Bipolar Junction Transistor Amplifiers:IC Power Output Stages

IC Power Output Stages

If an IC power amplifier is to occupy a relatively small volume, the output power will be limited. This is due to the limitation imposed on the thermal conductivity of a small-area power device. A discrete transistor mounted on a large heat sink will exhibit a much higher thermal conductivity than that of the smaller IC chip. This limits the power that can be dissipated by the output device or devices of the IC as the junction temperature will be higher with the lower thermal conductivity of the IC chip. Typically, this limitation leads to an IC power output stage [3] that is implemented in a high-efficiency configu- ration. Generally, the output stage for power outputs in the range of a few watts will use a class-B configuration, while those with tens or hundreds of watts will use a class-D configuration.

One of the most significant limitations on dissipation is the junction temperature. As the temperature rises, several potentially dangerous effects may occur. First, the solder or alloys used in the transistor can

be softened or even melted. Second, the impurity profiles in the doped regions can be affected if elevated temperatures exist for long periods of time. A third result of higher temperatures is the increase in collector leakage current. In power transistors, this current doubles for an incremental increase in temperature of 7–10°C. The leakage current, Ico, at temperature T2 can be related to Ico at temperature T1 by

where TK can range from 7 to 10°C. For a temperature increase of 100°C, the minimum factor of increase in Ico is 1024 (2100/10), whereas the maximum factor is ~20,000(2100/7). This marked increase in Ico can lead to increased power dissipation which leads to an increased temperature followed by a further increase in Ico. In some cases, this feedback effect is large enough to cause thermal runaway and destroy the transistor. This effect is minimized by placing a resistance in the emitter to decrease the forward bias on the base–emitter junction as current increases.

In other situations, the leakage current can approach the value of the quiescent collector current. Since leakage current is not controlled by the applied input signal, its effects can severely limit the amplifying properties of the stage. For these reasons, the manufacturer places a limit on the maximum allowable junction temperature of the device.

The maximum junction temperature is therefore an important quantity that limits the power a transistor can deliver. The junction temperature will be determined by the power being dissipated by the transistor, the thermal conductivity of the transistor case, and the heat sink that is being used. The collector junction is the point at which most power is dissipated; hence, it is this junction that concerns us here.

Basically, manufacturers specify the allowable dissipation of a transistor in two ways. One way is to specify the maximum junction temperature along with the thermal resistance between the collector junction and the exterior of the case. Although this method is very straightforward, it incorrectly implies that the allowable power increases indefinitely as the transistor is cooled to lower temperatures. Actually, there is a maximum limit on the allowable dissipation of the transistor which is reflected by the second method of specification. This method shows a plot of allowable power dissipation versus the temperature of the mounting base. Quite often this plot shows power dissipation versus ambient temperature, where an infinite heat sink is used. However, if the transistor could be mounted on an infinite heat sink, the ambient temperature would equal the mounting base temperature; thus both plots convey the same information. This second method indicates the maximum allowable power dissipation, in addition to the maximum junction temperature.

The maximum limit on power dissipation ensures that chip temperature differentials do not become excessive as excess power is dissipated in collector regions. It also minimizes the possibility of excessive collector currents in typical applications.

Thermal Resistance

The thermal resistance of a material is defined as the ratio of the temperature difference across the material to the power flow through the material. This assumes that the temperature gradient is linear throughout. The symbol q is used for thermal resistance, and P Power flowing through conductor The diagram of Figure 23.26 represents the thermal circuit of an IC chip, including the output transistor, surrounded by free air. Here, qJM is the thermal resistance from collector to mounting base and qA the thermal resistance of that portion of air in contact with the mounting base; TA the temperature of air far away from the transistor; TM the mounting base temperature; and TJ the collector junction temper- ature. The power P will be determined by the electrical circuit that includes the output transistor; P, in turn, will determine the temperatures TJ and TM. The temperature TJ can be written as TJ = TA + P(qJM + qA) (23.55) This equation shows that as more power is dissipated by the output transistor, TJ must rise.

For high-power IC chips mounted in a TO-3 package, qA is usually many times greater than qJM. If qJM were 1°C/W, then qA might be 5–20°C/W. Of course, qA depends on the area of the IC package.

The diagram of Figure 23.27 represents the thermal circuit of the IC chip mounted on a heat sink.

