|
|
This decrease in current will result in both output voltages being forced more positive than they should be. These unwanted fluctuations in drive voltages create unwanted current fluctuations through the output devices that cancel in a Class-A amplifier but are released to the output in a Class-AB, B amplifier. The current source maintains a constant current draw and effectively eliminates this distortion mechanism.
(Class-A does more than just provide a more linear transfer curve, it allows a push-pull amplifier to actually cancel both common mode noises and distortions. Unfortunately, amplifier noise is only measured with a shorted input. Consequently, the Class-AB amplifier real noise contribution is not measured, but certainly heard.)
100% Potential Gain Realization
We have covered three of the common circuits used to equalize the drive voltages to the identical output devices in an OTL push-pull amplifier. These three circuits shared a low output impedance unity gain follower output stage. The next three circuits have output stage that provides gain, but not a low output impedances or a low distortion figure. Consequently, these circuit demand a feedback loop to reign in the output stage.
The circuit below uses the split-load phase splitter to drive the two output devices. The output is phase inverted. The gain is roughly equal to output device's Gm against load in a Class-B amplifier.
|
|
|
|
In a Class-A amplifier, the gain is roughly equal to twice the output device's Gm against load. The zener-capacitor combination strives to eliminate the follower like action the top output device's placement seems to imply and thus it works to increase the gain and output impedance of the device. Shorting the output, reveals that both output devices work evenly in drawing and sourcing current from the grounded output, as both devices must see an identical magnitude input signal. Forcing the output +1 volts up cannot force the bottom device to see the +1 volt pulse at its input, as there is no mechanism to relay the pulse. Nor can the top output device see the pulse, as the zener-capacitor combination pulls the plate resistor voltages up by the same +1 volts.
Feedback can be applied by adding two resistors to this circuit: one entering the grid and one spanning the grid to the output. With this feedback loop in place, a positive pulse applied to the output will be fed back to the grid, which will relay it in phase to the bottom output device and out of phase to the top output device. The bottom device will conduct more and the top device will conduct less, both of which will buck the pulse. However, this simple feedback based amplifier works best with high transconductance output devices. Tubes (and to a certain extant MOSFETs) require an additional gain stage. Adding a simple grounded cathode amplifier is the obvious choice when using MOSFETs. When using tubes for the output stage, the better choice is two triodes in cascade or two in a cascode configuration or single pentode based grounded cathode stage. In fact, the later Futterman amplifiers that used sweep tubes as the output tubes used this very topology. (He ran the pentode first stage in near current starvation mode to further increase its gain, which is a testament to just how little transconductance the output tubes held.) In his amplifiers the output signal was voltage shifted higher and the fed to both the top pentode's screen and the to the top of the split-load phase splitter's plate resistor.
Using this topology with transistors or MOSFETs in the output stage does not require so much gain; a 5687 would work just fine. The final open-loop gain is given by the gain of the input stage against the transconductance of the output devices against the load impedance. Once again, the Class-A amplifier realizes twice the gain and half the output impedance that the Class-B amplifier yields.
|
|
