High-frequency performance of transistors has been greatly improved by several modifications of the basic n-p-n junction transistor. External dimensions have been decreased, the p-layer made very thin, and a fourth lead attached. The resulting tetrode transistor is expected to have important applications in high-frequency, broadband transmission systems.
The objectives of transmission research have been profoundly influenced by two new factors within the last decade. One of these is a very rapid increase in the demand for broadband transmission circuits and the other is the advent of the transistor. Within the last ten years, for example, the total bandwidth of the circuits linking New York with Boston has been increased from 5 to 25 megacycles, thus making possible a five-fold increase in the number of simultaneous telephone conversations over these circuits. This trend emphasizes the need for finding better and more economical broadband transmission methods to meet demands of the future. And in the new systems we can expect the transistor to play an important role.
Broadband transmission systems require transistors especially designed for good high frequency performance. Several kinds of transistors, potentially useful for performing different functions in such systems, are being studied at the Laboratories. Among them are the recently announced p-n-i-p, in addition to the junction tetrode transistor.
The junction tetrode transistor is a close relative of the junction transistor recently described in a RECORD article+ by M. Sparks. It differs from the earlier transistor in two respects: first, some of the mechanical dimensions have been made smaller, and second, a fourth electrode has been added.
Consider first the dimensional changes. It is found that high-frequency performance is improved as the germanium bar is made smaller in cross section. Only that part of the bar near the p-layer needs to be made small, however, and in the tetrode this is accomplished by a chemical etch. Figure 2(a) shows a junction triode transistor, and Figure 2(b) represents the same transistor with the center section reduced by etching. In this central part, the transistor measures only about 0.01 inch on a side. The electrical effect of this etching process is a decrease in the output capacitance, which is typically reduced to about one micro-microfarad.
Another dimension of the junction transistor that is very important to high-frequency performance is the thickness of the central p-type layer. The signal current is carried across this layer by the diffusion of electrons. If the frequency of the signal is too high, the electrons tend to arrive at the output side of the layer out of step with each other. As a result, the signal current is not properly transmitter at high frequencies. This aspect of transistor performance improved by making the p-layer thin. New techniques recently worked out in the transistor research areas of the Laboratories have produced p-layers about two ten-thousandths of an inch thick. When the best of this material is used to make transistors, as much as 85 per cent of the signal current is transmitted across the p-layer, even when the frequency is as high as one hundred million cycles. Figure 2(c) shows a junction transistor with a thin p-layer and with a small junction area. In spite of the low collector capacitance and good diffusion properties of such a transistor, it does not amplify at high frequencies. The difficulty comes from internal feedback within the transistor. Signals in the output circuit are fed back into the input circuit through a resistive element which is effectively in series with the connection to the p-layer. Current that enters the transistor through this connection and flows vertically through the thin p-layer encounters a resistance which increases when the path is constricted by making the p-layer thin.
A way of reducing this internal feedback had to be found before thin p-layers could be used to advantage.
In the junction tetrode shown in Figure 2(d), internal feedback is very greatly reduced by making two connections to the p-layer and by applying a few volts of bias between the two. The effect of this is to cause the signal current to flow in a constricted path very near one side of the bar as illustrated. All of the transistor action takes place within a fraction of a thousandth of an inch of the bottom base contact. Current entering the base contact in this case has a very short distance to flow (vertically) in the p-layer, and for this reason encounters very little feedback resistance.
Making connection to a p-layer that is only a few ten thousandths of an inch wide presents some interesting mechanical and electrical problems. Techniques for making this connection, however, have been simplified to such an extent that it seems feasible to do the job quickly and accurately with automatic machinery. Actually a number of transistors have been made on an experimental model of such a machine (Figure 5).
In either the manual or automatic process the connection is made by welding on a carefully formed gold wire. The end of this wire is formed into a "paddle" as shown in Figure 3. The paddle is placed in contact with the germanium bar, and with the thin edge parallel with the junction, it is dragged along the bar until the proper position for bonding is found. A carefully controlled pulse of current is then passed between the gold wire and the germanium. The resulting bond makes a long thin contact with the p-layer.
The most significant electrical property of tetrode transistors is their ability to amplify broadband signals and to operate at very high frequencies. Experimental transistors have been used as amplifiers at frequencies as high as 150 MC and have been made to oscillate at frequencies above 1000 megacycles. They have been used to amplify a 20 Mc band of frequencies centered at 70 Mc, and in this application they can produce about 9 DB of gain per stage.
Aside from improved performance at high frequencies, junction tetrode transistors are electrically rather similar to the familiar junction triode transistor. A feature which is important in many applications is the ability to operate on low values of voltage and current from the power supply. Typical power requirements for producing maximum gain are of the order of 10 volts and one milliampere, but values several times higher or lower than this may be used. As an example of the power used in experimental operating circuits, a 100-mc FM receiver in which there are 6 tetrode transistors requires a total of 10 milliamperes at 12 volts. Less than half of this power is consumed by the transistors. In another application, a 5-mc crystal oscillator in which there is one tetrode transistor requires one-tenth of a milliampere at 3 volts.
The noise generated in junction tetrode transistors is little enough to be acceptable in all but the most critical applications. Typical measured noise figures are, for example, 3 db at 10 mc and 10 db at 70 mc.
Among possible telephone applications of the junction tetrode transistor being studied in the Transmission Research Department, three are of special interest. The first of these is an experimental broadband carrier system that could be used in conjunction with a miniature cable. In this application the transistor has been used experimentally as an amplifier for a band of frequencies extending from 5 to 15 mc and as a crystal oscillator at 10 mc. A second experimental application has been in a pulse-code transmission system in which (See Figure 1) the transistor has been used to produce and amplify pulses only one-twentieth of a microsecond long and to manipulate these pulses at a rate as high as ten million pulses per second. Peak pulse power output as high as 500 milliwatts has been obtained without excessive heating of the transistor. A third telephone application has been in an experimental very high frequency radio receiver. in this case, small size and low power drain are of great importance.
The results described in this article have been obtained with experimental models of junction tetrode transistors made in the laboratory. The Transistor Development Department, however, has made great progress in learning to control the processes involved in making these transistors. Production by Western Electric Company is scheduled to begin this year, with initial production directed toward meeting the government needs.
.
R. L. WALLACE, Jr., received his B.A. and M.A. degrees in Physics at the University of Texas. From 1937 until 1941 he was a graduate student and instructor at Harvard University. During the war years, he was engaged in communications research for the O.S.R.D. at Harvard University. He joined the Technical Staff of Bell Telephone Laboratories 1946 and has since been concerned with problems in sound spectrograph studies, magnetic recordings and more recently with the circuit applications of transistors, particularly in their broadband applications. Mr. Wallace is a member of the Acoustical Society of America, The Institute of Radio Engineers, Phi Beta Kappa, and Sigma Xi
|
| The author (right) observing bonding operation. E. Dickten is clamping base wires in micromanipulator.
Fig. 2 -Evolution of junction tetrode by process of narrowing center section, making p-layer thinner, and attaching extra base lead.
Fig. 3 --Method of finding correct position for bonding gold wire to very thin p-layer.
Fig. 4 -- H. E. Bridgers (left) and E. D. Kolb operating crystal-growing machine used in making germanium crystals for tetrode transistor research.
Fig. 1 - R. L. Carbrey adjusting transformer in junction tetrode pulse generator. The oscilloscope at the right displays 0.050 microsecond pulses.
Fig. 5 -The experimental automatic bonding machine:L. G. Schimpf (left) and E. Dickten are loading machine with a "header," in which is mounted the junction bar.
|
