For example, the popular MJ15003 NPN power transistor lists a range of 25 to 150 for its h_FE. Thus, making a useable transistor-based buffer with just three transistors is difficult, given the actual potentially low h_FE of most transistors; with less than three transistors, it is near impossible.
      All of which leads to reader Tony's first circuit: a global feedback free, completely single-ended, Class-A, three-stage buffer. This circuit's goal is to maintain DC coupling from input to output, without adding a DC offset. Will it work? Maybe.

            Reader Tony's 3-stage, single-ended, Class-A buffer

      Hand selecting high-h_FE transistors might give an adequate current gain, but a safer bet would be to add a fourth transistor. An added transistor would not only increase the final current gain, but would also eliminate the need for the first stage's diode. Each transistor has about a .7 volt drop from base to emitter. Thus, cascading PNP to NPN to PNP to NPN transistors should keep the output's DC offset inline with the input. Now, adding more stages to an amplifier is usually fraught with danger, as each stage introduces some phase lag, which in an amplifier that uses a global feedback loop potentially spells oscillation. In an amplifier that forgoes global feedback in favor of local feedback, the phase shift is of no real concern, but at the cost of dropping frequency response at the extreme high frequencies and higher output impedance.   
      So are we done with improving this amplifier? No, not at all. We have one large problem to face: the limited voltage swing due to the resistor loading of the three-emitter followers in series. These resistors limit the maximum voltage swing of following stage, as each proceeding stage requires an increasing input current, which must diminish with the collapsing voltage across the resistor.

       One solution is to decrease the resistors' value, so as to allow for the increased current demand of the following stage. This would certainly help. However, it would also greatly increase the required current gain for the amplifier as whole and it would also greatly increase the dissipation of the transistors. Another solution would be to use a separate, higher voltage power supply for the input and driver stages. This would allow lower idle currents to be used and would increase the linearity of the emitter followers. However, it would also greatly increase the voltage breakdown requirements for the transistors.
     The last solution is to actively load the emitter followers with constant current sources. This would allow a full voltage swing into the following transistor's base and it would also allow the retention of the simple bipolar power supply. (The full output swing of this Class-A amplifier is set by the output stage's idle current, as the constant current source can only deliver its set current into the loudspeaker, which in this example is 3 A, which in turn equal 24 volts and 36 RMS watts into 8 ohms. Thus the full +/- 30 volts of the power supply will not be fully utilized. Thus, the voltage lost across the constant current sources is not a concern.)

< PREVIOUS

NEXT >