The preceding chapters discussed the basic operation, circuitry, applications and limitations of the transistor. This chapter contains important miscellaneous considerations, including transistor operation at high frequencies, i-f and r-f amplifiers, limiters, mixers, handling techniques, hybrid parameters, and printed circuits.
The Transistor at High Frequencies
Transit Time, Dispersion Effect.
In the earlier chapters it was noted that the low-frequency, small-signal parameters change as the operating frequency is increased appreciably above the audio range. Figure 7-1 illustrates the low-frequency equivalent circuit of the transistor including the collector junction capacitance C. At higher frequencies this equivalent circuit must be modified to include the effects of the current carriers' transit time on the transistor parameters. The transit time of the carriers (holes or electrons) is one of the major factors limiting the high frequency response of the transistor.
Fig. 7-1. Low-frequency equivalent circuit
of the transistor (including collectar junction capacitance).
Fig. 7-2. Transistor high-frequency
Fig. 7-3. Transistor high-frequency equivalent circuit.
Fig. 7-4. Typical transistor i-f amplifier.
The movement of holes or electrons from the emitter through the base layer to the collector requires a short but finite time. In the transistor, as noted earlier, the electron does not have a clear and unimpeded path from emitter to collector. As a result, the transit time is not the same for all electrons injected into the emitter at any one instant. The effect of an identical transit time for all electrons would be a simple delay in the output compared to the input signal. Because the injected carriers do not all take the same path through the transistor body, those produced by a finite signal pulse at the emitter do not all arrive at the collector at the same time. The resulting difference is very small and is of no consequence in the audio frequency range. At the higher frequencies, however, this difference becomes a measurable part of the operating cycle, and causes a smearing or partial cancellation between the carriers. Figure 7-2 illustrates the dispersion effect in a transistor at high frequencies. Notice that, in addition to the increased period, the signal has also suffered a reduction in amplitude (the time delay results in a phase shift) . The decrease in the output signal means a decrease in the current gain . The degradation in frequency response becomes steadily worse as the operating frequency is increased, until eventually there is no relationship between the input and output waveforms (and no gain).
Another factor that limits the high frequency response of the transistor is the capacitive reactance of the emitter input circuit, which behaves as if re is shunted by a capacitor. This reactive parameter is reduced if the source impedance is made as low as possible. Since rb is also effectively in series with the source, a good high frequency transistor must have a low base resistance. If the source impedance and base resistance are low, the upper frequency response limit is determined primarily by the collector junction capacitance and the variation in the current gain.
Alpha (a) Current Frequency. In view of these limitations, the basic circuit illustrated in Fig. 7-1 is not a useful approximation of transistor performance at high frequencies. To modify this circuit for accurate representation of high frequency equivalence requires that all of the internal parameters be specified in a complex form (magnitude and phase angle) as functions of the frequency. In most cases, however, it is sufficiently accurate to modify Fig. 7-1 to include only the variation of a with frequency, since few design problems justify the details required for exact equivalence. The variation in current gain can be satisfactorily approximated by the relationship:
where a is the current gain of the operating frequency fc ai is the low frequency current gain; and fe is the frequency at which the current gain is 0.707 of its low frequency value (3 db down).
As a numerical example of the above, compute the current gain tor a junction transistor having a low frequency current gain of α1 = 0.95, an a cut-off frequency of fc = 10 mc, and an operating frequency of 7.5 mc. Then
Including only the junction capacitance and variation in a in the low frequency circuit makes all the computed values far from exact. In addition to the capacitive reactance of the emitter, there is also considerable variation with frequency in the collector resistance and collector junction capacitance. The collector resistance rc decreases rapidly for a ratio of greater than 0.15, falling to about 10% of its low frequency value at and then remains at that value. The collector junction capacitance also decreases as the operating frequency increases above an greater than 0.15, but does not decrease as rapidly as rc. In a typical characteristic, Cc drops to approximately 75% of its low frequency value at and to about 50% at after which the curve levels out. Due to the coupling between the input and output circuits, the input impedance contains a reactive component beyond the emitter shunt capacitance. At the a cut-off frequency fc, the reactive component is approximately equal to the resistive input component. This causes the input impedance to be inductive for the grounded base connection, and capacitive for the grounded emitter connection (due to phase reversal).
High Frequency Equivalent Circuit.
