Minimising audio distortion will be discussed in this section. The main culprit seems to be the non linearity of the base- emitter junction. The techniques introduced will be predistortion, symmetry, feedforward and feedback.
While placing a resistor in series with the base emitter junction appears to swamp the non- linearity of the latter, this will also reduce the gain, or make the collector load uncomfortably high- thus this idea alone is not enough to fight audio distortion.
Once the designer is out of methods for reducing distortion in a single stage, the only resort is to use the distortion in one part of an amplifier in order to cancel the distortion in another; feedback involves subtracting (part of) the actual output from the desired output- a correction signal is thus created which can be added to the input signal in order to reduce distortion. The importance of feedback cannot be stressed too much- it will get its own section sometime.
Feedforward is similar to feedback; the error signal is mixed at the output of the amplifier, though, not the input. This second output amplifier is awkward to design, however, unless a bridge configuration is used: The low impedance speakers need to be driven by a very low impedance output stage.
Predistortion involves canceling the distortion in a stage by placing a stage featuring a complementary characteristic in front of it (or after.) A good example is the well known two- stage common emitter amplifier (figure below): In the first stage, the gain is slightly increased when the input voltage becomes more positive (because the emitter resistance is reduced.) The same is true of the second stage, considered individually, but the overall distortion is reduced: The output stage is fed in antiphase from the collector of the first transistor.
The cascode configuration is another widely known example: It is well established that the common base circuit is highly linear, because the input impedance is reduced in sympathy with increased gain- but the input impedance is poor. A preceding common collector stage is often used to overcome that. An increasing current in the input stage will only decrease the current in the output transistor, and vice versa, thereby minimising the aggregate distortion, again.
The distortion problem is nowhere more acute than at the output of a power amplifier: Several volts are needed across the low impedance loudspeaker. If predistortion cancels the overall distortion in stages in series, symmetry does the same for stages in parallel. Harmonics which are an even multiple of the fundamental frequency, and thus more insulting to the ear , are cancelled in npn and pnp devices driven in parallel.
If only 'n'- type devices are available, as in thermionic valves, it is still possible to take advantage of symmetry: The two output valves are driven in antiphase, feeding two halves of a transformer primary winding whose center tap is grounded. The speaker is driven by the secondary, which provides a current stepup and isolation from the high voltage needed for the valves.
Present Random Quote:
|
When all your opponents are gone, what use will you be? |