> I just put the ground switch back in the hole as requested, of course it doesn't do anything.
To be too cute: run a 0.1uFd 600V cap, a 1Meg, and a 10 Ohm in series from raw B+ to ground. This gives about 0.1mV of raw buzz. Return your first stage cathode networks to the 10r instead of straight to ground. Wire the ground switch to short this.
Now it does something. And something familiar to 1960s players. When switch is set "wrong", the amp buzzes more than when set "other way".
To be more authentic, set one end to short the 10r not with a short but with another 10r. Gives 0.05mV buzz.
Assuming the build has zero buzz (this gimmick only makes sense if the build is super quiet), then the 3-way switch gives 0.05mV, 0.1mV, zero mV buzz in the three settings. This is very-like how the original worked in many rooms with 2-pin wall plug (except "zero" buzz hardly ever happened).
___________________
> I'd suspect the tube first.
Agree (maybe) about keeping "original" tubes in the signal path.
However signal does not go *through* the trem oscillator. It just jiggles at the user-set rate. Effect on signal comes further down the line. As as Steve is suggesting, in this circuit a hardly-weak 12AX7 may fail to jiggle itself, or not at high rate. Unlike a signal stage where low gain is just low gain and you can cover it with a higher VOL setting, the trem jiggle has a _minimum_ gain below which it just won't work.
So find a tube which does what it has to do, and don't worry if you end up with microphonic asian junk or even a MOSFET; that's not "the sound", just a side mechanism.