Because of these factors, representation of the transistor high-frequency operation by any linear four-terminal equivalent network is at best a rough approximation over any substantial frequency range. This is especially true if the circuit is to be reasonably representative of the physics of the transistor, and if the number of Circuit parameters are to be kept within reasonable limits. One form of equivalent circuit, suggested by Dr. W. F. Chow of the General Electric Company, has worked out well. This involves the insertion of a low pass R-C filter network in the low frequency circuit, derived for an equivalent current generator in the collector arm (αie) . The modified equivalent circuit illustrated in Fig. 7-3 takes into account the variations of rc and Cc with frequency. This circuit provides a fair representation of transistor performance through the range below the a cut-off frequency. If the operating frequency is greater than fc, the low pass filter must be replaced with an R-C transmission line.
Frequency Comparison of Point-Contact and Junction Transistors.
At this point, a brief explanation of why the point-contact transistor is capable of a higher operating frequency than the junction type is in order. The high frequency effects on the equivalent circuit parameters are essentially the same for both types. Actually, the major difference is in the mechanics of conduction.
Point-contact transit time is determined primarily by the field set up by the collector current. In equation form, where S is the point spacing in centimeters, µ is the hole mobility in ρ is the germanium resistivity in ohm-cm, and I is the collector current in amperes. Typical transistive values are ρ= 12 ohm-cm, and Ic = ∂ ma, for which T = 1,570 µµsecs. Ignoring all other factors, this limits the upper frequency response of the point-contact transistor to about 600 megacycles.
In the junction transistor, movement of the current carriers is primarily by diffusion, and is not appreciably affected by the electrode potential fields. In equation form ime through the base layer, W is thickness of the base layer in centimeters, and D is the diffusion constant in cm2/sec. Typical values for a P-N-P transistor are W = 2 × 10-3 cm and D = 38 cm2/sec (for an N-P-N type, D is about 69 cm2/sec) , for which T = 0.121 µsecs. Ignoring all factors but the diffusion time, the upper frequency for this typical P-N-P type is approximately 8 mc, and for the N-P-N type about 16 mc.
High Frequency Circuits
In general, the upper frequency limit of the junction transistor is considerably lower than the limits of the point-contact type. On the other hand, the junction type has a lower noise factor, and better stability in some applications. These factors frequently make it advantageous to use the junction transistor in some high frequency applications even if an additional stage or two may be required.
Figure 7-4 illustrates one stable form of i-f amplifier stage using a WE 1752 N-P-N transistor. The operating frequency is 455 kc, and the gain is 18 db.
Due to the natural regenerative feedback path through the collector junction capacitance and the base resistance, and the close coupling between the input and output circuits, the circuit, when connected in tandem, is likely to oscillate unless the stage is carefully tuned. The alignment procedure is easiest if the last stage is tuned first. For an input resistance Rg = 500 ohms, the output resistance ro averages 12,500 ohms, and Cc is about 15 µµf.
The cascading of transistor i-f is more complicated than that of vacuum tubes. The main contributing factors are the effect of the output load on the input impedance, and the effect of the generator impedance on the output impedance. These factors show up largely in the design of interstage coupling networks.
Fig. 7-5. Transistor i-f coupling networks.
I-F Coupling Circuits.
For interstage coupling, an i-f transistor amplifier may use a series resonant circuit such as that illustrated in Fig. 7-5 (A) . The main requirement for this type of coupling is that the short-circuit current gain is greater than unity. Thus, the series connection in the case of the junction type may only be used in the grounded emitter connection.
Parallel-tuned resonant-coupling circuits are applicable in i-f strips, particularly when junction transistors are used. If point-contact transistors are used, special care is required to avoid oscillation due to the inherent instability of these types when short circuited. Several types of parallel-tuned coupling circuits may be employed. Figure 7-5 (B) illustrates one such possible circuit with the input of the coupled stage directly connected into the resonant circuit of the first stage. This direct coupling can also be used if the inductor and capacitor are interchanged. Figure 7-5 (C) illustrates another coupling arrangement with the input of the second stage connected to the junction of the two capacitors in the resonant output tank of the first stage. In this case, the capacitors can be used for matching the impedances between the stages. This coupling arrangement can be made inductive by reversing the reactive elements and connecting the input of the second stage into the tank inductance. This arrangement requires that a capacitor be inserted in the input lead of the second stage. This capacitor blocks the d-c bias and also helps to avoid the excessive loading of the tank due to the input circuit of the second stage. An alternate method is to couple the second stage to the tank inductively. If the inductive coupling is also tuned in the second stage, the circuit becomes the double-tuned coupling network illustrated in Fig. 7-5 (D) . The center tap in the inductance of the secondary circuit provides for the proper impedance match between the stages.
