Download this article in .PDF format
This file type includes high resolution graphics and schematics when applicapable.

To understand the characteristics of power amplifier transistors, and how parameters affect each other in combination with circuit trade-offs, the circuit designer should have a basic concept of device construction. The semiconductor die, resistor stabilization when applicable, and the packaging of the device are all very important at microwave frequencies. It is important to qualitatively relate device construction to parameters and circuit performance. Thus, one can intelligently evaluate different devices and the differences between devices of the same transistor family.

The currently most popular type of transistor construction will be described—planar epitaxial diffused junction. This type is used by all semiconductor manufacturers at present in the processing of high-frequency, power-amplifier transistors. The geometries may vary, but the concepts described here are qualitatively independent of the geometries employed. Parameters are affected by geometry, but the relationships are similar for all types.

Basic transistor construction

Fig. 2. Planar transistor construction. The basic elements are indicated.

The NPN planar transistor is constructed as shown in Fig. 2. Fig. 3 is a typical top view showing interdigitated (comb structure) connections for the emitter and base areas; which is one type of connection pattern.

Fig. 3. Top view of typical planar transistor showing interdigitated comb-like structure. This is one popular connection pattern.

Transistors for higher frequencies and higher power outputs in a given physical chip area must have increased active-to-physical-area ratio. To accomplish this, a finer geometrical structure for the emitter is necessary to increase the emitter-base periphery for a given physical area. This requires smaller emitters spaced closer together. This, in turn, requires tighter mask tolerances, creating more yield problems and requiring more careful processing; thus, a more expensive device is created. Emitter geometry definition of one micron of less has been achieved in small-signal, low-power devices (one micron = 10-6 meters or 3.95 x 10-5 inches). This mask-tolerance requirement, however, cannot yet be attained in very high power, large-area transistors. The present state-of-the-art dictates emitter geometries of 3 to 5 microns in width or site side for reasonable yields.

The active area concept

The base drive must approach the emitter-base junction from the side as visualized from the basic construction model. This base current has to go through the region under the emitter. The narrower the base width, the higher the lateral sheet resistance of the structure will be, or the higher the effective base resistance is and the more voltage drop there will be for a given base drive. Thus, the emitter-base voltage even at the edge of the emitter-base junction will not be as high as the emitter-base voltage applied across the external terminals of the device. Further, under the emitter, away from the base-contact area, less emitter-base voltage is available and the current turn-on is less as well. This is the “current pinch-off” effect.

Current pinch-off is a function of dc beta (hFE) since beta is a function of the base width. As the transistor is driven harder, the pinch-off effect becomes worse and the active area of the device (the emitter-area-carrying current) increases less rapidly than at lower current levels. What the active area looks like is roughly show in Fig. 4. Basically, this is a three-dimensional series resistance with a shunt capacitance. As frequency increases, this built-in low-pass filter allows less base-emitter junction drive. The active area decreases as the frequency increases. This, to the circuit designer, means that the transistor is shrinking in useful size or area. Unfortunately, the impedance levels are not changing as rapidly and the output capacitance is only slightly affected.

Fig. 4. Impedance representation in transistor model. The effect is a low-pass filter.

Fig. 5. Transistor current distribution for a simple transistor model.

How the current distribution looks at one frequency for a simple transistor model, is shown in Fig. 5. From this sketch, it can be seen how a higher-frequency transistor (a transistor with a higher active-to-physical area ratio) would require a finer geometry. By example, Fig. 6a shows a coarse geometry compared to Fig. 6b. For the same drive at the same frequency, the current distributions would be as shown. For a given physical area, there is much more active area in the transistor of Fig. 6b because of the finer geometry. This makes the transistor more useful at this frequency. The greater active area gives more power gain for reasons that will be discussed. The transistor of Fig. 6b, however, has problems involving the safe-operating area and uniformity of operation, over the area for which something must be done to give it equivalent safe-operating performance. This also will be discussed.

Fig. 6. How finer geometry increases active area. At a is a relatively course geometry. For a given physical area, b has a more active area because of its finer geometry.