place to start would be to design a competent push-pull tube amplifier and then disable the drive signal to one leg, unless the signal is high enough of a value to imply that the other leg is nearing cutoff.
    You have in the Dynaco MK 3 a fairly good push-pull. Why not forgo the trendy aspect of the para-feed for the ease and simplicity of just converting the MK 3 into an SE / push-pull amplifier by retaining the basic layout and functioning of the stock amplifier? To see what I am suggesting, imagine if you were to remove one of the coupling capacitors in the amplifier.
    One tube would still see the signal and would amplify that signal. But the other tube would only see the negative bias voltage and not amplify any signal. Thus, we would have half an amplifier. The output transformer will not saturate, as the disconnected tube will still draw the same idle current as before, which will match the idle current of the other tube and prevent any net magnetizing of the output transformer's core. The amplifier would then work in SE Class A mode, as just one device handles all of the output waveform.
   Now if we replace the capacitor only during that part of the drive signal when the normally operated tube ceases to conduct, the amplifier becomes a push-pull amplifier. (One point worth noting is that the impedance presented to each tube is just 1/4 the impedance the same output transformer would present to two tubes working Class A push-pull. Why? One half of the primary winding is used to couple signal to the speaker at any time. So the winding ratio has now dropped by half and the winding ratio squared times the load impedance is the reflected impedance and  [1/2] = 1/4.)
   The amplifier redesigned should meet the requirements for SE/ push-pull operation. The first change to note is the move to the pentode configuration rather than the ultra-linear of the output stage tubes. This was done to keep the normally idle tube idle, as the signal from the ultra-linear tap would stimulate it in the same way as signal at grid 1 would. The second change to note is the insertion of a choke at the idle tube's cathode. To better understand what is going on here, remove the diodes, zener and rectifier. Now imagine what the cathode signal

The output stage of a Dynaco MK 3 converted into an SE / PUSH-PULL amplifier.   

would be in the presence of drive signal at the grid of this tube. The answer is that the cathode will track the input signal at the grid as the choke represents a very high impedance at audio frequencies, which allows the tube to function as a Cathode Follower. As it is primarily the grid-to-cathode voltage that determines the current draw from a pentode, the constant voltage between grid and cathode keeps the current flowing through the tube constant.
   Now let's put the diodes back into the circuit. What happens when the cathode tries to exceed the diode string's breaking voltage? Well, the answer is that the diodes aggressively begin to conduct and lock the cathode at their threshold voltage and the constant current source aspect of the tube ceases and it begins to amplify. (The rectifier is added to prevent the zener from being forward biased when the drive signal travels negative relative to the ground.)
   The trick is picking the right value for the zener. The truth is although I often use zeners in the circuits found in the Webzine, I would not use one in an actual circuit I would build for myself. I would use a transistor and a two resistor voltage divider to create a virtual zener. In fact, the top resistor could be replaced with a potentiometer and a resistor to make an easily adjustable zener!
   Ian, there will more to come next month, such as a cathode biased version using a current source instead of a choke.


<PREVIOUS   Copyright © 1999 GlassWare. All Rights Reserved