The close coupling between the input and the output circuits of the transistor causes the resonant frequency of the coupling circuit to be particularly sensitive to variations in the input and output impedances. In general, the load impedance has a greater effect on the input impedance than the generator impedance has on the output impedance. For this reason, the best procedure to follow in aligning an i-f strip is to start with the last stage and work toward the first.
To avoid the critical tuning problem, the stage may be neutralized. This allows each resonant coupling circuit to be independently adjusted without introducing any detuning on or by the remaining stages. One form of neutralization is illustrated in Fig. 7-6 (A) . For reasons of clarity, only the 4-p circuit is shown. Neutralization is accomplished by balancing the resistor RB and the equivalent impedance Zc of Rc and C against the transistor base resistor rb and the equivalent impedance of the collector arm zc composed of rc and Cc. The balancing conditions are more clearly illustrated in the equivalent circuit of the neutralized circuit, as shown in Fig. 7-6 (B) . The circuit is drawn in the form of a
conventional bridge. The bridge is balanced when . Under this condition there is no interaction between the input and output circuits. Then, when the stage is neutralized, the output impedance is independent of Rg and the input impedance is independent of RL.
Fig. 7-6. (A) Neutralized amplifier.
Fig. 7-6. (B) Equivalent circuit of neutralized i-f
Fig. 7-7. Typical transistor r-f amplifier.
Fig. 7-8. Junction transistor mixer circuit.
In practical circuits the neutralization network design can be simplified by omitting the capacitor C if a point-contact transistor is used, or by eliminating Rc if a junction transistor is used. This changes the balance equations to for the point-contact types, and for the junction types. These simplifications are possible at the intermediate frequencies because feedback is governed primarily by rc in the point-contact transistor, and by Cc in the junction transistor. The network components are not very critical. Values within a 5% tolerance range are generally satisfactory.
Notice that the lower output terminal is connected to ground through RB. This makes it important for the value of RB to be small in order to avoid introducing too much noise through RB into the output circuit. For satisfactory operation, the value of RB should not be larger than the base resistance. This fixes the value of C in the range of Cc and Rc in the range of rc The loss in gain due to the neutralizing network will be less than 10% of the total gain in a properly designed circuit.
Transistor r-f amplifier circuits, like their counterpart vacuum-tube circuit types, are most often used for improving the gain, over-all signal-to-noise ratio, or selectivity characteristic of a multistage circuit. Figure 7-7 illustrates a typical transistor r-f amplifier circuit. The design is basically the same as that of an i-f amplifier. The chief problem is the selection of a transistor having a sufficiently high a cutoff. The power gain of a r-f amplifier is inversely proportional to the frequency. In a typical case a transistor having a maximum gain of 40 db at 10 mc will have a maximum gain of 20 db at 40 mc. There are two critical parameters, the emitter bias and the base resistance. The base resistance is determined by the physical construction of the transistor and, therefore, low base-resistance transistors, designed specifcally for high frequency applications, should be used. The importance of emitter bias was considered in the analysis of oscillator circuits. The bias should be selected to be far enough away from the unstable region of the characteristics to avoid oscillation, and yet not so far away that the gain is very low. Special care must be taken to avoid introducing stray capacitance into the emitter input circuit. These reactances tend to lower the input emitter impedance, and thereby decrease circuit stability.
Limiter circuits can be designed using transistors and germanium diodes. These circuits operate much like vacuum-tube limiters. In the grounded base connection, the input circuit acts like a diode when the emitter electrode is biased slightly in the forward direction. When the value of the input signal exceeds that of the emitter bias, the signal is rectified by the diode action of the input circuit. The resulting self-bias tends to keep the maximum emitter current constant. Since the collector current is proportional to the emitter current, the output signal is maintained at a constant level over a large range of input signal values. The input rectification action is considerably improved when the circuit is shunted by a junction diode. The diode performs two important jobs. It clips large positive input pulses, and prevents the coupling capacitor from charging on extraneous noise pulses. For optimum operation, the output resistance is matched to the load, and the generator impedance is kept as low as possible.
