Ferninando -- The biasing method used in the X-100-3/A produces a significant compromise against two opposing needs. It can be summarized as follows:
1. From a biasing only standpoint, each output tube in your amplifier should operate with a quiescent cathode current of 32 ma. This level of current flow will keep output stage distortion at bay, and based on schematic B+ information, will cause each tube to dissipate just shy of 11 watts at the plate.
2. However, the 12AX7 heater string needs at least 33 volts across it (11 volts per tube) to not only produce normal operation in the line and phono preamp stages, but also to let the unit warmup up in at least this millennium. The tubes will operate with as low as 10 volts heater, but this is based on an individual tube basis, with warmup time being quite significant. The problem is however, that when you connect a number of identical tubes in series, the heater characteristics of each tube are rarely identical -- and particularly so at a reduced operating voltage. And, the heater characteristics can vary from one tube manufacturer to another. As a result, at 30 string volts, one tube may in fact get 10 volts, but another 10.5, while another only 9.5. This will significantly compromise performance in the tube with the starved heater, and almost never let that tube fully warm up. By operating the string at at least 33 volts, there is enough current flow to ensure that ALL the tubes will operate properly even though the heaters are powered from a reduced voltage string arrangement. Fisher was able to push this concept to the edge, by using tubes of identical manufacture, and pushing a reasonable warmup time right to the edge of that definition. With the changing times as they were (coming into SS), Fisher likely considered that the tubes would outlast a reasonable period of use, and so any consideration for tube changes down the road that ended up with different manufacturer's tubes being used in the heater string received little concern.
3. As for R65, well it was an insurance policy. With no possible means of adjusting the bias, it helped to cover any sins of high heater current draw in V3, that would result in a low grid bias condition for the output stage, and the resulting high current draw/high dissipation that would produce. With the individual tube adjustment capability now being added, the need for R65 is moot.
You can see then that there is a gap between good output stage performance/small signal tube warmup time, and good output tube life. Frankly, this bias system worked better with the 7591 based amplifiers, because they naturally operate with a tad more quiescent current, which eliminates the overly long warmup time, without over dissipating the output tubes under quiescent conditions.
In your amplifier, there is a slight gap between what's good for the output stage tube life, versus what's good for output stage and small signal tube performance. The 33 volt cathode buss voltage I stipulated assures reasonable warmup time and good performance from the small signal tubes and the output stage, but causes each output tube to dissipate just under 12 watts at the plate. This is a dissipation level that the Russian tubes should be readily able to deal with, remembering that the more modern Design Maximum plate dissipation rating for the 6BQ5 family of tubes is actually 14.4 watts. I would suggest that you at least try this level of output stage quiescent current draw, and judge the overall performance it provides for yourself. You can always adjust down accordingly if you want.
In view of all the preceding, you might ask then what is the purpose of R66? All it would seem to do is to add even more current draw through an already nearly maxed out output stage, from a quiescent condition standpoint. Well, again, the heater characteristics of a vacuum tube are anything but linear with respect to voltage and current. As the amplifier is asked to produce more and more power output, current draw through the heater string bias system will cause the bias voltage to increase at a very nonlinear rate, which can endanger the heaters of the small signal tubes, and increase distortion significantly in the output stage. Therefore, to help keep a lid on this unacceptable behavior, R66 is used as a shunt to help contain the worst of the extreme voltage changes that can occur with such a bias system.
I address all of this even further in the thread I did on the Fisher X-101C. That amplifier uses the same bias system approach, although it does include a single bias adjustment pot to adjust the quiescent bias current of all four tubes at once. In any event, the original bias system was removed in favor of installing a separate DC heater supply, and EFB(tm) to bias the output stage. The changes that move made to the overall performance of that amplifier were quite significant.
I hope this helps!
Dave
