The output voltage of a tube stage is much higher than the bias needed for the next stage, we should eliminate the DC and let only the AC arrive to the next stage.
It is no more used a lot, quality transformers are rare and so
are their manufacturers.
A transformer cannot transform DC, only the AC can reach the
secondary. Another advantage is the possibility to use a
transformer that elevates the AC voltage, in this case the first
stage should have a low output impedance and the second one a
high one, which is usually the case. As the transformer is driven
by a DC current, there should be an air gap in its magnetic
circuit to avoid the saturation of its core which would give poor
dynamics and distorsion.
The DC is stopped by the capacitor placed serially with the
input.
It's the most common type in use.
The quality of this cap should be very high, the signal passing
through it.
The resistor value fixes the input impedance of the second stage
(the grid has a nearly infinite impedance, at least for the tubes
without grid current).
This circuit is a voltage divider where the first resistor is the
cap's reactance and the second the input impedance of the second
stage.
where C = coupling cap
Rg =
grid resistor
Fmin
= lower frequency to pass
The less components the signal goes through, the better our
ears feel ...
Thus the idea to suppress any coupling element.
To achieve this, we must increase the cathode voltage of the
second stage to keep a good bias voltage for the second stage.
That is to say, we must increase the second stage supply voltage
and increase the cathode resistor so that the voltage drop across
it corresponds to the first stage output voltage minus the bias
voltage needed for the second stage. For the direct heated
triodes, the heater voltage is polarised to achieve this same
condition.
The Loftin-White schematics use this principle.
Example with 2A3 tube.