The operation of the transistor in mixer circuits depends upon the rectification and non-linearity of the emitter circuit when it is biased slightly in the forward direction. Figure 7-8 illustrates one basic arrangement of a transistor mixer circuit employing a junction transistor. This circuit takes advantage of the relatively high gain of the grounded emitter connection by injecting the BFO signal into the common emitter lead. The junction transistor works well in mixer stages despite its relatively low α cutoff. This is possible because only the i-f frequency must lie within the useful frequency range of the transistor. The point-contact type also works satisfactorily in transistor mixer circuit. Its utility is limited to some degree by its relatively high noise figure and low gain.
Fig. 7-9. Transistor power supply.
As the number of applications for transistors increase, many new power supply systems will be required to fit in efficiently with the particular design. The power requirements of an individual circuit is very small, so small in fact, that quite often the life expectancy of the bias battery is the same as its normal shelf life. Nevertheless, in some applications it may be desirable to derive the power supply from an existing a-c source. Figure 7-9 illustrates an experimental power supply, fabricated for a particular application, where a bias of 30 volts and a drain of 10 ma were required. The circuit is a basic full-wave rectifier terminated in an R-C filter. With the values shown, the ripple is less than one percent.
Miscellaneous Transistor Characteristics and Handling Techniques
Transistor Life Expectancy.
One of the outstanding features of the transistor is its practically indefinite life expectancy. Long life was originally predicted on the basis of the transistor construction and its conduction mechanics, which indicate there is nothing to wear out. Although the transistor is still very young, enough experimental data is now available to back the initial long life predictions.
The usual transistor failure occurs gradually over a long period of time and after thousands of hours of operation. The performance degradation generally shows up as an increasing saturation current. (The effect of increasing saturation current was covered in the transistor amplifier chapter.) While the various self and fixed biasing methods may be used to minimize the effects of increasing Ico the system's efficiency and gain suffer. In an amplifier circuit, this factor decreases the available volume. Gradually, as the limit of the automatic biasing arrangement is reached, there is also a noticeable increase in the distortion content.
Another variation in the transistor performance characteristics is a gradually decreasing output resistance. In systems designed for an image impedance match (Rg = ri and RL = ro) , this change introduces a mismatch loss. In the usual amplifier design, however, the output resistance is in the order of 20 to 50 times the load resistance. The decrease in ro, therefore, is less serious than the increase in Ico. The best single maintenance check is a measurement of the current gain.
Sudden failures of transistors are not common in normal operation, although open emitter and collector junctions were not too rare in the early transistors. These defects were attributed to faulty assembly during manufacture. Present manufacturing and quality control techniques have practically eliminated open junction defects. Transistor shorts are more common since they are usually caused by overloading. When the transistor power rating is exceeded, the junction temperature rises quickly. The increased heating effect encourages diffusion of collector region impurities into the base layer, which, in time, will destroy the junction. In brief then, open circuited transistors generally result from poor production; short-circuited junctions generally mean improper circuit design.
Insofar as ruggedness is concerned, the superiority of the junction transistor compared to the point-contact type can be anticipated from a comparison of the basic construction details (Chapter 2) . The emitter and collector electrodes of the point-contact type depend on a force contact with the germanium surface. These cat-whiskers, it will be remembered, are fastened to the main electrode conductors which are embedded in, and held by, the plastic stem. It is possible, then, to vary the contact pressure of the catwhiskers by a twisting force applied to the plastic stem. This distortion can be introduced by direct mechanical force, humidity or temperature variations.
Most of the present transistors are hermetically sealed. Sealing is important because of the ease with which an unprotected junction surface may be contaminated by water vapor. The contaminating effects are particularly noticeable so far as the value of the saturation current in an unsealed unit is concerned. In a typical case, the saturation current of a junction transistor will increase one hundred times its dry air value when the relative humidity is increased by 50%.
The transistor can withstand shock, vibration, and drop tests far beyond those of the vacuum tube. However, it is a good plan to treat the transistor with reasonable care to avoid unnecessary damage. The effect of distortion of the stem on electrode contact pressure was noted in earlier paragraphs. Any damage to the hermetic seal is, of course, serious. Transistor electrode leads are generally as flexible as those of regular carbon resistors. These leads should not be subjected to continual bending or flexing, or to pulls greater than a half-pound.
