Objective...........................................................................................................................1 Class A Amplifier .............................................................................................................1 Transformer-Coupled Class A Amplifier..........................................................................4 Transformer-Coupled 'Push-Pull' Class A Amplifier........................................................5 Transformer-Coupled 'Push-Pull' Class B Amplifier........................................................6 Transformerless 'Push-Pull' Class B Amplifier.................................................................7 Quasi-Complementary Amplifier......................................................................................9 EXPERIMENTS ..............................................................................................................10 Class B DC Voltage Transfer Characteristic.........................................................10 Crossover Distortion.............................................................................................10 Class AB Operation..............................................................................................10 Feedback Correction.............................................................................................11 Overload (Short-Circuit) Protection......................................................................11 Objective To review the evolution of amplifier power (output) stages and to examine experimentally selected power amplifier configurations. Parts List BJT opamp resistors potentiometer diodes 2 of 2N3904, 3 of 2N3906 741 330 Ω, 4.7 kΩ 10 kΩ 2 of 1N4004 Class A Amplifier A 'Class A' (linear) amplifier is by definition one which is capable of amplifying bipolar input signals, i.e., amplifying whatever the polarity of the change from the quiescent operating point. For example for a sinusoidal signal input variation around the DC bias point both half-cycles would be amplified essentially in the same manner. Class A power amplifiers involve basically the same kind of analysis considered in connection with incremental parameter analysis, except that large-signal operation is involved. For incremental-signal amplifiers detailed power calculations are not particularly important. Indeed smallsignal amplifiers typically are very inefficient, but the power level involved is generally small and the efficiency is traded-off against other considerations. Part of the reason for this small-signal power inefficiency is the need to locate the quiescent point so as to support amplification of bipolar signals. Because of this there is continual collector power dissipation associated with the quiescent point current and voltage even in the absence of a signal input; signal power typically is small compared to this dissipation. However concern about dissipation and consequently efficiency becomes important when the amplified signal amplitude is large, and this is the context for the following analysis. While details do vary depending on a specific circuit context basic considerations can be explored using the CE amplifier configuration drawn in the figure to the right. Note for later reference that the amplifier load Power Amplifiers V2 15-1 Copyright 1996 © M H Miller resistor RL is distinct from the collector resistor RC. These resistors have different functions, a matter discussed below. Certain simplifications are commonly made in studying the general behavior of this circuit to suppress secondary considerations. For example the BJT characteristics are assumed to be adequately represented by a 'flag' model for which the collector current is ß ( assumed constant) times the base current. This neglects the small but nonzero saturation voltage of the transistor, with the effect of generally overstating efficiency somewhat. However this omission does not affect the basic principles involved. Similarly the nonlinearity in amplification at very low and very high currents is ignored, as is the Early effect, i.e., the transistor characteristics are assumed to be parallel lines equally spaced for equal base current increments, with zero reverse bias leakage current. These assumptions simplify the analysis considerably, making it tractable. Important relationships are retained in the model, and the omissions can be compensated in a trial design by a modest adjustment in the nominal design target values. Ultimately of course a computer computation can incorporate a more detailed nonlinear device model in a numerical analysis. The IC-VCE plane for the CE amplifier collector characteristics is shown in the figure to the left; the transistor collector characteristics themselves are not actually drawn but rather should be inferred. Because the AC circuit is different from the DC circuit there are two load lines to consider. The DC load line establishes the quiescent point, and is given by VCC - VCE = IC RC Incidentally note that emitter stabilization is not used. This type of stabilization ordinarily would be avoided because for large transistor currents it would contribute noticeably to the standby power dissipation. (For modest power levels ( ≈ 1 watt or less) emitter stabilization