In the above description, I'm working from the
B+ voltage to ground, the plate voltage will be several volts lower:
The output transformer primary winding has a DC resistance of its own, anywhere from under 100 ohms to over 200 ohms, epending on the OPT being used. So just think of it as behaving like a big wirewound resistor, at least for DC measurement consideratrions. This is the same, with either both push-pull and single-ended circuits.
Note, that in most OPTs, the DC resistance of the two plate connections of the primary, to the center tap will be about 10% difference, this is due to the layer-winding technique generally used, and each layer of wire and the insulating material will add up to create an increasing circumference, thusly a longer length of wire, even though the actual number of turns is identical on each side of the center tap.
The one exception to this will be in a bifilar-wound transformer, and these tend to be a lot more expensive "gold-plated" high end items.
So, as an example, the measured B+ votage at the OPT center tap (usually a red wire) is 370VDC, the winding from plate to CT is 200 ohms DC resistance, the cathode resistor is 400 ohms, and the DC voltage across the cathode resistor is 14 VDC, which indicates a combined plate and screen current of 35 mA.
In the SCA-35 or ST-35, with the ultralinear screen grid taps on the primary, the combined plate and screen currents will be the same 35 mA cathode current - this is all a series connected circuit, for our calculations. The screen current is generally quite low, so we can just let the plate take all the "credit" as far as current consumption, in order to keep the mathematics somewhat less complicated. Just think of the EL84 as behaving like a hefty power resistor between the cathode and plate, for static DC measurment situations. The plate dissipation is the product of the voltage drop and current between the cathode and plate
So, in our theoretical amplifier circuit, we have a voltage drop of 7VDC through the primary winding, and 14V across the catrhode resistor, for a total of 21 VDC, which is then subtracted from the 370VDC B+ voltage, which will be 349 VDC. Multiply that by 35 mA and the plate (and screen) dissipation is 12.215 Watts. OK, we're not doing brain surgery here, so we can dispose a couple of the less significant decimal places
12.2 Watts is more than close enough, or even 12 Watts - with tube circuitry, 5% tolerance is more than "close enough" precision.
I hope that this is somewhat clearer than mud
/ed B