Radio broadcast .. (1922-30)

.RADIO BROADCAST. tortion in the loud speaker can be reduced greatly by application of bias voltages which are large in comparison with the varying audio-frequency voltages supplied by the amplifier. Electrical Efficiency The method of test and the results ob- tained will now be discussed. For the particular case of a loud speaker, efficiency, as it is generally defined, is not of the greatest interest. Since it is the duty of a loud speaker to utilize all the available power that can be supplied to it by the input circuit, a ratio of output watts to watts input cannot adequately describe its worth. Therefore, in testing loud speak- ers, efficiency is defined as the ratio of the sound output in watts to the maximum power in watts the supply circuit is cap- able of delivering under the conditions of optimum impedance. This efficiency ratio in turn is usually expressed in Deci- bels, or ten times the logarithm to the base ten of the ratio of the output watts to the total available input watts. In order to test the efficiency of the -loud speaker, the authors devised the circuit shown in Fig. 6. The pro- cedure in making measurements was as follows: The oscillator was first set at a given frequency, as for example, 500 cycles. Switch A was thrown to position (1) and Switch B also to position (1). The alternating com- ponent of the voltage drop Ei, across the non-inductive resistor was read with the vacuum-tube voltmeter. The total power avail- able from the amplifier was then expressed by the equation, P, = E, 2 X K where K is a constant. Switches A and B were then thrown to position (2) and the sound-power output of the loud speaker was then expressed as: P 2 = Ez 2 X X where X is a variable function, depending upon the frequency only if the volume of the sound issuing from the loud speaker is kept fairly constant by using larger or smaller inputs. The re- sponse of the loud speaker in DB was then expressed by the equation: a high-quality modern electrodynamic loud speaker was placed in the same position as the electrostatic and the readings repeated over the frequency range of 200 to 6000 cycles—enough readings being taken to make an accurate curve. By plotting the POSITION 1 /•POSITION 2 curves of both electrodynamic and the electrostatic loud speakers on the same sheet, the difference in their responses in DB at any ordinate may be read off the curve, since both curves are raised or lowered at any ordinate by the same amount, namely 10 logiuX/K. Difficulties will be encountered, however, if attempts are made to find the difference in response R = 10 Lo glo = 10 Lo glo i X - 20 Lo glo 10 Log 10 Since the value of the term 10 Log 10 X/K depends upon the frequency only (assum- ing a reasonably constant sound output), it will be seen that the response curve of the loud speaker will be raised or lowered at any ordinate by the value of 10 Logio X/K. Readingswerethustakenover the fre- quency range of from 200 to 6000 cycles, and a curve of response ver- sus frequency as abscissas was plotted. Response Characteristics Since the actual power rep- resented by the sound coming from the loud speaker is only a few microwatts, it is ex- ceedingly difficult to calibrate the apparatus so that absolute , values can be obtained. In order, therefore, to get the response of the electrostatic loud speaker without actu- ally calibrating the apparatus, FREQUENCY- IN Fig. 5 at ordinates where the responses are of different sign. Fig. 3 shows two such curves plotted on the same sheet. It can be seen that the response of the electrostatic loud speaker is better than that of the electro- dynamic by —10 DB. at 200 cycles. At 420 cycles they are equal. At 500 cycles they are again equal. At 1000 cycles the electro- static response is less than the electrody- namic by —6 DB, at 1800 cycles and 3500 cycles the responses are equal, and so on. This set of curves indicates that the response of the electrostatic speaker was uneven as compared to that of the moving-coil type. In Fig 5 we have a set of curves of the response of the electrostatic loud speaker Fig. 6 plotted upon the assumption that the curve of the electrodynamic is a straight line. Curve 1 is for the loud speaker using circuit B (Fig. 2) and a bias potential of 25 volts. Curve 2 is for circuit B and a bias potential of 100 volts, and curve 3 is for the same circuit with a bias of 250 volts. Curve 4 shows the relative response of the electrostatic loud speaker when a 250-volt- bias and circuit C are used. Without going into a detailed discussion of these curves, it can be seen that, in general, the larger the bias potential used, the greater will be the response of the electrostatic loud speaker. It is also seen from curve 4 that changing the resonant frequency of the loud speaker circuit by inserting different values of capacity (0.5 mfd in this case) in series with the loud speaker, changes the shape of the response curve. In general, it is desirable to have the main resonant frequency of the loud speaker circuit considerably above the highest frequency at which it will be used. Volume vs. Bias While a bias voltage of 500 to 600 volts or more is desirable for good re- sults, fair reproduction is obtained with this loud speaker when only 200 or 300 volts is used as a bias. As shown in Fig. 5, the volume of sound delivered by the loud speaker for a given input voltage increases as the bias voltage is in- creased. It is also true that har- monic distortion decreases as the ratio between the bias voltage and the amplifier voltages applied is increased. Although in a.c. ma- chinery, only odd harmonics oc- cur, in the case of the electrostatic loud speaker both odd and even harmonics may be present. In other words, if an initial frequency of 500 cycles is applied to the loud speaker, we may have 1000 cycles, 1500 cycles, or any other frequency or combination of frequencies which are integral multiplies of 500 cycles. This combination of the original frequency with its har- monics, gives rise to a sound wave which is distorted with respect to the original frequency, and the ef- fect is called harmonic distortion. When attempts are made to re- produce music by means of a loud speaker in which harmonic distortion is present, the quality is poor. This is the case when the electrostatic loud speaker is op- erated at low bias potentials. If, however, the bias is increased, this distortion will be reduced. Let the bias potential across the loud speaker be denoted as E and the vary- ing voltage supplied by the amplifier ex- pressed as e Coscot, then the force upon the diaphragm tending to make it vibrate may be expressed approximately by the equation. F = 2KEe Coswt + J Ke 2 + iKe 2 Cos 2<ot It can be seen from this equation, that by increasing E, the harmonic MICROPHONE term ^Ke 2 cos 2o>t becomes less important, and if the bias potential is increased to a high enough value, it will be negligible. The humps in the curves in Fig. 3 are not entirely depen- dent upon the electrical char- acteristics of the loud speaker circuit, but can be seen to depend also upon the various mechanical resonances which may be present in the back plate and the diaphragm of (Concluded on page /22) DECEMBER 1929 • 107