might be considered. In such a case the analysis following is applicable on noting that the emitter and collector currents will be very nearly equal (ß » 1) and considering the emitter and collector resistances to be essentially in series.) The DC load line and the Q point are shown in the figure; this is same load line analysis as discussed and illustrated for the discussion of incremental parameter analysis elsewhere. The AC load line, by contrast, describes the relationship between changes in collector current and changes in collector voltage. Whereas the DC load line has a slope inversely proportional to RC the AC load line slope is inversely proportional to RL||RC. Since considerations of efficiency, more precisely maximum current transfer into the load, typically suggest making RL < RC there can be a significant difference in slope between AC and DC load lines. The AC load line is indicated in the figure also; it is defined by its slope and the Q point (through which it passes). Note that the intercepts of the respective load lines with the abscissa are labeled VCC and V* CC . For maximum symmetrical output voltage and current swing (ignoring saturation and low current constraints as described before) design the Q point to lie at the 'middle' of the AC load line, i.e., Power Amplifiers V2 15-2 Copyright 1996 © M H Miller The geometry of the shaded triangle reveals that Hence choose to locate the Q point as desired. Note that, as is to be expected The DC power provided by the supply is given by the product of the average current (here the Q point current because of the symmetrical variation assumed about the centrally placed Q point) and the supply voltage: Note that The average sinusoidal power in RL (not RC) is The instantaneous collector power dissipation is ICVCE = IC[ (RC||RL)( ICQ-IC)+VCEQ]; the term in the square brackets is obtaioned by substituting for VCE. Note that this power always is positive, and that the expression is zero for IC = 0 and also for IC = ICQ+VCEQ/RC||RL. It follows that the dissipation is a maximum between these intercepts. That maximum occurs at IC#=#ICQ and is ICQVCEQ. This is to be expected, since the fixed average supply power is dissipated in the resistors and in the transistor collector. Hence increasing the dissipation in the load necessarily reduces the collector dissipation, and so the maximum value of the latter occurs in the absence of an AC signal. The transistor operates cooler with a signal present than without! A useful 'figure of merit' for the amplifier is meaning among other things that the transistor selected for use in a Class A amplifier must be capable of dissipating more than twice the power that can be transferred to the load. Another more familiar figure of merit is the conversion efficiency: This has a (broad) maximum at RC = √2RL; the maximum is 8.6%. Power Amplifiers V2 15-3 Copyright 1996 © M H Miller The Class A amplifier is inherently an inefficient circuit, and its use is limited to relatively low power applications where efficiency and collector dissipation considerations are less dominant. The efficiency can be improved for a special circumstance which can be used in some low-power circumstances. This special case makes RL-> ∞, i.e., RC is used both to establish the DC bias point and the load. For this case the efficiency 'rises' to 25%. The increased efficiency here is primarily the result of not having to split the AC collector current between RC and RL, i.e., the total AC power is 'load' power. Transformer-Coupled Class A Amplifier In the preceding circuit the collector resistor is used to set the quiescent operating point so as to allow maximum symmetrical movement of the operating point. If the resistance is too large there is excessive dissipation by the quiescent point current; indeed the low efficiency of the preceding circuit is due primarily to just this dissipation. On the other hand if this resistance is too low the AC collector current division into the load resistor is degraded. This tradeoff is evident in the expression for the efficiency, which depends both on the ratio RC/RL and concurrently on the inverse ratio RL/RC. In other words it is desirable both that RL > RC and RC > RL. What would be useful is a 'resistor' which has a large AC resistance to improve the transfer of AC collector current into the load. and a small DC resistance to reduce losses in the process of establishing the Q point. The inductor is a circuit element with just this general property if we read 'reactance' for 'resistance'. Ideally the DC resistance is zero, and the AC reactance can be large (depending on the signal frequency and the inductance). The circuit to the left, below illustrates the modification. A further benefit, the ability to vary the slope of the load line for a fixed load resistor, accrues by using a transformer rather than simply an inductor. This more common configuration is drawn to the right. Power Amplifiers V2 15-4 Copyright 1996 © M H Miller Thus, for example, the DC relationship is VCE =VCEQ=VCC, i.e., the collector DC voltage is fixed; there is zero DC voltage drop across the (assumed lossless) transformer primary. On the other hand the transformer couples an effective resistance of n2RL into the primary; where n is the turns ratio of