more circuit changes

 A couple more things have changed in the version 2 design.  Updated schematic coming soon...

For one thing, in the fx loop return circuit, I'm using two inverting stages with negative feedback, to multiply gains together and create a clean and stable 100x non-inverting gain.  In the first version of the amp, I have used negative feedback triode stages in several places, and I just used a single coupling capacitor, shared between the output and the feedback loop.  However, I noticed that Howard Dumble used separate branches with a capacitor for each one; I got to thinking, maybe he knows something here, maybe it's better to keep the branches more "isolated" from each other.  So in version 2, I drew the schematics using the two-capacitor topology.  But then I encountered more discussion of NFB triode stages in the Aiken Amps pages (a great resource for tube amp design, along with Rob Robinette and others).  Aiken seemed to imply that the single-capacitor topology is best in most cases, since it's good to keep the output circuitry "within the loop".  So now I'm convinced the other way, and I'm going back to single, shared capacitors in the fx loop return stages.  (I'm very proud of myself, I came up with the general concept of using opamp-like NFB circuits and virtual grounds with tube stages by myself, before I ever encountered them in the literature.  So I'm happy to have confirmation from Dumble and Aiken, that these ideas really work with tubes.)

The other significant change will be in the v-mid eq circuit.  In version 1, I built a homebrew circuit which does work, but it's not very "effective"; i.e., the amount of boost or cut is not very pronounced, and also the frequency range is limited.  My circuit uses two triode stages, yet in a way it's effectively a passive circuit (in my estimation), because neither of the amplifying stages participates in frequency-dependent operation.  The first stage is basically a phase-splitter like one might find driving a push-pull output section.  The second stage is a NFB summing amp which combines the mid-eq signal (boost or cut) with a "dry" (flat) signal; this latter topology in particular limits the effectiveness of the eq.

So in the new version, I am going to try to run the eq in an active mode, as one sees with Baxandall active circuits.  One end of the level pot will be driven by the input (buffered by a NFB triode stage), and the other end will be driven by the output of the second stage.  The tunable R-C bandpass network is fed by the wiper of the level pot.  Output of the R-C network goes to the second triode stage input.  So the second stage is again a virtual-ground NFB circuit, but now the virtual ground is frequency-dependent.  And the "dry" bypass is eliminated.  This should permit more dramatic alterations of the eq curve, boost and cut, while still keeping the desirable center-flat position.  I will most likely prototype this circuit section before I build it into the new amp.