Radio broadcast .. (1922-30)

component must be equal to the r.f. re- sistance of the coil. Now the latter, with the usual design of coils, varies in a man- ner roughly proportional to frequency; but the reactance of the coupling induc- tance is also proportional to frequency. Therefore, the optimum coupling can be maintained over a wide band of frequen- cies. If capacity coupling be used, how- ever, this is impossible, for the reactance of a condenser is inversely proportional to frequency. It can be seen, therefore, that when it is desired to maintain op- timum coupling over the whole tuning range of a receiver, there is no alternative to inductive coupling. In most cases, however, it is desirable that the coupling should be greater than the optimum, which always gives a single- peaked tuning curve. With a greater value of coupling a double-peaked curve is obtained, and this can be used to compen- sate the usual loss of sidebands in follow- ing cascade tuned circuits. If the peaks are arranged to occur at frequencies 10 kc. apart, audible notes of 5000 cycles will be reproduced at greater strength than notes of 50 cycles; and when this is done, it is obviously desirable that the peaks should occur at the same distance apart over the whole tuning range of the re- ceiver. This, however, is usually impos- sible. In what follows it is assumed that tun- ing is carried out by means of variable con- densers; it is not applicable to the rare cases where variometers, or a combina- tion of variometers and variable condensers, are used. Band Width At whatever fre- quency wL = 1/uc, at a frequency 5 kc. dif- ferent from this the value of wL —1/iix; re- mains approximately constant; consequently, the quantity R 2 + X 2 of equation (6) must also be constant in value over the whole tuning range of the re- ceiver, if the band width is to remain constant. It will be seen that with inductive coupling this quantity is much greater at high than at low frequen- cies. Neglecting the effect of the resistance, when the two peaks are 10 kc. apart at a frequency of 600 kc. they will be 20 kc. apart at 1200 kc.; the increased r.f. re- sistance at high frequencies, which always occurs in practice, increases this varia- tion in band width. With capacity coupling, however, the case is exactly opposite. Neglecting the resistance, the band width at 600 kc. is double that at 1200 kc.; but in this case, the effect of the resistance is to reduce the variation in band width. The coupling reactance is smallest when the resistance is greatest, and vice versa; consequently, the value of R 2 + X 2 tends to remain con- stant. Indeed, by suitable design of the coils it is possible to make it quite con- stant, but this is not desirable, since the necessary increase in r.f. resistance at the higher frequencies would make the cir- cuit very unselective. It is quite possible, however, to affect a compromise. Coupling Variation When a capacitatively coupled filter is used, however, the coupling does not re- main constant; it is considerably less at high frequencies than at low, and this can be used to compensate the low selectivity of both primary and secondary circuits at high frequencies. It is quite possible to design a circuit which will give constant selectivity over the entire broadcast band Table II—Characteristics of English A.C. Screen-Grid Tubes From this it would seem that the only difference between capacitative and in- ductive coupling is that the variation of band width with frequency is less with the former. There is, however, another difference, and this also is due to the difference in the variation of reactance of the coupling components. It is well known that the ordinary tuned circuit is Table I—Essential Figures for Selectivity, Gain, and Sideband Variation less selective at high frequencies than at low, and this is partly due to the increased coil resistance. It is also well known that with any filter circuit a decrease in the coupling increases the selectivity. When an inductively coupled filter is used with ordinary coils, the coupling remains con- stant at all frequencies; both primary and secondary circuits are less selective at high than at low frequencies, and, since the coupling is constant, the variation in selectivity is accentuated. It is about the same as that with two cascade tuned circuits. 08*60 CAPACITATIVELY COUPLED FILTER AMPLIFIER Fig. 3 of frequencies, or even one which is more selective at the higher frequencies. In order fully to appreciate the differ- ences between capacitative and inductive filter circuits, it is necessary to compare the resonance curves at different frequen- cies. In Fig. 1 are shown three curves; A is for a frequency of 600 kc. and is for both inductive and capacitative filters, B and c are for a fre- quency of 1200 kc. and are for capacity and inductance coupling, re- spectively. For both filters the coil induct- ance was 240 mh., the r.f. resistance at 600 kc. was 10 ohms, and at 1200 kc. was 20 ohms. The coupling induct- ance in the inductive filter had a value of 4.7 mh., and the con- denser in. the capacita- tively coupled filter had a capacity of 0.015 mfd. —values which give the same degree of coup- ling at 600 kc. The essential figures for selectivity," gain, and sideband variation are given in Table I. Greater Selectivity The immense superiority of capacity coupling is at once evident; both filters give the same results at 600 kc., but at the higher frequency, the selectivity with ca- pacity coupling is nearly three times that with inductive, being 30 instead of 10.5. At 600 kc. it is 33 with both circuits. These figures for selectivity are obtained by dividing the filter output voltage at reson- ance by the output voltage at a frequency 40 kc. different from resonance, the input voltage being the same for both frequen- cies. This method of expressing the selec- tivity is used throughout this article. The amount of sideband cutting is called the sideband variation, and is expressed as a loss in percentage. With band-pass filters it is sometimes a high-note loss and some- times a high-note accentuation; the upper sideband Limit is taken as 5000 cycles. The ratio e/E has been defined earlier, but it must be remembered that it is not the actual amplification from the antenna to the grid of the first r.f. tube; the actual amplification depends upon the method and degree of coupling to the antenna, as well as the antenna constants. Returning to the consideration of Fig. 1. It can be seen that the amplification is • RADIO BROADCAST FOR APRIL • 32: