gridcurrent said:
1k seems to be have been a popular value, in both expensive (like the Marantz 8b) and less fancy amplifiers, both pa and hi-fi.
"Popular value" is somewhat meaningless, because the value must be considered in regard with the rest of the circuit.
The grid stoppers are used to introduce a pole in the open-loop response, that is essential for stability of an amplifier submitted to global NFB.
The frequency of this pole is determined by the value of the grid stopper resistor and the reflected Miller capacitance of the output tubes.
For example, changing 7591's for 6V6's, all other things being equal, the pole would move about at least an octave because the g-a capacitance of the 7591 is about 3 times less than 6V6.
But many other parameters are significant, such as open-loop gain, transformer leakage inductance and stray capacitance, as well as the general layout.
It is extremely difficult to predict accurately the ideal value, because many factors are either unknown or not well documented.
yet McIntosh avoided stoppers, at least in output stages.
McIntoshes use a such different topology it is almost incongruous to try to compare. The output tubes are driven by cathode-followers (partly in order to limit the issues of grid current), but the output stage (including transformer) is under nested NFB. The output tubes operate with loads shared between cathode and plate, similar to a cathodyne PI. The overall gain of the output stage is unity (well, two if you count both outputs). Apparently, this (combined with the extraordinary attention to detail in the OT) made unnecessary stabilization components (although some amps had to have a 100pF added to the bottom half of the PI.
Note the smaller MC225 (with 7591's) had 470r grid stoppers. That would be consistent with the much lower Cga of the 7591's.
messing with the ubiquitious Dyna Stereo 70 when replacing the 1k grid stoppers with 10k, thought it sounded more "effortless", just saying.
One could venture a tentative explanation; maybe the amp was not that stable with the 1k's and the 10k's tamed the unstability... Interseting to note the SCA35 had no grid-stoppers.
not many power amps with high value stoppers, the allegedly super wideband Citation II used 10k.
Note that the grids of the output tubes are driven with quite a low impedance, not only because of the unusually low plate resistors in the PI, but also because both sections of the PI are under significant NFB.
incidentally, that Citation used 270 ohms in series with the screens, ultra-linear configuration.
I believe with that one you're opening a big can o' worms...