Radio Broadcast (May 1928-Apr 1929)

SEPTEMBER, 1928 CHECKING UP ON AUDIO DISTORTION 291 reach it. This, of course, increases the steady plate current. Another remedy, and in this particular case a better one, is to increase the plate voltage to 135 or higher; it may be seen that this accomplishes the same purpose as above, by shifting the operating point to one of the higher curves and automatically increasing the distance to the lower bend of the curve. DISTORTION FROM A POSITIVE GRID AMPLITUDE distortion caused by cutting off of the positive peaks of signal voltage may be had at another portion of the curve — in particular, where the grid voltage crosses the zero line and goes positive with respect to the filament. To find out why this causes distortion requires that a start be made at the plate circuit of the preceding tube. The connections are those of Fig. 4A, which shows two tubes with resistance coupling. The coupling resistance, the coupling condenser, and the grid leak may be replaced by a single impedance, Z, equivalent in value; and the equivalent circuit is redrawn in Fig. 4B. The first tube has an alternating voltage generated in its plate circuit, due to the signal voltage at its grid. The generated voltage acts upon the internal plate resistance of the tube, Rp, and the coupling impedance in series, causing an alternating current to flow through them. Each element then has a voltage drop across it, the sum of the drops being equal and opposite to the impressed voltage. The proportion of voltage that exists across Z may be expressed very nearly by Z + R X total voltage. That is, the larger the value of Z, the higher the voltage existing across it, and vice versa. In the amplifier circuit, suppose the B voltage on the last tube to again be 90, but suppose the C bias has been dropped to only 9 volts, so that the operating point becomes "B" of Fig. 1. As long as the signal voltage swings within limits around "B" — namely, plus or minus 9 volts, there is no distortion. But if the signal amplitude runs higher, a direct current flows from filament to grid each time a signal peak makes the grid positive with respect to filament — exactly as a current flows to the plate when it has a positive voltage on it. When the grid is negative, no current flows to it, when positive, more or less current flows, depending on how far positive the voltage goes. This can only mean that the resistance from grid to filament has been decreased enough to permit current to flow; in fact, compared to the resistance when no current flows, the grid-filament resistance goes down tremendously when grid current flows. But the gridfilament path is in parallel with the coupling impedance; and when paths are in parallel, the equivalent impedance of the combination is always lower than that of the smallest element. Therefore, when the grid-filament path drops from a very high to a low value of resistance, it drags down the equivalent value of the coupling impedance, which in turn means that the a. c. voltage across the coupler decreases. It must be remembered that the grid-filament resistance is lowered only on the positive peaks, so it is at these points that the voltage is lost, and the tops of the input signal are cut off. Losing the tops of the wave means that the negative loops become relatively larger, so that the average of the plate current goes down. When amplitude distortion takes place because of the grid swinging positive, the reading on the plate milliammeter decreases. The remedy in this case is to increase the Cbattery bias, moving the operating point away from the zero line so that the voltage swings cannot reach it. THE PROPER REMEDY TT MAY be seen, therefore, that 1 there is an optimum value of C bias for each value of plate voltage. It is somewhat less than half of the bias necessary to give zero plate current, by inspection of the characteristic curves. The object is to the confine the grid voltage swings between the point of zero voltage and the lower point where the curve begins to bend — in other words, to the part of the characteristic where the "curve" is a straight line. If, after attaining the optimum bias, amplitude distortion still occurs, the remaining remedy is to increase the plate voltage; incidentally, this necessitates a new bias. The latter remedy is limited, of course, by the permissible plate voltage that may be applied to the tube. Above this power level, a new tube of higher power rating must be used; or, with the same tube, the condition simply means that the volume control on the receiver must be lowered, if amplitude distortion is to be done away with. In the case where the grid voltage goes positive a meter may be placed in the grid circuit and used to show the grid current that flows. If the amplifier is resistance coupled to the previous tube, the flow of current will be so small as to be indistinguishable on a one-milliampere meter. The reason is that only a slight value of current flowing through the high-resistance grid leak provides a negative drop that compensates for the positive peak of signal, and the grid cannot go far enough past the deadline to give an appreciable current. Amplitude distortion is present, nevertheless, just as much as with any other coupling system. Nothing has been said heretofore concerning filament saturation, but assumption has been tacitly made that in all cases the filament was capable of supplying all the electrons needed. If not, the characteristic curves shown in Fig. 1 would, at a given plate current, bend sharply to the right and continue horizontally. Such a curve would obviously cause serious amplitude distortion if the grid voltage variation came into the saturation region. The distortion would be analagous to that caused by grid current flow. Such distortion is often caused by an exhausted tube or a low voltage storage battery. Assurance must be had, then, that the filament of the tube is turned up to rated voltage, or that it has not lost its emission. A very approximate way of obtaining the optimum C bias is to short the grid to the filament of the amplifier tube, and read the plate current. Then remove the shorting wire and adjust the bias till the plate current is slightly more than half the former value. All makers of power tubes indicate the proper B and C voltage which should be used. The effect of amplitude distortion is to introduce in the output frequencies that are not present in the input voltage. A pure, smooth input voltage, such as those of Fig. 2, impressed on an amplifier in which amplitude distortion occurs, would come out with one or both the peaks flattopped, as has been seen. This sort of a wave is composed of more than is apparent from a casual inspection. It really has the original pure wave combined with a family of harmonics — waves 10 I -0+ j GRID VOLTAGE IE I , . 'Volts "j *jdl? I w I Input Voltage I FIG. 3 with frequencies that are 2, 3, 4, 5, and so on, times the original frequency. For sake of illustration, suppose the signal has its positive peak cut off and that it is cut straight across. Such a wave is shown by the heavy curve of Fig. 5. The finerlined curves show the components that are actually present, and which in the composite form the output wave. With this sort of action accompanying amplitude distortion, there is no wonder at the weird jumble issuing from some loud speakers when trying to reproduce musical selections, where each 'note proceeds from the loud speaker accompanied by its own little family of harmonics. A bit of experimenting with a milliammeter while listening to the reproduction will enable one to connect the results of amplitude distortion with the cause. There is an interesting connection between frequency distortion and amplitude distortion. With transformers in the amplifier which will not pass the low notes, the set will handle a given power level without distortion. If, in an attempt to better the fidelity of reproduction, transformers be installed which will pass the low notes, the effective maximum power capacity of the amplifier will be lowered because, as is more or less well known, the lower notes contain more energy or have a greater amplitude. The better class of broadcasters have the low-frequency, large-amplitude voltages present in their output. Where the original amplifier would not pass these notes, the better transformers enable it to do so; and although the higher frequencies still come through with the same amplitude, the newly introduced low frequencies overload the amplifier. It is well to make sure that the loud speaker can reproduce the lower frequencies before installing transformers to pass them. Otherwise the only attainment would be to reduce the power capacity of the amplifier, for even though the speaker cannot respond to the new low notes, it will respond to their distortion products. Distorted Output Wave