> I think you also get a fairly constant 600 ohm output impedance
Dim memory?
The 200's output runs about 90-110 ohms across the audio range, rising above 10KC or 20KC. It is a feedback amplifier, so basically low-Z, but in most of the several models the feedback is taken from a separate winding so the output winding resistance and leakage inductance are not cancelled. Leakage inductance is non-trivial, since to get good 20CPS response and low distortion they put many-many turns on the core.
Notice that a pair of 6F6 or 6AQ5 are good for 8 or 10 watts; the 200 "underloads" them with a "too-high" primary impedance so it "only" delivers 24V or 1 Watt in 600 ohms.
At most audio frequencies you can get around 2 Watts in 150-200 ohms: I have seen smoke from a 1W 220 ohm resistor left bridged across the high-power low-THD audio-only (20CPS-40KC) 200CD model (rated 40V in 600 ohms).
Any decent audio shop of the period would have a "Gain Set": a box of pads and meters that you would put between your wanky signal source (before the HP-200, most audio sources were wanky) to set your exact levels (and build-out impedance). Sadly all the Gain Sets that once existed have vanished; we saw one in the Worst Bench contest at the old place, and that may be the only one I have seen since the 1960s.
> at least two major versions of the circuitry, with the newer version using small tube, I believe, and providing better distortion specs.
The 200-series were in production about 20 years, very "exciting" years for electronics. So there are several feature sets (wide range, low distortion) and internally several generations of parts. Both Octal and Miniature tube-sets were used, and I think each type was produced in most of the feature-set models.
The 1950 models did not get the very low THD levels of the later models: no demand for sines much cleaner than tenth-percent. I'm amazed at how clean the 1960- models got, considering that the "lamp stabilization" won't work without large nonlinearity in the oscillator's amplifier.
> There is a light bulb in the oscillator ... DON'T mess with it!
AGREE AGREE AGREE AGREE !!! (Though if waveform is large and distorted, gently check that it is tight.) It will never burn out, it isn't supposed to light, the filament hardly gets warm, but is critical to getting the level right. Also do not fiddle with the trim-pot near the lamp: unless you have to replace the lamp, the factory trimmer setting is good forever.
> In most cases, I've found the fault to be with the capacitors (the fixed ones, not the big air-variable) in the frequency-determining Wien bridge circuit.
I have not seen those go bad. But they are the kind of part that would vary in production, and they could have got a bad design. Mine are ceramic body with mica insulator, lovingly built, and will probably last a lifetime.
You can get similar no-start trouble when the coupling caps dry up, especially the 20uFd(?) can-cap that couples the oscillator 2nd stage plate back to the Wein bridge. The exact value for this cap is not critical: I bridged mine with 47uFd.
A HP-200 will always bobble when set near 60CPS with the cover off. I once spent days trying to fix this with better power supply filtering. It goes away when you put the cover on. D'oh, that huge air-cap is super high impedance at 60CPS and sucks all the hum from the room to beat with the oscillator. The case shields it excellently. The frequency and level-vs-freq also shift when you put the cover on, so do your final check and calibration with the cover or you are wasting your time.