the transformer. Hence the AC load line, describing the relationship between changes in collector current and changes in collector voltage, has a slope equal to -1/n2RL. Since the AC load line passes through the Q point (zero change point) the load line is located as shown in the figure. The equation for the AC load line is For a symmetrical swing (ignoring saturation and low-current nonlinearities) set the Q point at the center of the load line: Ordinarily load resistance would be determined by the nature of the application and the transformer turns ratio specified to reflect that resistance into the transformer primary to provide the desired load line slope. The DC (supply) power is and the AC load power is Note that the collector dissipation capability required still is twice the AC power. If transformer losses are ignored the efficiency is 50%. The increase in efficiency is the result of eliminating the dissipation which would be produced by the quiescent current flowing through a collector resistor. Unfortunately another problem arises which limits the applicability of class A transformer coupling. The DC Q-point current flowing through the transformer primary contributes to the magnetization of the transformer core. For a power amplifier, that is an amplifier requiring a high DC current, the transformer core tends to saturate and degrade the coupling between the primary and secondary. Because of this the Class A transformer-coupled circuit is generally applicable to relatively low-power amplifies. Transformer-Coupled 'Push-Pull' Class A Amplifier The class A transformer coupled amplifier is limited indirectly in power output capability because of the quiescent collector current; this current flows through the transformer primary winding and adds a DC magnetic bias to the transformer core magnetic flux. This reduces the core flux change permitted for an AC signal current in order to avoid magnetic saturation of the transformer core and the consequent nonlinearity of a reduction in core permeability. The 'push-pull' circuit configuration to the right relieves some of this problem by an application of symmetry. The circuit consists of two distinct Class A amplifiers sharing a common load resistance. Although the output transformer used ordinarily has a center-tapped primary nevertheless it operates as two separate primary windings in series, coupled magnetically (same magnetic core) to a common secondary. The input transformer also commonly is center-tapped, but it also operates as two transformers with a common primary winding, coupled magnetically to separate secondaries. Because of the geometry of the input the two AC signals coupled to the transistor bases are of equal amplitude but are 180° out of phase with each other. The collector currents also are of equal amplitude and 180° out of phase (in the normal range of transistor operation). Hence there is no AC collector current flowing through the connection to Power Amplifiers V2 15-5 Copyright 1996 © M H Miller the emitter. emitter. Insofar as the AC collector current is concerned it flows from one collector to the other! What is particularly significant about this circuit is that unlike the AC currents the DC collector currents flow from the respective collectors to the center-tap, and then to the emitters. The magnetic effects of the DC collector currents tend to cancel each other since these currents magnetize the transformer core oppositely, but the AC currents couple in phase to the common secondary load. The design of this circuit is similar to two singletransistor Class A amplifier designs, except that the resistance seen in each collector circuit is RL/2n2. The reason for this is that both AC currents contribute to the magnetic flux linkage and so to the secondary current but only half the primary turns are used by each transistor. ( As indicated before the collector transformer operates as two transformers each with an n/2 turns ratio and with secondaries connected in parallel.) Since the total secondary power is supplied by two transistors rather than just one as before each one need dissipate (ideally) only half as much as in a single transistor circuit. The overall efficiency however remains at 50%. Transformer-Coupled 'Push-Pull' Class B Amplifier Push-pull Class A transformer-coupled amplifier efficiency could be improved considerably if the quiescent collector dissipation ICQVCEQ could somehow be eliminated. This dissipation arises largely because of the biasing needed to enable symmetrical collector current swings. Clearly however making VCEQ or ICQ zero does not allow bipolar operation. However it is possible to operate with ICQ = 0 and emulate bipolar operation if two transistors are used, each amplifying only one polarity of a symmetrical signal. A circuit to realize this is illustrated to the right; it is similar to the circuit of a Class A push-pull circuit except that there is no emitter junction DC bias. Each transistor therefore is biased at cutoff, i.e., with ICQ=0, and consequently with no DC collector dissipation. The AC base signals are 180° out of phase as before so that each transistor is turned on alternately, for alternate half-cycles of the input signals. Although the two collector primary windings both couple to the common secondary