Radio Broadcast (May 1928-Apr 1929)

A New Principle in Audio Transformer Design By KENDALL CLOUGH Research Laboratories of Chicago THOSE who go to the market for audio transformers have noted, no doubt, the disparity in size of the devices that are offered. It will also be noted that, as a general rule, the physical size increases as the claims for fidelity of reproduction increase. This article proposes to explain the basic reason for these variations in physical size as well as to describe a new principle in the design of transformers. By the use of this new principle it is possible to produce the ideal audio curve, impossible with the conventional transformer, as well as eliminate a type of distortion that is present in the latter but which is not indicated by its characteristic curve no matter how good it may appear. The circuit of Fig. I shows the representative stage of amplification as ordinarily used in radio sets. The signal voltage, eg, impressed on the grid of the first tube may be of any frequency between 30 and 10,000 cycles corresponding to the range of frequencies concerned in the reproduction of speech and music. The tube on which the voltage, eg., is impressed has an amplification factor of and a plate resistance, rp. In the plate circuit of the tube we have connected a transformer having an inductance, Li, and the turns ratio, N. Consider that the secondary operates into an infinite impedance, an assumption that is justified when the transformer operates into a tube with sufficient C bias. We will modify this assumption later for the high frequencies. fn such a circuit as this, the amplification may be expressed by the equation: N A = amplification ~ where to = 2 X f A = v fey + 1 Li = primary ir ductance This equation first defines amplification as the ratio of the voltage impressed on the second grid circuit to the voltage impressed on the first grid circuit. We note that the amplification is directly proportional to the amplification factor of the tube and the turns ratio of the transformer. This might lead one to believe that almost any degree of amplification could be accomplished by the use of a high-mu tube and a high ratio transformer. This is not the case. Using a transformer having a primary inductance of 44.2 henries and a turns ratio of three, we have, by use of the above equation, plotted in Curve 2, Fig. 2, which shows the amplification at various frequencies when used with a tube having an amplification constant of 8 and a plate resistance of 10,000 ohms. This corresponds approximately to a 201-A tube. It will be noted that the curve is not ideal by any means, for at 30 cycles (the lower limit of the frequencies we are concerned with for the reproduction of music) the amplification has fallen off to 15.5 while THE author of this article is no stranger to Radio Broadcast's readers. He has designed several of the popular receivers we have described, and is well known as an excellent radio engineer. In this article he discloses for the first time his new audio amplifying system on which he has been at work for some time. The advantages of this system are several. It provides an amplifier that is flat from 30 cycles to above 6000, or, if desired, a rising characteristic between 30 and 100 cycles. This is done at less cost than that of a conventional transformer-coupled amplifier, and with somewhat greater voltage gain. At the same time the introduction of harmonic frequencies in the output due to saturation of the core of audio transformers due to the d.c. plate current which flows through them, is avoided. Although we have not seen the data, we understand that a study of this effect at Cruft Laboratory proved that at times the third harmonic in the output of an amplifier in which d.c. flows through iron cores rises as high as 50 per cent, and that no adjustment of bias or plate voltage would eliminate it. Mr. Clough's system prevents such distortion. He is at work upon another article which we hope to present soon. — The Editor. FIG. I the amplification at the high frequencies is 25.0. Now as was suggested, we might believe that we could secure a higher amplification such as is shown in Curve 3, Fig. 2, using a tube having a higher amplification factor, such as 30. Such a curve could be secured if we had a tube with an amplification factor of 30 and plate resistance of 10,000 ohms. This would be a very difficult tube to design and produce, however, and we see by reference to a tube chart that the 240 type ([J. = 30) has a plate resistance of 150,000 ohms. Thus we can compute such a curve as is shown in Curve 4, Fig. 2, using the same transformer as was used in Fie. 2 and the 1000 FREQUENCY CYCLES PER SECOND FIG. 2 '33 new value of plate resistance, i.e. 150,000 ohms. We see in Curve 4, Fig. 2, while the amplification has been improved at the high frequencies, the amplification of the lows is very poor indeed. The nature of the equation above indicates that we could improve the amplification of the bass notes shown in Curve 4, Fig. 2, by an increase in the primary inductance say to 672 henries. All of this seems very simple but there is a practical aspect worthy of our consideration. BETTER RESPONSE WITH INCREASED PRIMARY INDUCTANCE? I N MAKING the latter transformer we would 1 require many more primary turns than we did on the 44.2 henry design, in fact about 3.9 times as many. In addition, to keep the ratio 3:1 we would need 3.9 times as many secondary turns. We assumed that the secondary operated into an open circuit. Now, if we attempted to build the transformer with a primary inductance of 672 henries we would find that the secondary turns would have a large self-capacity and that at the higher frequencies we could no longer assume operation into an open circuit because the self-capacity would practically short-circuit the secondary causing the curve to slump off as is shown in Curve 5, Fig. 2. This drop in amplification at the higher frequencies is not noticeable to the ear so long as the frequencies below 4000 or 5000 cycles are not materially impaired. But we see that Curve 5, Fig. 2, begins to drop badly at 1000 cycles so that the actual transformer fails to approach the ideal at either the high or the low frequencies. This is the reason why a transformer for the 240 tube has never appeared on the market. Referring to the data that was used to prepare the Curve 2, Fig. 2, we might consider the possibility of producing a transformer which would have a high turns ratio in order to raise the amplification. Thus let us assume that we desire a transformer of 5:1 ratio. This would necessitate increasing the number of secondary turns by the ratio of 5:3 which would cause the amplification curve to appear theoretically as in Curve 6, Fig. 2. However, in attempting this, we find that when the transformer is actually measured the curve would drop off at the higher frequencies as shown in Curve 5, Fig. 2, due to the fact that the secondary distributed capacity would reduce the secondary voltage at the high frequencies in the same manner as was shown in connection with the curves on the 240 tube. Practical experience indicates that a transformer with the degree of bass amplification shown in Curve 5, Fig. 2 cannot be built at ratios in excess of 4:1 without impairing the high frequencies more than the average ear will allow. Refer to Curve 2, Fig. 2. Let us consider what 10 000