The power that can be dissipated without exceeding the maximum junction temperature is found from solving the preceding equation to result in

The values of qHS presented between the chip case and free air might range from 0.4°C/W for air-cooled systems to 2°C/W for flat vertical-finned aluminum heat sinks to 8°C/W for cylindrical heat sinks that slide over the chip package. For each thermal circuit, the amount of allowable power dissipation is fixed.

Circuit or Conversion Efficiency

The efficiency of a power output stage is a measure of its effectiveness in converting DC power into AC output power. It is defined as

A useful relationship between the allowable dissipation of the chip and the maximum output power can be found by assuming that the power delivered from the DC source is dissipated by the output transistor and the load. This assumption would be very inaccurate for a class-A, resistive load stage, in which the resistor dissipates significant DC power. For many class-B or class-D stages, the assumption is not unreasonable. In equation form, this assumption is expressed by PS = Pout + PT (23.58)

where PT is the actual dissipation of the output transistor or transistors.

In terms of the circuit efficiency, the transistor dissipation can be written as

The effect of circuit efficiency on output power can be demonstrated by assuming that the maximum allowable output stage dissipation is 5 W. If the circuit efficiency is 50% or h = 0.5, the maximum output power calculated from Eq. (23.60) is also 5 W. If the efficiency is increased to 78.5%, the maximum efficiency of an ideal class-B stage, the maximum output power is found to be 18.26 W. Increasing the circuit efficiency to 98% a figure approached in a near-ideal class-D stage, the maximum output power becomes 245 W. In this case, an output stage that can dissipate 5 W delivers 245 W to the load. Although these calculations are based on some idealizations, it clearly shows the importance of using a circuit configuration that leads to a high value of efficiency. This explains the popularity of the class-B stage and the class-D stage in IC design as opposed to the class-A stage.

Class-B Output Stage

A class-B stage that can be integrated is shown in Figure 23.28. The current source I1 consists of an IC current-source stage to provide a small bias current through diodes D1 and D2. A small quiescent collector current I2 is necessary to reduce crossover distortion at the output. As vin swings positive the input to T2 goes negative, turning T2 on while shutting T1 completely off. A negative-going waveform at the amplifier input drives the bases of T1 and T2 positive to shut T2 off and pass the signal to the output through T1. The amplifier can be made short-circuit proof by limiting the output current that can flow if the output terminal is accidentally shorted to one of the supplies. Figure 23.29 indicates the additional

circuitry required for this purpose. The emitter–base junctions of transistors T4 and T5 are driven by the voltage drops across the resistances RE1 and RE2. Under normal operating conditions, these voltages are too small to turn T4 and T5 on; thus, circuit operation is unaffected. If the output is short-circuited to the negative supply voltage, serious damage to device T1 could result if transistor T4 were not present. A large voltage would appear across T1 and the base–emitter junction would be forward-biased resulting in a high value of emitter current. The excessive power dissipation could destroy T1 if the output were to be shorted to the negative supply or even to ground. When this occurs for the circuit of Figure 23.29, T4 becomes sufficiently forward-biased to divert the base-current drive from T1. The maximum current that can flow in the output circuit is then limited to VBE4/RE1. Typical maximum currents for the short- circuit case range from 10 to 50 mA for modern IC amplifiers.

The VBE multiplier circuit often replaces the two diodes in Figure 23.28 and Figure 23.29 to get better cancellation of the crossover voltage of the output devices. This circuit appears in Figure 23.30. If negligible base current flows in T3, the voltage across R1 and R2 is

The voltage VCE3 is used to eliminate crossover distortion and can be adjusted by the ratio of the two resistors. Whereas the absolute values of individual resistors cannot be accurately determined in standard IC fabrication processes, the ratio of two resistors can be determined to the required accuracy.

There are several other output configurations based on the complementary emitter follower. A popular one is the Darlington output stage modified for IC amplifiers as shown in Figure 23.31. The current gain of this stage is very high, approximately equal to b2. If high-gain pnp devices are available, a Darlington pair similar to the upper npn pair can replace devices T2, T4, and T5. For the Darlington pairs, a larger difference of input bias voltage must be provided to the inputs of the respective pairs due to the larger voltage drop between each input and output which is now 2VBE(on) instead of just VBE(on). The VBE multiplier circuit can be designed to generate this increased bias voltage.

Class-D Output Stages

IC class-D amplifiers using PWM have been reported with efficiencies of 90% or more at 10-W output and a frequency response from 20 Hz to 20 kHz. Many IC chips are available that drive power BJTs or MOSFETs, delivering 30–50 W to a load. Larger discrete circuits report audio amplifiers based on the class-D stage that deliver 600 W per stereo channel [9]. This type of amplifier is often used in low-end car radios.