they do so on alternate half-cycles. In this circuit the output effectively uses two transformers operating disjointedly, each with an n/2 turns ratio. The effective turns ratio then is n/2, so that the load resistance reflected into each collector is n2RL/4. Ignoring saturation and low current effects the peak AC voltage is VCC, and the peak current then is VCC/(n2RL/4). The average current supplied by each transistor is Power Amplifiers V2 15-6 Copyright 1996 © M H Miller and the combined (total) AC power is The collector dissipation (each device) is Differentiate the expression in the brackets to determine that the dissipation is a maximum when Note particularly the ratio of the peak collector dissipation to the average power output; the transistors used for a given power output need have an order of magnitude less dissipation capability than before. Note also the improved efficiency. There is however one inherent flaw in the basic Class B operation, and that involves the biasing of the transistors at cutoff. Since the transistors do not turn on until the emitter junction bias is above the 'knee' of the diode characteristic there is a 'dead zone' around zero crossings of the input signal which is not amplified. This 'cross-over' distortion can be significant in some applications and arrangements must be made to mitigate it. One method of doing this simply is not to use strict Class B biasing, but rather to bias the transistors so that the quiescent point involves a low rather than zero DC current. This reintroduces some DC collector dissipation, which is the price for ameliorating the 'cross-over' distortion. Operation in this modified fashion usually is referred to as Class AB operation. Transformerless 'Push-Pull' Class B Amplifier Transformers generally are both bulky and heavy, characteristics preferably avoided in electronic equipment in general and for small/portable equipment in particular. Methods to combine the contributions of the transistors in the output stage without using a transformer find particular application in many integrated circuit devices where use of a transformer simply is not feasible. For example the simplified 'complementary symmetry' Class B output amplifier illustrated to the right uses NPN and PNP transistors in such a way that each conducts on alternate half cycles of the input signal. Both transistors operate as emitter followers, with RL as the shared emitter resistor. During Power Amplifiers V2 15-7 Copyright 1996 © M H Miller the positive half cycle of the input signal the NPN transistor conducts, providing a current source to drive the load. For the negative half cycle it is the PNP transistor that conducts, serving as a current sink to pull current from the load. There is however crossover distortion in the configuration as shown, corresponding to the 'dead zone' around the zero crossing of Vin during which the transistors do not conduct. A modified circuit (left) adds biasing to provide Class AB operation. The use of biasing resistors generally is not desirable in integrated circuit fabrication because they occupy relatively large portions of the chip area, and a modified biasing method of the sort illustrated to the right often is used. The diodes provide, in a manner of speaking, a substitute for batteries to bias the transistor emitter junctions. The quiescent state of the amplifier is set by adjusting Vin (DC level) so that there is no voltage drop across RL. Assuming matched diodes and emitter junctions for simplicity the quiescent current of the transistors then is the same as the diode current I (neglecting the relatively small base currents). Suppose Vin then is increased from its quiescent value. This raises the voltage at the NPN base, and since V (see figure) is substantially constant the base voltage of the PNP transistor also is increased. The effect is to cause the NPN device to carry a larger current, and the PNP device to carry a smaller current; the difference current flows through RL. Note that the increase of NPN base current to support the larger current comes from the current source I; the diode current is reduced accordingly. This means that I must be at least large enough to provide the base current needed for the largest NPN emitter current to be drawn. Conversely, decreasing Vin increases the PNP current while decreasing the NPN current, with the difference current flowing out of RL. Power Amplifiers V2 15-8 Copyright 1996 © M H Miller The circuit to the left provides additional flexibility (at a cost) in design by replacing the diodes with a transistor and resistive voltage divider. The transistor emitter junction provides a reference voltage of one diode drop across R2, with the voltage divider action (neglecting the base current) providing a proportional voltage across R1. Thus the voltage V can be adjusted by the choice of R1/R2 resistor ratio. Quasi-Complementary Amplifier The Class B/AB complementary push-pull power amplifier configurations assumed implicitly that the characteristics of the NPN and PNP devices are matched so that, for example, both half cycles of a sinusoidal input signal were amplified substantially the same. High power matched discrete complementary devices are available commercially. However for integrated circuits, where