Radio Broadcast (May 1928-Apr 1929)

onie Experiments With Bamd^Pass Filters By KENDALL CLOUGH Engineering Department, Silver-Marshall, Inc. "pOR months we have been trying to secure ■*■ quantitative data on band-pass selectors and Jitters for use at broadcast frequencies. Not only does Mr. Clough, who is Chief Engineer of Silver Marshall, Inc., give the result of his laboratory work, but he gives some idea of how the home experimenter may play with the circuit for himself. Mr. Clough promises more interesting how-to-do-it material for an early issue. — The Editor. A CONSIDERABLE amount of material has appeared in the engineering press on band-pass filters for radio-frequency tuners. Principal among these is the circuit discussed by Dr. Vreeland in the Proceedings of the Institute of Radio Engineers, March, 1928. In his paper Dr. Vreeland points out very completely the advantages of the use of a band-type filter in the tuner of the receiver, but for the benefit of those who have not had access to this paper, these advantages are redescribed here. If we were to connect a stage of radio-frequency amplification, as shown in the circuit of Fig. 1, and run a resonance curve at 1000 kc., we would find that the circuit responded at and about resonance as shown in curve a of Fig. 2. Now in receiving a signal from a transmitter at 1000 kc, we would find that, in the course of modulation, frequencies varying from 995 kc. to 1005 kc, had been combined with the carrier. Obviously, if the reproduction is to be of the best, a band of frequencies must be transmitted from the antenna to the loud speaker with equal amplitudes rather than the single carrier wave only. Just how wide this frequency band should be has been the point of many discussions, some contending that we need to regard only a band 5 or 6 kc. either side of resonance, while others believe that a band 10 kc. wide each side of the carrier is necessary for perfect fidelity of reproduction. The finest audio equipment manufactured to-day is designed on the 5 kc. basis, so it seems superfluous to consider a band of greater width than this. We are not concerned here with the actual band width, however. The fact remains that, whichever stand one wishes to take, the resonance curve a of Fig. 2 does not permit the free passage of a band of frequencies of any M = 20uLh FIG. I A transformer with a tuned secondary constitutes the coupling device used between one radio-frequency amplifier tube and another in the vast majority of present-day receivers. THIS PICTURE SHOWS THE EXPERIMENTAL BREADBOARD RECEIVER WHICH THE WRITER CONSTRUCTED TO TEST THE BAND-PASS PRINCIPLE appreciable width. It will be seen in the curve that a frequency 5 kc. off resonance is amplified only 83 per cent, as greatly as the carrier, and a frequency 10 kc. off resonance only 62 per cent, as great. RESULTS WITH THREE STAGES PROM an interference standpoint the single 1 stage of amplification would be far from adequate for modern conditions, so we have shown the resonance curve b in Fig. 2 which was obtained by cascading three of the circuits. It can be seen that the selectivity to an interfering station is greater, while the 5 kc. amplification is only 60 per cent, of normal and the 10 kc. amplification 25 per cent, of that of the carrier. The operation of this receiver would be equivalent to the use of a tuner with a perfect band pass and an audio amplifier having good amplification of the bass notes and falling to 60 per cent, of the bass amplification at 5000 cycles, and 25 per cent, at 10,000 cycles. It should be remarked that the receiver having the resonance curve b of Fig. 2 would not be considered a particularly selective receiver, so the reader can judge for himself the side-band cutting that is going on in the high-grade selective outfits. The ear is a tolerant device, and never seems to miss that which it has not heard. Now, it may be demonstrated that the resonance curve shown is a definite geometrical shape. By this we mean that there is no adjustment (such as the resistance of the coil, the primary coupling, or the L/C ratio) which will cause the circuit to admit, say, a 5 kc. band with more facility without admitting an interfering station 10 or 20 kc. off resonance with corresponding facility. Thus, the only circuit of the usual resonant type which would provide perfect fidelity would be a circuit infinitely broad, a mathematical fiction which would be worthless in reality. This indicates that an entire change in the shape of tuner response would be desirable. The dotted-line rectangular curve of Fig. 2 would be the theoretical ideal shape. This shape is not capable of attainment, but there is a cir 104 cuit, old in the art, which under proper conditions will produce a response approximating this curve more or less closely. Dr. Vreeland has discussed a similar circuit (Fig. 3) in detail in the paper' mentioned, but it can be shown that this circuit is the analytical equivalent of the circuit with which we are to deal, Fig. 4, and which has been covered theoretically in all standard texts. So thoroughly has it been discussed that there is little we can add to the treatment other than to present curves and observations made in the laboratory. It is hoped that certain readers will find sufficient material and interest in these notes to enable them to go on with the experiments in this interesting field of band-pass filters which is far from a state of perfect practical application. The theory of this circuit indicates that, when the coils, coil resistances, and condenser capacities are identical in each circuit, both circuits are tuned to the same frequency (due to the identical construction) when operated independently. 8 70 o 60 v 1 4 i / / t— .X / > N \ * 1 1 -30 -25 -20 -15 -10 -5 0 +5 +10 +15 +20 +25 +30 FIG. 2 If the voltage across the condenser in Fig. i (e->) were measured as the frequency of the voltage input to the preceding tube was changed, a curve similar to "a" in this graph would result. If three stages were used, the selectivity would be greater, as shown by the decreased response at points jar away jrom resonance in curve "b."