The observant reader will have noticed that in the last schematic that the plate resistor's value is much lower than in the original circuit. One problem with the stock Transcendental OTL is the ridiculously low plate current for the 12AX7 based input stage: 0.4 mA. Why was this trickle current chosen? The answer is that the designer hoped to develop all gain he could out of this stage. Higher current would have meant a smaller valued plate resistor and thus less gain. Why was the extra gain desired? Probably to lower some of the distortion and to extend the smaller bandwidth that resulted from using so little current. Huh? Much like a dog chasing his tale, we create distortion in the attempt to lower distortion. Starved of current, a triode produces more distortion, as it must operate in the bottom, gooey region of its plate curves. Additionally, the wimpy current is insufficient to charge the following stage's input capacitance (the Miller effect capacitance) quickly enough to ensure a wide bandwidth. Our solid-state brothers figured this out long ago. If an amplifier cannot charge its internal capacitances quickly enough, the feedback is going to send the amplifier into a slew-limiting distortion, e.g. SID.
   What's the solution? Replacing the 12AX7 and 12AU7 with different tubes and increasing all the idle currents are a good start. Tube myopia is a debilitating disease that inflects more audio designers than can be imagined. (I remember John Atwood and myself promoting the 5687, 5965, 6072, and 7062 back in the 80s and meeting blank stares. I remember being told, "Don't you understand, only the 12AU7 and 12AU7 can be used in audio equipment; the tubes which you like best are meant for industrial uses and cannot be used in audio equipment." Today, I hear, "Only the 300B or 2A3 can be used in a single-ended amplifier.")
   A better choice might be 6N1Ps or 7062s throughout. These triodes could easily handle sufficient current to make the amplifier sing. But I am also troubled by the first stage's power supply filtering: just a large valued resistor and a large valued (22µF) capacitor.

     Improving this situation can be easily accomplished. A shunt regulator can be made out of the input tube's unused triode. Or a constant current draw stage can be made by adding a cathode follower to the input tube's output. For as long as input stage and the cathode follower shared the same valued load resistor, their combined current draw will always equals a constant, as the cathode follower's current conduction is in anti-phase to the grounded-cathode amplifier.

< PREVIOUS

NEXT >