Radio Broadcast (May 1929-Apr 1930)

Data on Making Measurements NOTES ON A. F. TRANSFORMER DESIGN IT IS the purpose of this paper to present in a simple, practical form the general considerations involved in the design of audiofrequency transfer apparatus. The, curves accompanying this article are first hand, except when specified otherwise, are solely for illustrative purposes, and are intended to show comparative rather than absolute values. They have been taken in the radio laboratory at Columbia University with the cooperation of the Engineering Department of the Pacent Electric Company. Measurement of Output Voltages rr,HE measurement of output voltage of volt-1 age transformers must be done without excessive loading. This eliminates ordinary a.c. instruments. The electrostatic voltmeter might be used but the voltages to be measured are normally too small and the internal capacity of the meter too large. The vacuum-tube voltmeter in some of its many circuit variations is very satisfactory. Regardless of the accuracy of the measuring voltmeter or its sensitivity, true curves cannot be obtained unless actual operating conditions obtain, or are simulated. With normal grid bias on an amplifier tube the loading of the transformer by grid conductance is far greater than it would be with a onetube V.T. voltmeter load. The grid conductance changes sufficiently with a.c. voltage applied to make it quite desirable to specify actual loading conditions, when making the test. The one-tube voltmeter circuit falls down from the very start. The diagram in Fig. 1 shows a circuit which has been used with most satisfactory results by the writer. To avoid any error in calibration of the voltmeter, a comparison method is used. This permits of a very high degree of accuracy without extreme care in setting the input voltage. An accuracy of 1.0 per cent, is easily obtainable. In the input circuit, the applied voltage, d.c, and equivalent tube resistance may be adjusted to simulate any type of tube. The current, /, through /?i is set to give the desired voltage on the primary of the transformer. With the switch down the reading at Ij> is noted. Then the switch is flipped up and R adjusted until the same reading is obtained. By rapid flipping of the switch and a low-inertia meter a very accurate setting may be made. Figs. 3, 4, 6, and 8 show the effect of variation in applied a.c, d.c, load, both resistive and capacitive, and grid bias of the loading tube on the characteristic curve. They supply ample evidence to justify the specification of measurement under actual load conditions. By J. KELLY JOHNSON Fig. 8 was taken with a single-tube voltmeter using the method of comparisons which greatly increases the obtainable accuracy. The high bias used rendered the effect of grid conductance negligible. The improvement in the transformer characteristic with increase of applied voltage is quite pronounced. To make characteristic curves on transformers com In this article are presented considerable valuable data on audio transformers, showing how the conditions under which the transformer is measured effects its characteristics. The frequency characteristic, for example, is shown to vary widely with the load into which the device works, the value of the a.c. measuring voltage, and the d.c. current in the primary. The definite quantitative figures given by Mr. Johnson will be found very useful. — The Editor. parable, therefore, the same a.c. voltage must have been applied to their primaries. In Fig. 3 is shown the effect produced by d.c passing through the primary coil. The dotted curves show that a transformer with a heavy silicon-steel core was not much affected, while the curve for the high-permeability alloy core transformer was badly damaged by the d.c. The value of current used was about normal for a d.c. -type tube and considerably less than those encountered with the new a.c. types. The transformers were recommended by their respective manufacturers for use in the same point in the circuit. Fig. 4 shows the effect of a resistive load placed on the secondary of a transformer. The peak voltage at resonance is cut down appreciably by even a load of very low conductance. The next , Fig. 6, shows the effect of grid conductance in flattening out the transformer curve. Although the grid did not swing positive, the conductance caused sufficient UX-201 regulation to flatten the curves as shown. It must be noted that the previously mentioned effect of increased voltage applied on the primary raised the low-frequency portions of the 0.5 and 1.0-volt curves and lowered their peaks. Special Test Apparatus MEASUREMENTS on power audiofrequency transformers may be made using the same apparatus as described above. Ordinarily, however, the actual loading conditions require the employment of higher voltages than are conveniently measured with this layout. The fact that these are power-transfer instruments immediately suggests the use of standard alternating current voltmeters, ammeters, and wattmeters. But the power required to operate the average a.c. ammeter is often as high as 5.0 watts, and the voltmeters are but slightly more sensitive. Wattmeters, partaking of the characteristics of both voltmeters and ammeters, consume about twice this power. Thus, it is seen that ordinarily the total output of the transformer under normal loading is insufficient to operate the meter alone. In addition, the inductance of the iron vane, or dynamometer type of meter movement is quite appreciable, and would give very great variation in indicated voltage values with frequency. The ordinary meter is not accurate beyond one-hundredtwenty cycles. The circuit used in Fig. 5 is probably the simplest and the first to suggest itself. It requires a thermocouple meter and resistor to fix the input voltage, and another to measure the output voltage. This makes necessary a "mark" reading or requires an oscillator of a high degree of stability and high power output. The V. T. voltmeter comparison method can be used to advantage, a single-tube voltmeter with a direct indicator in the plate being sufficient, as the output voltage of the transformer is high, and the added loading of the measuring device low enough to be neglected. It is possible that an inverted V, T. voltmeter might be evolved which would simplify measurements. As regards the simulation of loads for power transformers, the curves of Fig. 5 show how a characteristic is altered by the secondary load. At the low frequencies the load reflected from the loud speaker would give a very lowimpedance load on the tube, while at high frequencies the opposite is true. A resistive load, on the other hand, will not have unpredictable resonance points as shown on the loud speaker Fig. 1. 1 0,000 i r / r \ 1 1 ' 0.1 ' c V ■ cc RE Sj 1 ^TU RATION EFFEC • \ _Y ,|:. 1 I,,.'. — HigtwMlloy Core - — Iron Core \ \ 300 500 FREQUENCY 3000 5000 10,000 1 1 INPUT TRANSFORMER Effect of Resistance Loading at Peak Frequency )0G~>, MEGOHMS LOAD Fig. 2. Fig. 3. Fig. 4. • may, 1929 page 34 •