Radio broadcast .. (1922-30)

.RADIO BROADCAST. detector, the straight line action operates in such a way as to make the static noise only 10 times the signal strength. It is obvious that the static will create less disturbance in the latter case. A very important property possessed by the linear detector, and one which is not generally appreciated is that when two signals are simultaneously applied to a detector and the weaker of the two is an undesired signal, but yet is strong enough to be heard as an interfering background, the action of the linear detector is such as to suppress the weaker of the two signals, and to prevent it from being heard! This rather surprising result can be explained most readijy with the aid of the diagrams in Fig. 3 in which A shows the desired signal, while B is the weaker interfering signal having a different frequency from the first, and c, which is A and B added together, represents the voltage actually applied to the detector input. With a straight-line detector the audio-frequency output is proportional to the envelope of the wave in c, resulting in the audio- frequency output shown at D. It is ap_- parent that this output contains no contri- butions introduced by the weaker of the two signals other than the inaudible super- sonic beat note between the two carriers. The total result is that with a straight- line detector the strong signal prevents the weaker superimposed signal of a different frequency from being detected. [This wHI not be the case if the detector characteristic is other than linear. Then the weaker signal will be rectified and the rectification products will be heard even when_the strong signal is present. — Editor.] This is, in effect, increasing the selectivity of the receiver, but has the great advan- tage over selectivity gained by tuned circuits in that there is no sideband trim- ming involved. The full benefit of this increased selec- tivity is not realizable in practical receivers for several reasons. In the first place, no detector has a perfectly straight-line characteristic, and, unless the character- istic is absolutely straight, the weaker signal is not completely suppressed. Also, in order to obtain full suppression of the weaker signal it is necessary that the strong signal always be larger than the weaker one. If the strong signal is modu- lated 100 per cent, there are times when its amplitude goes down to zero, and at these moments, of course, there is no suppression, or rather, what is normally the weaker signal may momentarily sup- press the normally stronger one. The net (A) DESIRED SIGNAL (B) - UNDESIRED WEAKER SIGNAL DESIRED AND UNDESIRED SIGNALS SUPERIMPOSED Supersonic beat Frequency / Audio Frequency 1 • Component (D) WAVE C AFTER LINEAR RECTIFICATION Fig. 3 — Suppression of detection of weaker undesired signal by linear rectification. result is that with the linear detector, as used in practical receivers, strong signals are able to reduce the efficiency with which weak but interfering signals are rectified T o <=> !?TL 0.0001 50,000 Cycle Source Fig. 4 — Circuit for demonstrat- ing how a strong signal can sup- press detection of a weaker one when a linear detector is used. to an extent that materially increases the apparent selectivity of the set, although complete suppression is not obtained. The possibility of suppressing a weak signal by a strong signal of different fre- quency was tested with the aid of the circuit in Fig. 4, in which a grid-leak power detector was used. The test was made by tuning in a moderately weak broadcast signal when no 50,000-cycle current was flowing through the resistance R. With the music coming through satisfactorily as determined by listening in the head phones a 50,000 cycle voltage was superimposed upon the broadcast signal by running a current through the resistance. As soon as this was done the music practically dis- appeared. It was found that two or three volts across the resistance were enough to eliminate substantially all detection of the broadcast signal, although the detector was adjusted to take inputs several times this amount before overloading. It is not necessary to use 50,000 cycles for the sup- pressing frequency, and any frequency, other than one giving an audible beat note with the broadcast signal, would have given the same results. This little experi- ment shows that a strong signal is really able to suppress a weaker signal when the two are superimposed and applied to a linear detector. This comparison of power and weak- signal detection shows that the former is superior in that it introduces less dis- tortion, is a more efficient rectifier, gives less disturbance with strong static im- pulses, and results in an increase in the effective selectivity. The linear power detector is obviously here to stay, and the future will see it used more and more. It has even been suggested that some day the input to the loud speaker will be ob- tained by rectifying a very large radio- frequency signal of perhaps 100 volts, using a vacuum tube, or perhaps a copper- oxide element, without the use of any audio-frequency amplification. Power detection requires more radio- frequency amplification than does the weak-signal detector, and not many years ago this was a real disadvantage. The screen-grid tube has altered the situation, however, by making it comparatively simple to obtain high amplification per stage without trouble from regeneration. Inasmuch as it is still necessary to use the same number of tuned circuits in screen- grid sets as before, in order to obtain the necessary selectivity, the additional radio- frequency is so easy to obtain as to be an advantage. DESIGNING THE POWER-SUPPLY CIRCUIT (Continued from page 48) sound, and accurate limits set for accept- able hum. as this depends upon frequency, type of loud speaker, efficiency of loud speaker, and cabinet resonances. How- ever, after a while the operator will be able to pass very accurately on the ac- ceptability of any particular job. To analyze the hum coming from our set and power pack we can short circuit various tubes, cutting out certain parts of the total hum. For instance, if we short circuit the primary of the last a.f. trans- former all the hum we get is coming from the power tubes alone. If we put a very large by-pass condenser across the grid- bias resistor we can tell whether it is grid ripple or not, and, if it is not, then we can increase the capacity of C? (see Fig. 1). If this helps we know we have excessive plate ripple and either Ci or Cz or both must be increased to secure satisfactory filtering. However, if this does not help then our hum is coming from the filament supply and may be due to unbalanced tubes, or unequal halves of the filament- supply winding. However, it will be found in the great majority of cases, especially when using push-pull, that the hum will be very low when the last a.f. primary is shorted. To locate hum coming by induction from power transformers and choke coils we can remove various tubes and place resistors across the primary of the suc- ceeding transformer. Thus, if we remove the detector tube and place 20,000 ohms across the primary of the first a.f. trans- former, we can tell how much induction there is by rotating the transformer, and if it is serious we can locate the apparatus properly to minimize it. It is obvious that our first a.f. transformer will give more trouble from this source than the second due to higher succeeding amplification. Having finally chosen our capacity values it is only necessary to specify their working and break-down voltages. For instance, we found 613 volts neces- sary across Ct at a 115-volt line. The highest line voltage we might encounter is about 130 volts, and, if no regulator were employed, this would result in an r.m.s. potential of almost 700 volts, or a peak of 980 volts. This, then, is going to be our maximum working voltage and Ci, should be specified for 1200 volts to have a sufficient margin of safety. Of course, we must also check all our filament and plate voltages accurately to make sure our computations were correct. We have now finished our power pack, have a job producing the required volt- ages, giving acceptably low hum, keeping within safe limits of heating and break- down, and costing as little as possible. Our job then, is finished, and we turn our pack over to the production department to do with as they will. 50 • NOVEMBER 1929