Radio Broadcast (May 1928-Apr 1929)

A NEW AUDIO SYSTEM FOR DYNAMIC SPEAKER REPRODUCTION 3] 57 tu. This does not include the detector audio gain. The use of the same resonating idea in the second transformer is unnecessary, since we' are interested only in obtaining a perfectly flat characteristic with a very large voltage gain. One difficulty is that the d.c. voltage drop through a shunt resistance would be enormous unless small values were used. For example, 3 milliamperes through 100,000 ohms would cause a 300 volt drop. The heating effect would be troublesome also since nearly one watt would have to be dissipated in the form of heat. If low values — less than 30,000 ohms — were used, the audio-frequency loss through the shunt resistance would be a large percentage of the gain of the transformer. In that case an ordinary low-ratio, high-quality transformer might as well have been used. THE OUTPUT TRANSFORMER I N CONNECTION with the complete amplifier * system, the last, and perhaps the most important link in the chain, is the output device. It was found that the best output transformers on the market were quite unsatisfactory below 200 cycles per second when the full plate current of a type 250 power tube was flowing through the primary. Core saturation takes place and even with a good-sized air-gap in the core few notes below 200 cycles reach the loud speaker. However, an output transformer is quite necessary with a dynamic loud speaker in order to match properly the load to the power tube. By keeping the d.c. current out of the output transformer by means of a choke coil, L2, and condenser, Q, as shown in Fig. 3, the transformer may be made very excellent for even the very low frequencies. The condenser, C2, may be made to resonate with the transformer primary to compensate for the loss due to the shunt choke coil, L2, and series condenser reactance on the low frequencies. Another advantage is that the audiofrequency path is through the condenser and transformer primary back to filament instead of through the power-supply unit. This prevents audio feedback to preceding stages. The result is a very stable high-gain amplifying system which has more gain than even an audio amplifier using a screen-grid tube in the first stage. The use of a screen-grid tube, even as a space-charge amplifier, means that impedance coupling should be used between that tube and the power tube, which gives, roughly figuring, i X 3 X 50X 1X4X5= 100 volts across the output device. This assumes a J-volt detector output, a 3 to 1 transformer, a gain of 50 in the screen-grid tube, a gain of 4 in the power tube and that § of this voltage appears across the output device and f across the power tube plate resistance. For the amplifier described, using 45 and 6| to 1 ratio transformers and tubes giving a gain of 8 and 4, the total voltage appearing across the output would be iX4iX8X 6| X 4 X 5 = 1 56. The ordinary amplifier using 3 to i transformers and tubes giving the same gains of 8 and 4 would give an output of iX3X8X3X4Xj = 48. Ratio 4)5:1 300 A MODIFICATION FOR A.C. OPERATION THIS amplifier having such an excellent frequency characteristic should be used preferably with d.c. filament tubes in the r.f. stages, detector and first FIG. 3. SCHEMATIC DIAGRAM OF COMPLETE REMLER AMPLIFIER audio stage, since the amplification is so high at 60 cycles. However, experiments with special amplifiers cutting off very sharply just above 60 cycles per second, have shown that the hum when using a.c. tubes may be made nearly negligible. The reason for cutting off above 60 cycles is apparent when it is remembered that ordinary loud speakers are excellent harmonic producers on low frequencies. The use of raw a.c. on the filaments of the tubes causes a 60-cycle component to be impressed in the grid circuits with the result that if the amplifier system is good as low as 60 cycles, this frequency will be amplified and reproduced by the speaker. The speaker, in case it is a cone or small horn will reproduce it as 120 cycle and higher tones. The point is that by not amplifying the 60 cycle component, practically all of the composite a.c. hum in the loud speaker is eliminated. Immediately the idea of resonating the primary of one of the transformers at about 80 cycles, was used to make the complete amplifier cut off sharply just above 60 cycles per second. The cutoff is a great many times sharper below resonance than when the transformer itself is made to have a fairly good audio response. The audio voltage drop across the resonating condenser increases as the frequency becomes less, giving a very sharp cut-off belcfw resonance. By designing the circuit constants properly, one transformer may be made to compensate for the other as shown in Fig. 4 from about 5000 or 6000 cycles down to at least 80 cycles per second. This makes an ideal arrangement for receiving sets using a.c. _. INI Over-all char. icteris lie ft r amplifier \ i V1ERS WITH SHARP CUT-OFF ABOVE 60 CYC Detector: RP 2000 ohms First audio tube Rp 5000 ohms.Ip 6ma Power tube CX-371 A rRANSFOR »^ First-tit age tra nsfor 1{ r fia 10 3%:' \ \ -^Second-st" ;e traru form "erk — r I \ FREOUENCY, CYCLES PER SECOND FIG. 4. RESPONSE CURVES OBTAINED WITH 3^ TO I RATIO TRANSFORMERS tubes, since the low notes are well reproduced when a good dynamic speaker is used. Nearly 95 per cent, of the very low notes such as those from an organ, as played over the radio, are between 80 and 200 cycles per second. Very few people realize this and most people will quite willingly swear that they hear 30 to 50-cycle notes in their radio sets when in nearly all cases their loud speakers and amplifiers will not reproduce anything below 100 cycles per second. Another advantage of cutting off sharply a little below 100 cycles is that a smaller power tube may be used, such as a 171 tube. The apparent room volume of sound with the a.c. system will be about the same as with the first d.c. amplifier system developed, since in this case the lowest note reproduced will be about 70 cycles as against 25 or 30 cycles in the d.c. system. This makes more power available for the tones which are reproduced. Incidentally, the cost of manufacturing such transformers is less. Most a.c. set manufacturers use audio transformers which will not pass 60-cycle signals, in order to minimize a.c. hum and in so doing generally lose in efficiency up to 300 or 400 cycles. This is bound to happen because it is not possible to obtain a really sharp low-frequency cut-off using ordinary transformers. The resonant primary principle should go far towards solving the a.c. hum problem in a.c. receivers. It will be noticed that only the transformers working out of the detector tube in both the d.c. and a.c. systems have resonated primaries. This arrangement keeps the d.c. plate current out of the primaries and so prevents core saturation and generation of harmonics due to this effect. By the use of very large cores of nickel-steel alloy, the chances of core saturation from d.c. are practically eliminated. If small cores are used and large primaries, that is, a large number of turns, core saturation may take place with bad distortion effects. The transformers described are built along ample lines to overcome the possibility of trouble from core satuation. The possibilities of audio amplifier systems using the low-frequency resonated primary principle are many and it is quite likely that this scheme will be widely used to make better audio amplifier systems. It is the engineer's problem to forever strive towards perfection, never reaching it but always advancing; and this design, it is hoped, is a step in that direction. 000 o o Q 000 in %o 50 30