The motion picture projectionist (Nov 1929-Oct 1930)

October, 1930 Motion Picture Projectionist Response Curves for Sound Apparatus By C. H. W. NASON IN the design of units for sound systems it is essential that perfection of each independnt element be striven for. It was the practice in the radio engineering field up until a year ago to meet the deficiencies in certain classes of equipment with compensating measures at other points in the chain. The lack of low frequency response in amplifier circuits was compensated for by designing dynamic reproducers which had marked resonance in the lower register. So also was it the accepted practice to peak the response characteristic of the amplifying transformers in the higher range so as to compensate for losses encountered in the preliminary circuits of the receiver. Advances in design during recent years gave transformers and tuning circuits which did not have the deficiencies noted above, and it was found desirable to seek for perfection in each unit rather than to attempt the socalled "matched unit" procedure which had been the rule. Certain manufacturers discovered that the public had become educated to the booming, ill-proportioned bass of the earlier speakers, and this practice was continued until it became evident that the more musically inclined laymen recognized the fallacy of the procedure and would have none of it. Today radio equipment is designed with the logical idea of rendering as faithful as possible a reproduction of the original as is technically possible. Some months ago James J. Finn wrote a short paragraph concerning the "Perfect Speaker" which attempted to show in a few words just what should be expected of a reproducer. In this note the statement was made that the frequency band between 15 and 15,000 cycles was necessary to the accurate delineation of speech and music, but that the human ear could not detect the difference if this range were restricted from the ideal to 30 to 10,000 cycles, and that the change could only be detected by direct comparison with the original should the range be still further restricted to an upper limit of 7,000 cycles. The note also stated briefly that the response should be "fiat" over the range covered. Now comes Frank Reichmann of the Oxford Radio Company with a letter occupying three columns of type which Mr. Finn published in the issue of August last. Mr. Reichmann takej up arms over the statement regarding the "flat" characteristic in a letter so replete with fallacies that it is difficult to realize that it was written by a man old in the practice of the engineering art. This "Flat" Characteristic With all the recent propaganda regarding "canned music" in the theater I believe that it is self-evident that the entire chain from microphone to loud speaker should be so perfect as to amount to no chain at all, but to a clear channel from one to the other. I believe that when we mention a flat characteristic we have just that in mind — that no one frequency will come to the audience in an improper relation to the others. There is just one condition under which this premise becomes of a doubtful character. Under certain conditions it is necessary to offer to the audience a rendition differing in volume or power of sound from the original. In the small theater the recording of a brass band must certainly not be run off with the sound from the loud speaker equivalent in power to the actual output of the band. Also, in a large theater it is necessary to boost the level of ordinary speech so as to have the sound reach the furthest recesses of the house. In the accompanying illustration (Fig. 1), there are two curves show 16 32 64 128 256 512 1024 204B 4C96 8192 16364 FREQUENCY O.V. Fig. 1. Relative effect of various sound frequencies upon the human ear. (From Fletcher's "Speech and Hearing," D. Van Nostrand, N. Y.) ingthe response of the human ear to various frequep.cies Of sound. This figure is drawn bodily from Speech and Hearing by Dr. Harvey Fletcher of the Bell Laboratories. For our purpose we will refer to the lower curve which shows the absolute mechanical power required of a sound wave if it is to produce a perceptible effect on the human ear. We are particularly interested in the range between thirty and eight thousand cycles. You will note that the pressure required at eight thousand cycles is a bare fraction of that required at the lowest point in the required range. If a sound system having an essentially flat characteristic is employed to transfer the sound of a singer from one point to another, and the gain through the circuit is adjusted so that the pressure from the speaker is the same as that from the singer's own vocal apparatus, the reproduction will be the same in all respects as the original. The apparent effect at each frequency is proportional to its pressure — above the threshold of audibility, not to its actual pressure above zero. If the singer's voice includes all frequencies over the stated range at an average pressure of one dyne, all the components of the original will be audible. Now suppose we reduce the output from the reproducer so that the average pressure reaching the ear is of the order of .01 dyne. Reference to the curve will show that all frequencies below 192 cycles will be below the limits of audibility and will not be heard. In order to preserve the illusion it will be necessary to so reapportion our electrical circuits as to make the entire band of frequencies have the same proportionate audible effect. It is not enough that they retain their proportionate values with regard to power. In order to preserve the illusion it will be necessary to raise the power output at the low frequencies so that they will be heard in their proper relationship. In the other possible case where the sound output from the reproducer is greater in its average power over the frequency range than the original, it will be necessary to effect a correction by lowering the bass response. This instance is the only one, other than the well-known cutting off of the higher frequencies to eliminate background noise, which points to the desirability of a response curve deviating from the "flat." Even in this case the correction is necessarily variable, being dependent upon the relationship between the average power of the original sound and that of the reproduction. Such a correc