Generally, junction transistors (Raytheon types 720, 721, 722, Germanium Products Corporation types 2517, 2520, 2525, Western Electric 1752, etc.) have long pigtail leads. These types can be soldered directly into a circuit. However, due to the temperature sensitivity of the transistor, solder connections must be made quickly. It is always a good idea to heat sink all solder connections by clamping the lead with a pair of long nose pliers connected between the soldered point and the transistor housing. This provides a shunt path for a large part of the heat introduced at the solder joint. If it is at all possible, transistors with short leads should not be soldered directly into the circuit. Several types of sockets will accommodate these short lead types. For example, the Cinch type 8749, type 8672, and regular 5-pin subminature tube sockets will handle point-contact transistors similar to the Western Electric 1698, the General Electric G 11A, etc.
The physical location of the transistor is not critical with respect to its mounting position, and since the heat generated by an individual transistor is small, many may be packed together. However, since the transistor is sensitive to the ambient temperature, hot spot locations near tubes and power resistors should be avoided. In this regard, a word of precaution on collector dissipation ratings is in order. The maximum collector dissipation is specified at some definite temperature (usually 25°C) . This value must be derated if the ambient temperature is greater than the specified rating temperature. Usually this amounts to a 10% decrease in power dissipation for each 5°C increase in ambient temperature. As a numerical example, assume that the maximum allowable collector dissipation for a transistor rated at 250 mw at 25°C is required when the ambient temperature is 40°C. The operating temperature represents an increase of 40° — 25° = 15°C. The power handling capacity should be derated 10% for each 5°C increase or 15/5 x (10) = 30%. 30% of 250 mw is 75 mw. Thus the maximum collector dissipation at 40°C is 250 — 75 = 175 mw.
Whenever a transistor is operated near its maximum rating, it is always good insurance to tie it to a metal panel or chassis. This connection provides a large radiating surface which permits the collector dissipation to be maintained at higher levels. In typical cases, this procedure increases the transistor power dissipation rating from 20 to 50%.
In addition to its power handling limitations, the transistor is susceptible to damage by excessive values of current and voltage. It is particularly important to protect the transistor from those transient surges which may be caused by switching or sudden signal shifts. Transient effects are particularly predominant in oscillator, i-f, and high frequency amplifiers due to the storage capacity of the reactive components. Limiting devices are usually incorporated into the circuit. The series resistors in the emitter and collector arms of the base-controlled negative-resistance oscillator are typical examples. In more complicated circuits, transient limiting elements are usually selected on the basis of tests made on experimental breadboard models. If a scarce or expensive transistor is involved, the equivalent passive "T", made up of standard carbon resistors, can be substituted for this measurement. When connecting a transistor into a live circuit, the base lead must always be connected first. In disconnecting the transistor from a live circuit, the base lead must be removed last.
It is an easy matter to mistakenly reverse the polarities of bias supplies, particularly in complementary-symmetrical circuits. reverse polarities will not impair the transistor as long as the maximum ratings are not exceeded. It is always a good plan to check for proper capacitor polarity, since almost all of the circuits require polarized types.
Significance and Derivation.
The open-circuit parameters, r11, r12, r21, and r22 are used exclusively throughout this book primarily because they are the most familiar four-pole equivalents. Some engineers prefer the short-circuit conductance parameters g11,
g12, g21, and g22. The conductance parameters serve well for the junction transistor, but do not work out too well for the point-contact type, which inherently exhibit short circuit instability.
The disadvantages in both the r and g forms suggest a combination or hybrid type of representation which will be applicable to all transistor types without requiring elaborate measuring techniques. The so-called 'h' or hybrid parameters are becoming more and more popular. Since many of the manufacturer rating sheets now specify the h parameters, it is important to be able to convert the hybrid values into the more familiar r form for use in the design and performance equations. On a four-terminal basis, the hybrid parameters are equated as:
The basic circuits for measuring the h parameters are illustrated in Fig. 7-10, which define the values of the parameters in terms of the input and output currents and voltages as follows:
Fig. 7-10. Basic circuits for measuring four-terminal h parameters.
Notice that two of the measurements are made with the output short-circuited, and the remaining two are made with the input open-circuited. Furthermore, none of the parameters are exact equivalents, since r11 is a resistance (ohms) , h22 is a conductance (mhos) , h12 is a numeric (voltage ratio) , and h21 is also a numeric (current ratio) .
