Radio Broadcast (May 1928-Apr 1929)

196 RADIO BROADCAST JANUARY, 1929 minimum sound variation perceptible to the average ear in a musical selection, a 2 ru variation may be taken as a very conservative limit on allowable distortion. Distortion becomes more serious at 18 to 22 watts, although the average person will not be disagreeably affected even at such an overload. While Fig. 1 is the power'output curve for the final push-pull stage selected, it was actually taken on the whole amplifier, so that it is also a measure of such overloading as might occur in preceding stages. ANALYSIS OF AMPLIFIER AN ANALYSIS indicated that the signal voltage applied to each 250 tube at the 14.35-watt point on the curve of Fig. 3 was 90 volts. (The advantage of the push-pull stage is here well demonstrated — the grids were actually 10 volts positive over the 80-volt bias, yet distortion which would have been serious with such an overload on one tube has not become perceptible to most ears with the push-pull stage.) With go volts required at the 250 grids, and allowing conservatively for a 0.2 volt-signal at the input of the amplifier the voltage gain needed to develop 90 volts from a 0.2 volt-signal would be 90 0.2 = 450 times. This is a conservative figure, since the averge detector tube will turn out 0.3 to 0.4-signal volts, and average pick-up unit will deliver as much as 1 to 2 volts. Further measurements indicated that thestandard Clough system transformers (described in July, 1928, Radio Brocadast), and one 226-type tube would give a gain of 120.5, obviously insufficient to operate the push-pull stage to capacity. At this point, one 226-type tube was found to be unable to operate the push-pull stage to capacity without overloading. A iyiA-type tube preceding the push-pull stage was tried with considerable success, as well as two 226-type tubes in pushpull. The latter were found most desirable, due to the simplification of filtration and isolation that the push-pull stage permitted, as well as its greater undistorted output. Suitable transformers of the Clough type were then designed to feed from one standard amplifier tube into a push-pull stage, and to feed from one push-pull stage into a second push-pull stage. These transformers were provided with a low-frequency cut-off below a hump ranging from 65 to 200 cycles and with a flat curve up to about 8000 cycles. (Incidentally, they will be welcomed by the many fans who have written to the designer asking for just such transformers.) An amplifier was then set up and measured, using one 4.3 : 1 -ratio transformer feeding into a 227-type tube, one 1.75:1 push-pull input transformer feeding into two 226-type tubes in push-pull, and one 1.75:1 interstage push-pull transformer feeding from the 226-type tubes to the 250-type tubes. The output of the 250-type tubes was fed, through the specially designed adjustable output impedance previously mentioned, into several different speakers, and measurements were made. Fig. 2 shows the result in the form of an overall gain-frequency characteristic from input to power tube grids. The desirable bass hump is present, and above 3000 cycles common coupling through the power supply produced a second hump which was left in the amplifier to compensate the sideband cutting in radio reception. Many observers liked the effect produced by the rise at high frequencies but if the writer's dislike of this effect is shared by others, a needle-scratch filter across a record pick-up, or a 0.0001 5-mfd. condenser across the input transformer secondary, will flatten it out effectively. This hump would not appear in battery operation, and its effect consequently must not be exaggerated by the reader who has previously been shown only curves taken on battery-operated amplifiers, which curves would not show regeneration almost sure to develop where standard power-supply units were substituted for batteries. The curve of Fig. 2 is not a "hand-picked" laboratory product — it is a true measure of performance at signal voltages developing actual operating output powers in actual loud speakers. A.C. OPERATION A DAPT1NG a high-gain amplifier giving good response at 60 cycles to a.c. operation was no easy task. Fortunately, two push-pull stages simplified the process, but the high overall gain made special precautions necessary for the input stage. The B supply for the push-pull output stage was found to need no filtering, while a resistance-capacity filter in the B wire was adequate for the intermediate push-pull stage. Filament balances were non-critical on both-pushpull stages. Such good fortune did not hold for the input stage. A very high inductance choke, in an unusual resistance-inductance-capacity filter proved necessary for B and C supply, and in addition the input transformer had to be oriented to minimize induction from the powersupply transformer which was 18" away. In the final models a.c. hum was reduced (when using airchrome dynamic or cone-type loud speakers) to a point where it was not objectionable for home use. Final models, operated with one dynamic speaker placed at an open window on a crowded boulevard, provided understandable speech and good music over traffic noise a city block away. Twelve air-column speakers distributed about three floors gave such effective coverage of 30,000 square feet of factory floor space that one could not hear conversation at normal speaking volume. The assembly of the amplifier is well illustrated in the accompanying picture and the physical layout will be found to follow very closely the general arrangement of the' schematic wiring diagram in that the amplifier progresses from left to right with the power-supply apparatus at the extreme right of the wooden baseboard. The 255 transformer is not screwed directly down to the baseboard, but should be wired into circuit loosely so that it can be adjusted for the minimum-hum position in actual operating tests. After its proper position has been determined (and outlined upon the baseboard with a pencil), it is clamped down by means of the steel tie-bar and two of the short-threaded brass rods with their nuts, the holes for them in the baseboard being counterbored. The wiring of the amplifier is comparatively simple and is accomplished by using the flexible hook-up wire cut to proper lengths with insulation pushed back and ends soldered to proper soldering lugs or fastened directly under tube socket terminal screws. Amplifier grid and plate leads should be isolated as far as possible from each other and from other wiring, and can be made quite short due to the layout of parts. All filament and power wiring should preferably be run in a common cable as far as possible, which may be laced with waxed shoemaker's thread after testing. The two loud speaker connections to the S-M 248 universal output choke should be terminated in battery clips so that they may be moved about to the different groups of soldering lugs on the choke in preliminary tests. list of apparatus THE following is a complete list of the apparatus employed in the construction of the power amplifier described in this article: Ci, C2, C3, C4 Potter condenser bank, type 673; Li One S-M Universal output choke, type 248; Ri One Carter potentiometer, type AP-15; R2 One Yaxley resistor, 2000-ohm, type 72000; Ri, Ri Two Frost tapped resistors, type FT-64; R3, R4 Two S-M resistors, type 659; R7, Rs Two Polymet resistors, 750-ohm; S! One S-M tube socket, type 512; to S7 Six S-M tube sockets, type 511; Ti, T4 Two S-M a. f. transformers, type 255; T2 One S-M push-pull transformer, type 257; T3 One S-M push-pull transformer, interstagetype, type 227; T6 One S-M filament transformer, type 247; T6 One S-M power transformer, type 328; One phone cord and plug, five-foot; One roll of S-M hook-up wire, type 818; Five Fahnestock clips; One S-M wooden chassis, 2i-tV' x fV" x J". The total cost of the apparatus is $93.36. FIG. 3. COMPLETE DIAGRAM OF PUBLIC-ADDRESS AMPLIFIER