|
Well, sort of, the problem with feedback lies in that it can do all of these very desirable things, but not all at once. Without a load the open-loop gain was 10, but with a 32 ohm load the gain may fall to 2. Now if all of the output signal is fed back into the amplifier, the output impedance will only be halved. But the distortion may even be greater than in the unloaded, open-loop state. Why? The output stage might have left Class A operation, all of the tubes might have left their linear region of use, and the power supply might have begun collapsing because of the increased current demand and its noise component greatly increased. In other words, the feedback could be used to lower the output noise and extend the bandwidth of the amplifier with no load, or it could be used to maintain a constant unity gain at the output in the presence of a load.
Realizing the limits of feedback to convert sow's ear unto silk, we must make the amplifier as close to silk as we can before we add the feedback. This makes noise reduction of prime importance to designing this amplifier. By using as many noise canceling techniques as we can, the feedback will have much less work to do. In order to balance the output stage, the top pentode must see no noise on its grid and the bottom pentode must see all of the negative power supply rail noise on its grid. Why the dissimilar demands? In fact, there are not any.
Think in terms relative to each pentode's cathode. The top tube's cathode connects directly with the output of the amplifier; therefore, its cathode should have zero noise on it as should its grid. The bottom tube's cathode connects to negative power supply rail; therefore, it will share all the noise on it and, thus, so should its grid. If both tubes have their grids tied to their cathodes and if this two tube series is placed across a bipolar power supply, the noise at the junction of the two tubes will null and equal no noise. (Imagine two rubber bands tied together and pulled equally in opposing directions.)
|
|