Resistance Parameters in Terms of Hybrid Parameters.
The relationship between the r and h values can be determined by straightforward substitution and the simultaneous solution of equations 7-1 and 7-2, as follows:
As a numerical example of these conversions, the manufacturer's rating sheet for the G.E. type 2N45 specifies the following typical values for the hybrid parameters:
Printed Circuit Techniques
One of the most promising features of the transistor is its ability to fit into the new prefabricated wiring techniques, by which the maze of hand-soldered wires normally associated with electronic equipment has been eliminated. Basically, a printed circuit starts with a metal foil bonded to one or both sides of an insulating plastic material. The metal foil may be copper (most popular) aluminum, silver, or brass. Most types of laminated plastics are suitable as the base insulator. The circuit is drawn on the foil clad laminate with an acid resistant ink. The complete assembly is then dipped into an etching solution which removes the metal not protected by ink. Holes are then drilled or punched into the assembly at appropriate points, and into these holes the various circuit components are inserted and soldered to the metal foil. If the circuit is at all complex, hand soldering is extremely tedious and difficult, and the dip soldering technique is used. In this method, components with preformed leads are inserted into the holes, either manually or by an automatic process. After fluxing, all the connections between the component leads and the circuit pattern are accomplished by a "one-shot" dip in a molten solder bath. Those portions of the circuit which must be left free of the solder are coated with a protective lacquer or masked before the solder bath. Dip soldering assures very reliable solder joints in one simple operation, and also permits a greater reduction in size by means of stacking techniques, which were previously limited by the space requirements for hand soldering operations.
Complex circuits are normally laid out on both sides of the laminated base. Connections crossovers may be made by several methods. The most common is by means of a tined eyelet. This is of particular importance in those cases where connection is made to a component which may be soldered and unsoldered several times during the life of the equipment. Repeated soldering at the foil will eventually cause it to lift from the plastic base.
Fig. 7-11. Experimental transistor i-f amplifier.
which may be soldered and unsoldered several times during the life of the equipment. Repeated soldering at the foil will eventually cause it to lift from the plastic base.
In spite of the small cross-sectional area of the foil conductors, the current carrying capacity of the printed circuit is good, due to the relatively large surface area and the heat conduction by the base material. A 1/32-inch copper foil conductor, for example, can safely handle about five amperes. Increased temperatures caused by current overloads causes the metallic conductor to buckle and separate from the base.
One of the major advantages of the printed circuit is its uniformity from unit to unit. For example, the distributed capacitance between foil conductors is in the same order of magnitude as that of a carefully hand wired assembly. In the prefabricated type, however, the value remains constant from unit to unit because they are all produced from the same master design.
Figure 7-11 illustrates the front and back of an experimental printed circuit type of transistor i-f amplifier. The component arrangement can be seen at the left of the illustration and the printed wiring can be seen at the right. Miniature components for use with transistors are shown in Fig. 7-12. The top row of the figure shows a miniaturized transformer and three resistors. The bottom row illustrates an inductor, a capacitor, two junction transistor sockets, and two point-contact transistor sockets.
The marriage of standard and miniaturized components with the basic printed circuit is, in essence, the "autosembly" technique devised by the Signal Corps Engineering Laboratories. This method is best suited to present production facilities, since it utilized components with
proven reliability. However, the recent progress in the development of printed components indicates that most of the applications of prefabricated circuits are still to come. Printed resistors having values of 10 ohms to 10 melts and which are sprayed onto an area of 1/16 of a square inch have been used successfully. Small inductance coils, having values up to 20 µh, can be etched into the printed circuit, and capacitors ranging from 10 µµf to .001 f can be incorporated in the printed circuit by etching opposite sides of foil-clad glass-cloth laminates.
Fig. 7-12. Miniature transistor components.
The transistor, because of its mechanical ruggedness and long life expectancy, is well adapted for direct assembly into printed circuit patterns. The minute heat generated by the transistor makes its future use in compact packaged equipment particularly promising. The prefabrication techniques will initially reduce the out-of-service time considerably, since complete circuits will be encapsulated in units no larger than present vacuum tubes. On the other hand, assembly repairs will require great skill and technical knowledge due to the complex arrangement of the miniaturized components.