PWM is used to reduce the power dissipated by the transistor while delivering a large power to the load. The varying load signal is applied by means of output devices that switch between on and off states. The resulting load voltage has a rectangular waveform that contains an average or DC value dependent on the duty cycle. In addition, the load voltage would also contain several AC components, however, these unwanted components can be filtered before reaching the load.

Any periodic waveform can be represented by a Fourier series consisting of a DC component (if present), a fundamental frequency component, and higher harmonics of the fundamental frequency. A rectangular wave switching between +V and -V that remains positive for t+ seconds is said to have a duty cycle of

The average value varies directly with d. The Fourier coefficients of the AC components also vary as d is changed and new frequencies may be introduced, but in general, these components are of little interest to us as they can be easily eliminated. If, for example, the repetition frequency is 200 kHz, all AC components of the waveform will be greater than this value and far out of the audio range. Now if we vary the duty cycle sinusoidally at some low frequency, the average value will also vary sinusoidally. Mathematically, we can express this by saying that if d = 0.5 + ksin wt, then the average value is

The waveform with variable duty cycle can be filtered by a low-pass filter to eliminate all frequencies above w. The result is a low-frequency output sinusoid with an amplitude that varies in the same manner as the duty cycle. A block diagram of PWM amplification is shown in Figure 23.33.

The voltage control circuit must have a 50% duty cycle when the input signal is at 0 V. As vin becomes nonzero, the duty cycle varies proportionally. The high-power switching stage amplifies this rectangular wave and applies the output to a low-pass circuit that allows only the changes in average value to pass to the load. The output signal is then proportional to the input signal, but can be at a much higher power level than the input signal. The power output stage may dissipate only 5–10% of the power delivered to the load.

The major advantage of PWM is that the output transistors need to operate in only two states to produce the rectangular waveform: either fully on or fully off. In saturation, we know that the very small voltage drop across a transistor leads to very low-power dissipation. A very small dissipation is also present when the transistor is cut off. If switching times were negligible, no device power loss would occur during the transition between states. Actually, there is a finite switching time and this leads to an increased total dissipation of the output stages. Still, the efficiency figures for the class-D amplifier are very high, as reported earlier. This leads to higher possible power outputs and smaller chip areas for integrated PWM amplifiers. The stages can be direct-coupled to the load, which eliminates the necessity of capacitors. Nonlinear distortion can be less than that of class-B stages, and matching of transistors is unnecessary.

In contrast, the disadvantages of this amplifier ultimately dictate the limits of usefulness of the PWM scheme. The upper frequency response is limited to a small fraction of the switching frequency. The operating frequency of power transistors generally decreases with higher power ratings. It follows

that the upper corner frequency of the amplifier may be lower for higher power transistors. Further- more, a low-pass filter may be required to eliminate the unwanted frequency components of the waveform. The generation of radio frequencies or electromagnetic interference by the switching circuits can also present problems in certain applications.

In addition to compound emitter followers, the power output stages can be designed in several arrangements. Figure 23.34 shows two possible configurations. The diodes appearing across the output transistors are present to protect the transistors against inductive voltage surges. If the filter is inductive, the current reversals that occur over short switching times generate very large voltage spikes, unless the protective diodes are used.

00000000In the push–pull circuit, the low-power, pulse-width modulated input turns T1 and T3 on when the signal is at its maximum value. Transistors T2 and T4 are off at this time, and current is forced through the load. When the signal switches to the minimum value, T1 and T3 go off, while T2 and T4 turn on to pull current through the load.

Figure 23.34(b) shows a bridge circuit that can drive a floating load with a single power supply. When the input signal reaches its maximum value, T1 and T3 are on while T2 and T4 are held off. The input signal is inverted and applied to the bases of T5 and T6. This inverted signal is at its minimum value during the time when the normal input is maximum; thus, T5 and T7 will be off while T6 and T8 are on. Current will leave the collector of T3, flow through the load, and enter the collector of T8. When the input assumes the most negative value, T1, T3, T6 and T8 turn off while T2, T4, T5, and T7 turn on. Current now leaves the collector of T7, flows through the load, and enters the collector of T4. During this period, the load current flows in the opposite direction to that flowing when the input is maximum. The load current then reverses each time the input signal makes a transition.

In some applications, such as motor control or high-output audio systems, the load serves as a filter of high frequencies since these particular loads cannot respond to the switching frequencies.