both PNP and NPN devices are fabricated in the same substrate, NPN devices generally have notably superior characteristics. One (of several) alternative configurations used for higher powers is the 'quasicomplementary' output stage illustrated to the right. The PNP transistor does not provide a sink for load current directly as before but rather provides the base drive for an NPN device, thus providing a composite PNP-NPN device with the input characteristics of a PNP transistor and the amplification capabilities of an NPN device. Power Amplifiers V2 15-9 Copyright 1996 © M H Miller EXPERIMENTS Class B DC Voltage Transfer Characteristic The circuit drawn to the right is used to illustrate various aspects of a discrete device complementary 'pushpull' power amplifier. A 'current source' (about 2 ma for the transistors specified) for biasing is provided by the PNP current mirror. The signal voltage source is simulated by the (741 OPAMP ) voltage follower; for this experiment a DC input voltage is provided by the potentiometer arrangement. Measure the DC voltage transfer characteristic (#voltage across the 330 Ω output resistor vs the voltage at point A). Make sure your data shows the 'dead zone' around the zero crossover adequately. Compare the measured data to theoretical expectations; use a computer generated graph as the reference on which to plot experimental data. Note: The transfer characteristic of the output stage will flatten as the output transistors saturate; this would be at (approx.) ±10.5 volts respectively (why?). Actually the 741 OPAMP output stage has a similar output configuration and probably will saturate at a somewhat lower voltage. Crossover Distortion Replace the DC input voltage by an AC sinusoidal signal (zero DC offset) from the function generator at a nominal frequency of 1 kHz. Starting with a small signal amplitude (< 0.7 volts peak) observe the output waveform as the input amplitude is increased. Include with your report a computer-generated plot of the output vs time for a sinusoidal input with amplitude 2 volts and compare computed results with the corresponding experimental observations to this plot. Input a small-amplitude AC signal , and vary the signal DC offset voltage; observe the effect on the output waveform. In your report explain as well as describe your observations. Class AB Operation Modify the previous circuit by replacing the base-tobase short-circuit connection (points A'-A in the circuit diagram) with the diode crossover circuit illustrated to the right. Measure the DC voltage transfer characteristic again, with special attention to the range of input voltage corresponding to the dead zone in the previous circuit. Similarly observe, describe, and comment on the AC output waveform observed as the input amplitude is varied. Power Amplifiers V2 15-10 Copyright 1996 © M H Miller Feedback Correction Modify the circuit as indicated by the heavy lines in the circuit diagram to the right. Replace the diodes by a short-circuit to restore the propensity for crossover distortion, and remove the voltage follower connection from the OPAMP output and connect it to the 330 Ω as shown. The high gain of the OPAMP tends to force a very small voltage difference across the OPAMP input terminals, concomitantly forcing the output waveform to track the input wave form. The OPAMP output varies so as to correct (in large measure) for the crossover distortion. Observe the output AC waveform with the original and the modified voltage follower connection for a sequence of signals of increasing amplitude; describe and compare these observations in your report. Overload (Short-Circuit) Protection If mischance causes the output to be short-circuited a large current demand may be placed on the output stage; for a power amplifier this can be destructive. Hence means often are added to prevent a current overload. In general there are two mechanisms needed for a protection mechanism; a means for detecting the event guarded against, and a means to prevent damage from the event. Short-circuit protection for positive signals (only) may be added to the circuit as illustrated to the right. Detection of excessive current is accomplished by the small resistors R added in the emitter paths; when the voltage drop across these resistors is sufficiently large (approximately 0.5 volt) an overload transistor (only one for NPN current is shown) is turned on. The protection then is implemented by bypassing the NPN base current to the output; the protection transistor essentially acts as a current sink to drain off base current.. The effect is equivalent to a rapid reduction in the value of ß, so limiting the short-circuit current to a designed safe value. Select a value for R to limit the short-circuit current to 15 ma; add your design to the circuit. Check your design carefully before testing it; the test very likely may be unforgiving of a design error. Note also that the PNP output transistor is NOT protected, so it is necessary to be careful not to apply an input voltage which turns on the PNP output stage. That is why the circuit diagram shows the negative end of the input potentiometer moved to ground rather than left at the negative rail. Be sure to make this change. Power Amplifiers V2 15-11 Copyright 1996 © M H Miller