I wrote this years ago. I’m pasting it here. Demystifying the magic of transformers one post at a time.
And before the trolls begin trolling I do have an MSEE so be quick to think before you hit quote and reply to rebuttal any of this. It was my area of specialty in graduate school.
———————
The transformer doesn't have an impedance of its own; it only reflects a voltage (or impedance - just the square the turns ratios) from the primary winding to the secondary windings through a ratio of the number of turns of the secondary compared to the primary. Magnetic flux linkage in a ferromagnetic material due to Hopkinson's Law is the theoretical baseline as to why transformers work the way that they do.
The amount of magnetic flux in the ferromagnetic core is dependent upon current flow within the primary coil of wire, the number of turns on the primary, and the reluctance (magnetic resistance) of the ferrite core of the transformer. The current flow generated within the coil generates a magnetic field, which causes magnetic flux to follow the path of least reluctance. The magnetic flux then traverses through the ferrite material to the secondary windings; generating a current within the coils of the secondary through flux linkage. Voltage is then produced across the secondary terminals due to Faraday’s law of Induction.
In summary, transformers are passive lossy devices. Power in is power out minus losses. The impedance you’re changing on the secondary are strictly for maximum power transfer, stepping the primary tube impedances down to speaker load impedances. In essence, the transformer takes high voltage swings of transient plate currents from the tubes and transforms (hint, transformer) them into low voltage high current sources for the speakers.
I also want to note that there is a lot of magnetic flux stored in a transformer core - which can cause havoc to circuits when impedances are mismatched (primary OR secondary). This means mismatching secondary speakers to secondary windings reflecting an intended impedance from the plates, which may or may not be correctly matched with tube swaps to a particular transformer’s intended primary impedance, can cause transient responses which over frequency can be damaging at high volumes. Just keep that in mind.
—————
I’m in the camp that it’s perfectly fine to pull two tubes to double the primary impedance and then counteract the higher induced primary impedance by halving the secondary impedance. Just note that it’s harder on tubes with higher wattage speakers since they can inductively reflect a larger impedance swing back onto the plates of the tubes and short them out due to overvoltage. Those voltage limits in datasheets are stated as limits for a reason and the stored magnetic flux in an output transformer is no joke.
For simplicity sake it’s best not to pull tubes and half the secondary impedance when you have:
1) Old gassy tubes
2) Amps that have really high B+
3) Amps that are triode strapping the screen grids
4) When using dumb high efficient speakers with the capability of large inductive kickback.
Or any combination of the above. It’s a good way to short a tube out and blow a fuse.