With 42 vdc across the heater string, it means each tube on average is operating with 10.5 vdc -- which surely does make the tubes slow to warm. The benefit however is that operating the tubes at less than rated heater voltage improves the signal to noise ratio of the tube with virtually no loss of gain. It also means that at 10.5 volts, the typical 12AX7 will draw 135 ma, rather than its rated 150 ma at 12.6 volts.
But there is also that 560 Ω resistor. At 42 volts across it, the resistor is drawing 75 mA. There's another 4 mA flowing through the bias divider resistors. Adding all of this up, it means that there is a total of 214 mA flowing though all four output tubes, or on average, 53.5 mA per tube. Allowing 3.5 mA to account for screen grid current, it means that the plate current in each tube is 50 mA. With a schematic indicated plate-to-cathode voltage of 393 volts, this places the plate dissipation for each tube at 19.65 watts, or 103.4% of rating. The tubes are getting cooked -- and particularly so if the unit is operating from today's typical AC line voltages, where the individual tube dissipation will be even higher.
Fisher was really threading a needle with this design, but with tubes still being plentiful at the time, yet the end of the tube age in sight, it was the decision made.
If you up the voltage across the heater string even more then to improve the warmup time or operating characteristics, it simply cooks the output tubes even more. At 10.5 volts a piece for the 12AX7s however, weak tubes that would otherwise still perform well at rated heater voltage, may have problems as Larry noted. Typically, about 11.0-11.5 volts for 12AX7s produces a very good compromise between warmup time, good performance with a wide range of tubes, and low noise.
Now making the bias circuit adjustable as was done in the X-100B will certainly address the making-the-bias-adjustable part of the issue. But the tubes still cook.
I addressed the ultimate answer to all of this in an old thread of mine here:
http://www.audiokarma.org/forums/index.php?threads/kx-100-fishers-intro-stratakit.673654/
While the modifications address all of the issues, and adds some significant muscle along the way, it also requires a significant amount of work. Short of that, you can adjust the compromises made to potentially help the issue.
The 560Ω resistor is the real issue here -- but its required because the output stages increase their current draw rather quickly with increasing power output. The resistance of the12AX7 heaters on the other hand goes up in value as the current flow through them increase -- working backwards against the needs of the output tubes which are trying to draw more current to produce the increased power. So, the resistor works to keep the voltage across the tube heaters somewhat constant in spite of increasing current draw -- so that the output tubes can then draw the current they need. But the current flow of the required resistor, and that of the heaters, makes the quiescent current through the output tubes too great. The resistor can be adjusted somewhat, but not too much or the voltage across the 12AX7 heaters will soar with increasing power output. If normal use will only require no more than 5 watts or so per channel, then you might try this:
Increase the 560Ω resistor to 1.5K at 10W, and then bias the output tubes to 44 vdc. This will reduce the output tube dissipation to about 80% of rating, and slightly up the heater voltage to each 12AX7. It will do so however at the expense of maximum power output, since the heater voltage to the 12AX7s now has poorer regulation with the increased fixed resistance value.
As I said, the real answer is to execute the modifications of the thread above. However, increasing the fixed resistance and making the bias adjustable may be a workable compromise as well -- as long as you understand the consequences of the trade offs being made.
I hope this helps!
Dave
