Radio Broadcast (May 1928-Apr 1929)

Answering the Question ARE FILTERS NEEDED IN A. F. AMPLIFIERS? By KEITH HENNEY Director of the Laboratory OSTENSIBLY, a two-stage transformercoupled audio amplifier is a perfectly simple assembly of apparatus for the set-builder to put together and operate. All that it is necessary to do is to mount the transformers, sockets, binding posts, and C-bias resistors, if a.c. operated, on a baseboard, and wire them up. But, is it as simple as this? Suppose you have the manufacturers' curve on a single transformer, giving its frequency characteristic, have you any assurance that the complete amplifier will have such a curve, or will it have additional humps and hollows in it, and will it tend to sing at some high frequency, or will it "motorboat" if you try to run it on a none-too-good B supply? For years George Crom of Amertran has been trying to educate experimenters up to the point where they will "filter" their amplifiers. The use of such filters keeps the a.c. where it belongs, and prevents it from roaming through the B-supply unit where it would become mixed with a.c. from other circuits. According to Mr. Crom, the characteristic of a two-stage amplifier will be that of a single stage squared provided — and only provided — it is well filtered. Just what does this mean? The circuit of a well-filtered Amertran DeLuxe amplifier is shown in Fig. 3 and a picture of the unit will be found in Fig. 4. A list of parts used in constructing it in the Laboratory will be found at the end of this article. To determine the value of Mr. Crom's suggestions regarding filtering, we took this amplifier into the Laboratory and measured its characteristic by putting constant voltages on the input through 12,000 ohms — to simulate the detector out of which it ordinarily works — at various frequencies, and measuring the current into a non-inductive output resistor of 4000 ohms. According to tube experts, the greatest amount of undistorted power output from a 2000-ohm tube (cx-301a) will be secured when the load into which it works is equal to 4000 ohms, and while the loud speaker into which the amplifier works will not have a constant impedance equal to 4000 ohms at all frequencies, we cannot hope to simulate it + B Det. Fig.. 1 — When filtering is employed in the detector plate circuit, two wires are needed to connect the detector with the amplifier input George Crom of the Amertran Company has maintained for a long time — he first stated his position in an article in Radio Broadcast for October, 1925 — that audio amplifiers and their filter circuits could not be considered separately. It is fair to say that his thoughts have not been as generally appreciated as they might have been. We went into our Laboratory and this interesting article, with accurate and quantitative data is the result. It all means simply this: it is not enough to build your amplifier from the best of units, you must also use proper filtering. From the information given here, everyone can go over his own amplifier and apply the suggestions given. With audio amplifiers and reproducers now covering a really, satisfactory scope, real fidelity can be attained, where it was not possible several years ago. This article is the first to give, we believe, any considerable quantitative data on this important subject. — The Editor. better than this. If the amplifier itself has a good characteristic when operated into a resistance load of the proper value, the problem is then up to the loud speaker designers to make a unit that will give equal results. In these tests we were not concerned with overall amplification nor with the power output, but for simplicity of measurement we calibrated our output current meter — a Weston thermocouple — in db (tu) up and down from 5 mA. which, into 4000 ohms, is equal to an output of 100 milliwatts. How Tests Were Made To prevent any of the a.c. in the primary of the first transformer from entering into the B supply we placed a 1.0-mfd. condenser, Gi, as shown which provided a low-impedance path as compared with the 50,000-ohm resistor. The grid circuit of the first tube was filtered by means of a high series resistance, Ri, and a low-reactance condenser, C2. A.C. voltages appearing across the 2000-ohm resistor, R3, were not able to enter the grid circuit, first because of the high resistance in series with it and secondly because the lower end of the audio transformer secondary is practically short circuited to the filament, so far as a.c. is concerned, by the condenser C2. Similarly, a.c. currents in the 227 tube circuit were kept out of the B supply by filtering the plate circuit of the first audio tube by means of the condenser C3 and the resistor R4. This means that the a.c. voltages across R3 were very small. The grid circuit of the power tube was filtered in the same manner as the 227, and to keep the a.c. currents of the last tube from wandering around through the B supply the loud speaker was connected directly to the center-tap of the filament of the 171. All the a.c. currents that entered the B supply from the last tube, were those which passed through the choke Li, and these were very small as compared with the currents going through the loud speaker. That was the purpose of the choke, to keep the a.c. currents going through the loud speaker and not through the choke. The purpose of the condenser, Ce, was to keep the d.c. flowing through the choke, and not through the loud speaker. The above paragraphs describe how our apparatus was arranged. Now what happened when we placed tones on the input, and measured them in the output? The amplifier was flat from 100 to 8000 cycles (curve A, Fig. 5), it. went down about 1.0 db at 60 cycles, and up about 8.6 db at 10,000 cycles where the capacity across the secondary resonated with r I "'HE source of tones was a beat-frequency J oscillator which would function down to 60 cycles easily; its output was impressed across 10,000 ohms and 500 ohms in series, and the voltage drop across the 500 ohms (General Radio resistance box) was impressed on the amplifier. The B supply was a Majestic power unit which used a gaseous rectifier tube, and which had been found to have an output impedance respresentative of all such devices. To obtain plate potentials lower than 180 volts — which was applied to the 171 tube — we used resistors of 25,000 and 50,000 ohms, respectively, at R4 and Ri. Since 12,000 ohms is much lower than the d.c. resistance of a detector, we placed only 22.5 volts on this tube so the current through the primary of the first DeLuxe transformer was about 1.0 milliampere. The 227 tube (a deForest) was supplied with 90 volts on the plate and about 5 volts on the grid. The 171 tube (a Raytheon) had 180 and 40.5 volts on its plate and grid, respectively. Fig. 2 — When an output transformer is employed the filter system illustrated above should be used in the plate circuit of the power tube february, 1929 . page 250 •