Radio Broadcast (May 1928-Apr 1929)

Checking Up on Audio Distortion By G. F. LAMPKIN F THE two kinds of distortion that may occur in an audio-frequency amplifier, the one that usually receives the lesser attention is the one easier to correct. If an amplifier does not produce an output that is an exact, enlarged reproduction of the input — i.e., if the output signals are not proportional in strength to the input signals at the broadcasting microphone, no matter how weak or strong those signals may be — obviously, distortion is taking place. Such action is designated as amplitude distortion; and the intelligent use of nothing more than a milliammeter will show the causes of this kind of distortion, when appropriate remedies may be applied. The correction of the other kind of distortion — frequency distortion — necessitates the use of more extensive equipment. This type of distortion is characterized by unequal amplification in different parts of the frequency spectrum. It is the type that has received somewhat greater attention in the design of audio transformers, coupling impedances, etc., with good characteristics. Suffice it to say that, in the analysis of a receiver for frequency distortion, the frequency characteristic of a single coupling unit may mean little or nothing. The frequency characteristic of the loud speaker may mean much more than the characteristic of a single audio transformer, for example. Only the overall frequency characteristic of the receiver, from the antenna to the loud speaker, should be relied on as a figure of merit. It is impossible to make each unit of the receiver perfect; but, fortunately, it is not impossible to make the composite performance of all the receiver units approach perfection. Making one unit imperfect in the opposite sense to another will cause the defects to cancel, and so leave the near perfect characteristic. In any case, successful correction of frequency distortion requires the use of more than a milliammeter. THE CAUSE OF AMPLITUDE DISTORTION AMPLITUDE distortion is the result of amplifier overloading. A given tube is constructed to handle a certain amount of audiofrequency power without overloading. When this power limit is passed, overloading and amplitude distortion occur, so that the two go hand in hand. The more or less familiar curve that shows the relation between the plate current and the grid B Voltage = 180V. PLATE -CURRENT GRID-VOLTAGE CHARACTERISTICS -50 -40 -30 -20 -10 0 HQ GRID VOLTAGE FIG. I REQUENCY distortion in audio ampli•*■ fiers is primarily a question oj poor design, and is somewhat beyond the province of the set builder. Amplitude distortion, on the other hand, is usually the result of improper operation; its detection, as Mr. Lampkin explains, may be accomplished by the use of a milliammeter, and its correction by the proper adjustment of the operating voltages of the amplifier. Any radio fan possessing a milliammeter can make the measurements described in this article, and in this way not only get better results from his amplifier, but also learn a great deal about audio amplification in general. —The Editor. voltage of a vacuum tube can be used as a basis for showing how amplitude distortion takes place, and how it may be eliminated. Typical characteristic curves for one of the so-called power tubes are given in Fig. I. They show how the plate current of the tube would vary when the grid voltage is swung through various values, for any particular B voltage. With a 135-volt B battery, if the C voltage is minus 30 volts, the plate current is 13 milliamperes: if minus 10 volts, 36 milliamperes, and so on. [Mr. Lampkin's curves are theoretical and must not be assumed to be those of a 1 12 or 171 type tube, whose characteristics might seem similar. They are the socalled "static" curves, and really for purposes of this argument should be "dynamic" instead. However, since the plate currents for these curves are so low, and because dynamic curves would differ for each load in the plate circuit — ■ each transmitted audio frequency, for example, if the tube were working into a loud speaker — it has been thought much simpler to use static curves as shown. The argument would be the same in any case; overloading distortion could be predicted from inspection of dynamic curves, due to the fact that these curves are not exactly linear, although they are much straighter than the static curves shown. — The Editor.] Suppose this tube were placed in an amplifier, with 90 volts on the plate and a C battery of 22 volts. The steady plate current would be 6.3 milliamperes, as indicated on the curve by the point "A." The audio-frequency signal voltages, when impressed on the grid of the tube, would alternately add to and subtract from the bias voltage. In Fig. 2 the curve is redrawn for 90 volts battery, and the point "A" is indicated. About this bias of 22 volts swings the audio voltage, alternately positive and negative, or to right and left, as shown along the extended vertical line. For each value of the resultant voltage on the grid, that is, the steady bias voltage plus or minus the instantaneous a.c. voltage, there is a corresponding instantaneous value of plate current. The curve drawn through the latter points, along the extended horizontal axis, is the form of the output current. So long as the signal voltages are comparatively low, say plus or minus four volts, the reading on the plate milliammeter will remain invariant. Although the plate current alternately increases and decreases, it must be remembered that the alternations take place one hundred or more times per second, and are much too fast for the ordinary milliammeter needle to follow. If the grid voltage does not swing beyond this plus or minus four volts the plate current on one half the cycle increases as much as it decreases on the next half cycle, so that the average force on the meter needle does not change and the reading stays constant. If a very low audio frequency were impressed on the grid of this tube, for example, a 10 cycle note, the milliammeter needle would actually wobble up and down from its average value, indicating that the instantaneous plate current varied in accordance with the incoming signals. However, when a stronger signal comes through and forces the grid voltage to swing to wider limits, the reading on the milliammeter will change. This greater input voltage may swing from plus 16 volts to minus 16 volts, and would have a graphical representation as in Fig. 3. In this case, when the total grid voltage reaches minus 3 5 volts (that is, minus 22 due the C battery plus minus 13 due the signal), the plate current becomes zero, and remains zero as long as the voltage is more negative than this value. The curve of output current shows clearly the portion of the cycle during which the current is cut off. Such an output, for a strong signal, is decidedly not an exact, enlarged reproduction of the input. The graph of plate current shows that the positive halves of the wave are larger than the negative halves, for the latter have their tops chopped off. On the curve, the horizontal line at 6.3 milliamperes represents the plate-meter reading when no signal is impressed. On the strong signal the average value of the plate current jumps to 9.6 milliamperes, and this deflection is taken up by the meter. Thus the fluctuation of reading on the plate-current meter is coincidental with amplitude distortion, and may be used as an indication of the latter. The remedy for distortion of this sort, where the meter deflects up on strong signals, is to decrease the C-battery voltage. The chopping off of the negative loops of plate current is due to the lower bend in the characteristic curve. Decreasing the C bias moves the working point away from this bend, that is, up the curve, so that the negative grid voltage swings cannot FIG. 2 290