Radio Broadcast (May 1929-Apr 1930)

-Z RADIO BROADCAST 70 60 W50[ §40 g30 Liar 20 10 0, 0.4 0.8 POUNDS 1.2 1.6 2.0 2.4 2.8 — ' * '50 mrr fundar fd.vern nental t ier aero ank eirc ss 50 mm: dverni uit across crystal i ( height on top plate of_ crystal holder 10 20 30 40 50 VERNIER COND. SETTING Fig. 8 60 70 untouched during a test run, while a variable on the other was changed. Then the variation in beat note between the two crystals was due solely to the change in the variable on one crystal. Since variation in frequency of the second harmonic was measured, the values given are twice the variation that occurred in the fundamental frequency. Because the crystal oscillators were to be supplied with operating voltages from a 60cycle a.c. source, the parallel-cut and perpendicular-cut types were compared as to frequency variation when the line voltage changed. The oscillator tube was supplied with 5 volts a.c. on the filament, and 200 volts d.c. from a B-power unit. Both these voltages had as their source an a.c. supply which could be varied from 110 to 130 volts. In the case of the parallel-cut crystal this change in line voltage caused an average frequency shift of 20 cycles. The same test applied to the perpendicular-cut crystal caused the frequency to change 35 cycles in two or three seconds, and then drift slowly to a total change of 92 cycles in some three minutes — again average values of several trials. This comparison, coupled with the fact that the parallel-cut plates gave a higher radio-frequency output, was the reason for confining subsequent measurements to this type of crystal. 1 Variations Possible IN FIGS. 8 and 9 are shown the frequency variations that may be obtained by three means of intentional tuning. In the dash curve a small semi-circular plate condenser of 50 mmfd. maximum capacity was connected across the crystal itself. Variation of this capacity caused a nearly proportional change in crystal frequency. The total variation in frequency when the capacity went from 0 to 30 mmfd. was 450 cycles. At the same time, however, the second harmonic output voltage dropped from 67 to 38 volts, peak. The variation of frequency resulting when tbe vernier condenser was placed across the fundamental tank circuit was determined, and is given in the dot-dash curve. The total change in frequency with timing of the main condenser, from the point where the oscillations started to where they broke, was some 800 cycles. (For the perpendicular-cut crystal, this figure was 250 cycles.) The frequency varied in much the same manner as did the fundamental voltage, in curve Eu Fig. 5. That is, the change in frequency was slight at first, then became greater until, at the point of oscillation cessation, the frequency change per increment of setting was a maximum. The dot-dash curve in Figs. 8 and 9, then, is a small portion taken from this larger curve that would be obtained were the main condenser varied. The hump in the second-harmonic output-voltage curve is also a portion of the larger curve E2, Fig. 5. Variation of pressure on the crystal surface constitutes another means of shifting the crystal frequency. The total change experience when the weight on the top plate of the crystal holder was run up to 2.8 pounds was 845 cycles. This represents a greater change in frequency for a given drop in output voltage than either of the other methods of timing. The relation between pressure and frequency increment is nearly linear, so that tuning in this manner is better than, for instance, the use of a 50-mmfd. vernier across the fundamental tank circuit. Another ad Weight on top plate of rt>6 mmfd.vernieracross it ■-^ 50 mmfd.vernier *v CLK. M 2.0 10 2.4 60 2.8 70 0.4 0.8 1.2 1.6 POUNDS 20 30 40 50 VERNIER COND. SETTING Fig. 9 vantage incurred by the use of pressure on the crystal is that it makes for stability. Jars or bumps suffered by the crystal and mounting are not so likely to disarrange the holder. The latter point was one of the most troublesome in bringing and keeping the two crystals to zero beat. For each slightly different position of the holder plate, the crystal took up a new oscillation frequency that was possibly two cycles, or as much as two thousand cycles removed from the original. It is physically impossible to grind the two surfaces of an 80-meter crystal parallel within five or ten "cycles." A ten thousandth of an inch on such a crystal represents some 13,000 cycles. It seems that for each new position of the plate a new portion of the crystal has a major effect on the frequency. It was neces 5? 20 5 t>40 60 50 100 150 D.C. PLATE VOLTAGE 200 250 Fig. 10 sary to place and replace the top plate of the mounting till the frequency of one crystal came within a hundred or so cycles of that of the other. A holder designed to maintain the relative positions of the crystal and plate within close limits; or one designed to give intimate electrical contact over both crystal surfaces, should aid in overcoming the difficulty. In the case of the longer-wave crystals, it is possible to operate them with no contact whatsoever between plates and crystal. Maintaining Frequency Constant rI^HE frequency shift which occurred when J both the filament and plate voltages on the oscillator tube were changed has been mentioned. In Fig. 10 the change in frequency that is due to plate voltage variation alone is shown. To maintain the frequency constant within one cycle the plate voltage must not be allowed to vary more than 1.5 per cent. To hold the frequency invariant to the same degree, temperature variation cannot exceed 0.66 degrees Centigrade. Thus, the major problem in frequency stabilization is that of temperature control. Other factors such as operating voltages and circuit constants can be fixed so that frequency shift due to them becomes negligible when compared with that due to temperature variation. Fig. 7 contains the points, plotted, which were obtained when the temperature inside the container for the crystal mounting was increased slowly by means of a resistance heater. Readings of temperature and frequency were taken at intervals of five minutes over the period of 45 minutes which the curve data cover. Over the straight-line portion of the curve the frequency changed 15 cycles per degree Centigrade. The lower bend in the curve furnishes another example of the discontinuities encountered in crystal operation. As the temperature slowly approached 35° the frequency began to change more rapidly, and at that point the beat note suddenly jumped to 2973 cycles, where it had been 429 cycles previously. For no apparent reason the frequency shifted 2544 cycles. Thus, not only must the crystal temperature be held within extremely close limits, but the absolute value of the temperature is important. It is not safe to calibrate the crystal in one range of temperature, and assume that the frequency in another range can be obtained by application of the temperature coefficient of frequency. The author's crystal-oscillator unit • may, 1929 The vacuum-tube voltmeter shown schematically in Fig. 2 . page 25 •