverse the air gap twice, it seems that the saturation would have little effect on the static voice coil impedance, especially at frequencies below 1000 cycles.*
-

The driving unit was a Jensen 12A with a resonant frequency in free space of 63 cycles. The measured stiffness of suspension when new was approximately 3300 grams per centimeter. The force factor, Bl, or flux density times active length of voice-coil conductor was measured and found to be 1SX106. The voice coil resistance was 6,5 ohms.

From these factors and from horn theory, taking into account the mass reactance of the moving system, the suspension and back-of-cone air chamber compliance, and the horn throat impedance calculated from the dimensions, the motional impedance was computed.

The computed resistive component of the motional impedance is plotted (solid curve) in Fig. 11, together with the rneasured resistive component (dotted curve).

The measured curve indicates better low frequency performance than the computed curve. While there is only a fair agreement between the two curves, it is believed that the discrepancies are explainable. First, the horn theory must be in considerable error in the immediate vicinity of cut-off. (10)

Second, vibration of the thin side baffles can readily cause the peaks and hollows above 80 cycles. Third, the air chamber between throat and cone was of complicated shape and the estimated volume may have been in considerable error, giving rise to error in the computed impedance above 200 cycles.
-

The cut-off of the horn is 47 cycles and the computations show poor performance below 55 cycles. But the measurements and critical listening tests show good performance down to 40 cycles.

The peaks in impedance above 80 cycles probably do not indicate increased efficiency but rather increased losses due to mechanical vibration of the thin side baffles.

In spite of the rough appearance of the measured impedance curve, listening tests with an oscillator showed no pronounced peaks or hollows below 350 cycles, and none at all below 200 cycles.

For some time this experimental model was operated without the back-of-cone enclosure. In fact, it had originally been intended to derive tht high frequencies from the side of the cone facing away from the horn in the manner of Olson and Hackley. (11) However, the low efficiency in the range wherein the speaker was not loaded precluded satisfactory performance and a high frequency unit to cover the range from 400 cycles up was added, and a cross-over network used to allocate the electrical power to this two-way system.