I suppose that another approach might be to use the whole power amplifier as the Baxandall virtual earth stage. Then the power amplifier would need to be inverting overall. The Baxandall network would be at the input, partly in the forward path and partly in the feedback path. But when all said and done, I think it would be hard to argue persuasively against strict demarcation between tone control and power amplification.
As seen, Armstrong had used rather ersatz implementations of the Baxandall tone control in some of its radio/radiogram chassis in 1953, followed by a “proper” implementation in its FC48 radio/radiogram chassis of 1954. Between those events it had stepped into the hi-fi market proper with its A-10 amplifier and control unit, released circa 1954 June. For that it used a passive tone control circuit that is described as being of the Voigt type, but which was more like the James type.
Insofar as the Baxandall tone control was thought to produce a better set of curves than the passive type, the choice of the latter seems to have been a retrograde step. As typically deployed, the Baxandall required an extra triode or pentode as compared with the passive type, so the latter was favoured in situations where cost constraints favoured minimizing the number of valves used.
It could have been that Armstrong nonetheless simply preferred the curve set produced by its passive tone control. The A-10 was in any event a little unusual. The control unit employed a single ECC83 double triode, which provided initial gain, equalization where required, and the low-pass filter. This was followed by the passive tone and then the volume control, from which the AF signal was fed to the power amplifier. The core of the latter was fairly conventional, comprising an EF86 gain stage, half of an ECC82 as a phase splitter and an ultra-linear EL37 pair output. But that was preceded half an ECC82 gain stage that was outside of the main feedback loop – a pre-stage, as it were. I imagine that that made up for the loss in the passive tone control network (say 14 dB, perhaps more), and avoided the need for higher gain in the power amplifier. Normally one would have expected that extra gain stage to have been provided in the control unit, not the power amplifier. But the spare triode, as it were, happened to be in the power amplifier, which required only one half of an ECC82 for its phase splitter. So that might have been the reason for placing the gain stage there. And it just might have been the reason for the choice of a passive tone control. A Baxandall control would have required another triode (or pentode) in the control unit, and given that the A-10 was intended to be keenly priced for a 10-watt hi-fi amplifier, the budget may not have allowed for that.
Armstrong then used a similar passive tone control in its later radio/radiogram chassis, certainly in the Stereo Twelve/Jubilee/Stereo 44/AF208 generation, but I think before that in the PB409. Why Armstrong referred to it as being of the Voigt-type is unknown. Perhaps it saw that as the original of its kind, with others, such as the James, simply being a variation, in which capacitors rather than resistors were used to avoid the bass potentiometer from affecting the treble.
Cheers,