Analog Circuits - Understanding & Designing / Home
- Part of ELEKTA -

15th Chapter : Transistor Amplifier

A small single battery cell AF amplifier able to drive a low impedance 8W load speaker will be presented. Although 3V or 6V batteries are available the 1.5V operation has some advantages: The efficiency is higher for low and medium sound levels and for the same battery size a 1.5V cell has a much larger capacity which results in longer battery life. Of cource you can not get much power from a 1.5V supply. If the bias output voltage is ½V_CC=0,75V, you may get theoretically a maximum swing of ±0,75V. But due to saturation voltages of the transistors and battery voltage drop only about V_P=±0,45V are realistic which is equivalent to 15mW. This sounds not so much but is fully sufficient for normal room sound pressure level if the efficiency of the loadspeaker is not too low. Because not only the output voltage but also the intermediate transistor voltages are limited by the supply voltage for optimum voltage efficiency the output transistors should work in common emitter configuration. This is quite different to AF power amps for higher supply voltages where in most cases the common collector configuration is prefered.

The maximum output current is I_P=V_P/R_L=62,5mA. If the transistor beta is B=150 then the maximum base current is about 400µA. This is quite much for an amplifier input so a preceding stage is very useful. IC designers normally would even build a 3 or 4 stage amplifier (and e.g. a 20 transistor bias circuit), but our PA should be as simple as possible. For maximum power gain in a direct coupled (best for low component count) amplifier the common emitter configuration is best suited, and a CB or CC stage is much worse. At low voltages a CC also has the disadvantage that the output voltage smaller than the input voltage by V_BE. If you want to drive a CE stage from a CC stage you need an input voltage of 2·V_BE»1,4V, which is nearly impossible form a single cell supply. Two CE stages are much easier to connect, but this makes feedback more difficult because the total amplifier will be noninverting. A solution will be a differential pair which has nearly the same characteristics like a single CE stage but features an inverting and a noninerting input. A further advantage is that the inherent temparature compensation makes the biasing more stable. All this ideas will result in this 1st schematic :

Figure 1 : Principal AF PA configuration

You will see the two stages and the feedback path formed by a resistive divider, but some things need to be changed. Very critical is the base drive of the output transistors. The upper transistor need a base voltage of V_CC-V_BE and the lower one needs V_BE. Both conditions can only be valid at a very small supply voltage range. Even small changes in V_CC will change the output stage bias current quite dramatically. At high bias currents the battery will be discharged very fast and at low bias levels the amplifier will have large cross-over distortions. A much better circuit is show in figure 2.

Figure 2 : The complete audio PA circuit

Because the maximum output stage base current is about 500µA the differential pair is biased at 1.5mA. To improve the coupling and biasing one part of the differential pair is doubled. Also the load resistor is replaced by a current mirror with pnp transistors. For biasing the output stage no additional circuitry is needed because 1..3mA bias current is pretty good and some changes can be tolerated. If you like you might introduce an adjustment point by connecting a 50W poti between the emitters of the current mirror output. The over-all gain is set to V_CL=5 via R3 and R4. Without feedback you will get the open-loop gain which is much larger :

V_OL » S_Diff·RL·ß » I_CDiff/U_T·R_L·ß » 25

R_in » 2ß/S_CDiff » 2U_T/I_CDiff·ß » 10kW

With feedback the input resistance is about 50kW, so R2 dominates the total input impedance.

What Simulation Not Says

Oscillations are often a problem designing amplifiers. In this design there are really critical because there are only two stages. The Miller capacitances gives some margin. But also important is decoupling V_CC via a 4,7µF capacitor. Because a real loadspeaker behaves not like a 8W resistances a series RC combination (22W+10nF) should be connected to the amplifier output.

Figure 3 : Photo of the first prototype

Some hints on realization

What Transistors are best?

The FET should have a low I_DSS and V_TO (BF256A or BF244A/BF245A). The bipolar transistors should have a high beta (e.g. >250), so the BC549C (npn) and BC559C (pnp) are a good choice. The output transistors also need a low saturation voltage, so the BC327-40/BC337-40 are better. For SMT designs BC849C/859C and BC808-40/818-40 are well suited.

What changes are needed for 3V?

The most important change is to set the output bias voltage to 1,5V for large output swing. This is e.g. possible with a resistor connected from the inverting input to ground.

How to get a better sound?

The optimization of the loadspeaker and its enviroment is more important than any PA changes. The easiest way to improve the amplifier is to increase the bias currents and to replace R6 by a current source. For the actual circuit THD is about 3%.

What are the applications?

Maybe a AM radio or an measurement amplifier to “hear” into other audio circuits. You can also use the amp as a pre-amplifier or with other transistors as an broadband RF amplifer(e.g. with BF224/324). Of course there are also apllications possible in which no good sound quality is needed but high efficiency, like a sirene.

Link : Basic Analog Course

To EE Practice

Partly published in ELEKTOR Vol. 9/2000.

If you need more information, please send me an email!

To Top - To Homepage - To EE Software

Copyright © 2000 Stephan Weber. All rights reserved.
Stand: May 24, 2000.