The Technician Explains

Radio Record, Volume III, Issue 12, 4 October 1929, Page 30

The Technician Explains

The Efficiency of Parallel Feed

(By

CATHODE

YEADDRS will be familiar with the type of high-fre-quency coupling described by the term "parallel-feed"; for a parallel-feed BrowningDrake has previously been described in the "Radio Record’s" technical pages and has received very favourable comment. The circuit is one concerning which a great deal of misapprehension seems to exist; it is hoped to make short work.of the task of removing this misapprehension, the arising of which is an apt illustration of the truth of the ancient saw-"A little ‘knowledge is a dangerous thing." Hxamining a_ typical parallel-feed coupling such as that illustrated in Fig. 1, we find in the plate circuit of the high-frequency valve a choke and a tuned circuit which are effectively in parallel; the return path from the choke to the filament circuit is, of course, via the B battery, which may be regarded as a continuation of the wiring. Now it is fairly common knowJedge that the amplification obtained from a valve depends upon the magnitude of the impedance in its plate cireuit. Taking, for instance, a resistancecoupled audio stage such as that shown in Fig. 2 and applying the well-known formula :- Re Amplification = g * ----- Ri + Re where g = amplification factor of valve Re = external resistance Ri = impedance of valve. we find, assuming the valve to have an A.C. resistance (or impedance) of 20,000 ohms, that an increase in the coupling (or external) resistance from 10,000 ohms to 40,000 ohms will result in the amplification being doubled. Suppose now that we endeavour to tse a resistance coupled stage such as that of Fig. 2 for high-frequency amplification in the broadcast band. What EE

will be the result? We know only too well. Amplification will be practically nil, And why? Because the various _sStray circuit capacities and the grid fiament and grid-plate capacities of the succeeding valve are in parallel with the coupling resistance. Bearing in mind that the combined value of two impedances in parallel (excluding resonace effects) is always less than that of either, and remembering further that the impedance presented by 20 or 25 micromicrofarads of stray capacities to broadcast frequencies (say, 300 metres) would be only in the region of 7000 ohms, it will be appreciated that we could scarcely expect any worth-while amplification. Replacing the plate resistance by a high-frequency choke as in Fig. 3 does nothing to improve the position. The choke will, of course, be substantially equivalent to a high resistance at broadcast frequencies, the impedance presented at differing frequencies varying in the manner depicted in the lower curve of Fig. 4, which is the impedance curve of a very high quality commercial choke. It will be seen that the impedance presented at broadeast frequencies is only about 20,000 ohms, 2 value which, as it stands, is not such as to secure anything approaching maximum amplification from a valve; when, in addition, we consider the effect of the stray capacities in reducing what amplification we might otherwise obtain, the position again appears rather hopeless. Yet certain Bnglish journals have lately been endeavouring to popularise circuits of the Fig. 3 type; it seems to the writer only fair to warn construc- —

tors that their use can only result in impaired amplification and an intolerable lack of selectivity. Returning now to Fig. 1, we still find a high-frequency choke in the plate of the valve, the only distinction between this circuit and that of Fig. 8 being that in Fig. 1 the grid leak which appears in Fig. 3 is replaced by a tuned circuit which, like the grid

leak, is effectively in parallel with the choke. Seeing that the impedance presented by the tuned circuit, even if it be of fair efficiency, is unlikely to exceed 200,000 ohms, while the grid leak which it replaces might have a resistance of 2 megohms or more, one might readily assume that the only effect of intnoducing the tuned circuit, leaving selectivity out of the question, is to reduce amplification still more. I say one might readily assume this, and a great many people who should know better have assumed it, but in point of fact, as a practical test will soon show, amplification will be very considerably increased by the change. In order to appreciate the reason for this increased. amplification let us redraw Fig. 1 in the form shown in Fig. 5. Here we have the high frequency choke stripped of its self-capa-city which is shown separately at Cl. Similarly the grid-filament and gridplate capacities of the next succeeding valve are shown as C2 and C3 respectively. Now we are probing earnestly at the solution of the increased ampli-

fication. It will be clear that these three capacities, which have previously operated to seriously curtail amplification, are now in parallel with the tuning condenser, their effect being merely to increase the capacity of this component very slightly. This, then, is the explanation. The self-capacity of the choke and the. various stray capacities absorbed. into the greater capacity .of the tuning condenser, increasing this slightly, while the impedance presented by the choke becoms ~,,that which would be presented by ts edingly high inductance were it rely devoid of self-capacity. Just how high this impedance is may be ascertained from the upper curve of Figure 4; it will be seen that, over the broadcast band, it varies from about three-quar-ters of a megolm to a megohm and a quarter, values so high as to reduce only slightful the amplification which could be obtained from the tuned circuit alone were this connected directly in the plate circuit. Since we have gone to all this trouble to show that the parallel-feed circuit is, as regards amplification, subsequently equivalent to the admittedly good tuned plate coupling, we have imposed upon ourselves the further task of showing what, if any, advantage the parallelfeed circuit has over the tuned plate. It has, in fact, two advantages. In the first place, high-frequency currents a to some extent kept out of the "B" ply which, if it possessed any substantial impedance to high-frequency cur- ;

rents, as a battery not in its first youth or a mains supply readily might, would in all probability introduce instability of the worst kind. Secondly, it permits tapping the grid coil in the manner y illustrated in Figure 6, a procedure/ which itself confers two benefits, firstly, that of increased selectivity, and, secondly, that of permitting the proper matching of the impedance of a threeelectrode valve; where a screen grid is used, there is no object in tapping the grid coil unless it..is desired to

‘Tsecure enhanced selectivity at the cost of 2 drop in amplification. A comparison of parallel feed with igh-frequency transformer coupling does not show any substantial advantage in favour of parallel feed, assuming that the high-frequency transformer used, ix one of real efficiency. There is this to be considered, however: that the’ construction of a highly-efficient transformer is a matter of some delicacy, the completed component being anything but robust, while the construction of a simple tapped or untapped coil is a matter of the utmost simplicity, and will result in a product capable of suStaining any amount of knocking round.

There is one other point which might be dealt with, namely, the ascertaining of a suitable tapping point on the grid coil to match a valve of any particular impedance, It is felt, however, that this can be better dealt with when the writer has gone a little further with the papers on coil and high-fre-queney transformer design which are at present appearing in the form of a short series, and readers will doubtless condone the slight delay in dealing with this particular matter. As a final word, it is desirable to stress the necessity of using a choke of the highest possible quality in paral-lel-feed circyits. The effect of any substantial loss in a choke thus employed is to seriously reduce the effective impedance in the plate circuit of the valve, and with it the amplification. For the enthusiastic constructor it is proposed Oo prepare in the near future compléte ata on the design and construction of high-frequency chokés having some pretensions to efficiency, and this will be published in due course. We have learnt from our simple analysis that to be of the utmost service in a parallelfeed circuit, the high-frequency choke employed must posséss @ very high indyctance (that is, it must have a great many turns), while a certain amount of gelf-canacity will not be so serious as might be the case in a different circuit. These things will be borne in mind when ¢he process of design is embarked upon, "Different chokes for different purposes" will be the watchword of the future.