Let's have a look at that amplifier/metering circuit Martin posted a few posts back.
T1 is not just a transformer, it is also a resonator. Job number 1 is to provide isolation from the HT hanging about on the anode of the valve under test. Job number 2 is to filter off mains frequency components because this is an AVO valve tester and the electrode voltages are unsmoothed mains waveforms from tapped transformers. Maybe this is the reason for 15kHz, to get the Gm test signal further away from 50Hz and its harmonics? Job number 3 is to filter off any harmonics of the 15kHz test signal, otherwise valves with lousy harmonic distortion behaviour will score erroneously wrong better Gm figures. Hopefully the Gm test signal is small enough that job 3 is trivial, but job 2 is very important.
R1 is critical, it is the resistor across which the valve anode current is sensed. Accuracy affects results proportionately.
The 1.99uF (unnumbered, but I assume C1) resonates the input (primary by definition) of T1 as a series tank. The slotted tuning slug in T1 allows the tank to be peaked up on the frequency of the oscillator board, so long as it's within a couple of percent.
D3, D4 are clamping diodes to limit the size of transients, eg from people operating range switches, protecting the transistors. Mounted on the underside of the board they look like later additions.
C2 1990pf 2% acts like a minor part of the overall tuning capacitance. ~The primary and secondary of the transformer are so tightly coupled, magnetically, that despite two windings they act as a single resonator, giving a single pole response.
(By contrast, a traditional radio IF transformer has its windings spaced apart to give rather weak coupling. Each is resonated with its own capacitor almost independantly of the other. There are thus two lightly coupled resonators giving a two-pole response which falls off twice as fast as a single-pole response.)
C2 won't work terribly well, because it is working in series with the rather uncertain input impedance of VT1, but it's an attempt to sample off only a small fraction of the power in the resonator.
R2 C3 is an extra well decoupled derivative of the power rail so that less power rail noise gets to the bases of VT1 and VT2 through their bias resistors.
VT1 is just working as an emitter follower to make as high an input impedance as possible so as not to load the resonator. Note that there are no RF stopper resistors applied to this transistor and its collector goes straight to the (decoupled) power rail. The base sees a low Z ar RF through C2 and the stray C of the transformer. Could this be an area which goes silly if a high performance modern transistor is dropped in, replacing the 2N2926?
VT2 is a common-emitter amplifier (we'll come back to what C7 is doing) which drives VT3, another common emitter amplifier with its emitter resistor fully decoupled to get every ounce of gain out of it.
VT3 drives D1 and D2 (their numbering is another clue that D3 D4 were fire-fighting fixes). These drive the meter with a fullwave rectified version of the signal.
Now comes the clever bit. C8 and C9 act as the capacitance across the meter to smooth the signal going to it. They also provide the DC block missing from the path from VT3 collector to the subsequent circuitry. By having two capacitors in series, they also give a centre tap effect for the signal, combined with the diode action, this point reconstructs the full AC
current waveform in the output circuit. The meter sees it rectified, but the voltage on R14 which senses the full AC component is there to be fed back through C7 into the emitter of VT2
So the input of VT2 senses the difference between the signal voltage coming from the emitter follower, VT1, and the current actually passing through the meter.
VT2, VT3 are a feedback amplifier with plenty of gain. The feedback loop compensates the diode drops in D1 and D2, it even compensates for meter resistance.
The purpose of C10 is now revealed, it is a compensation capacitor to roll off the loop gain so that it is stable at high frequencies.
So there you are. No rocket scientists were slaughtered in designing this circuit. There are a few omissions by modern standards (we've learned a bit since then) but it's quite clever.
David