On transistors with output transformers. "To begin with, my explanation won't satisfy the academics. So why use a step-down transformer in the output stage of a class A amplifier? If the winding ratio is 2:1 for example, the output voltage will be 50% of the voltage at the transistor. Accordingly, the loudspeaker impedance as seen by the amp is multiplied x 4. A speaker with an average 6Ω impedance will load the output transistors at 24Ω. Into such a higher load, a transistor usually moves to superior linearity. Better put, the potential voltage swing in that more linear operating region is higher, so of benefit especially to a class A transistor amp with only two Mosfets per channel which must produce high voltage swing whilst remaining in their linear area across the entire swing. This is best managed when a transformer matches the load impedance to the transistors' output impedance.
"Then why is this solution so very rare? Because transistors can work well directly into speaker impedances. It simply means global feedback to stabilize the circuit and reduce distortion but with it also bandwidth and gain; plus a big output capacitor to further degrade the sound. Of course a transformer costs 100 x more than a capacitor, weights 1000 x more and has more limited bandwidth. And, it is almost impossible to find a maker capable of winding such a specialized transformer. Basically you're on your own. But note that based on just cost, weight, dearth of suppliers and potentially poor technical data, the use of an output transformer can't be justified. With the technical models we use to describe the behavior of capacitors and transformers, it is very easy to demonstrate how any film capacitor will always beat the very best of transformers. There's simply no comparison if one runs an oscilloscope and signal generator. It's child's play to prove. So the conclusion could be that a cheap capacitor always and easily beats the best transformer if measured with a signal generator on one side, a fixed 8Ω power resistor on the other and an oscilloscope and spectrum analyzer at the output.
"But that's only part of the real picture where we have an actual complex transient music not steady-state input signal, loudspeakers at the output, sound traveling through the air and finally our human brain decoding it all. I have much experience on the matter including double-blind tests with people that were entirely unaffected by audiophilia. Listening to playback of any music program, anyone can easily tell that the sound arrives with different loudness patterns. With the combined sound of an orchestra playing in the background and a fortissimo dynamic from any solo instrument passing through a capacitor, it seems that the instruments move toward you so the illusion of the stereo image constantly shifts with loudness. Amplifiers with output transformers maintain a stable stereo image. This isn't a personal opinion or wishful thinking. Many double-blind tests with ordinary listeners confirm it.
"Just so, I don't know of any technical model we use to design amplifiers which explains this effect. We can very easily explain why a cheap capacitor will always beat the best transformer and this remains true for applications in the industrial, medical and aerospace sectors. But when listening to music, it's evident that this model doesn't account for everything. To the point, Olimpico could work just fine with a big coupling capacitor. At the time I did exactly such tests with it and LaScala. There simply was no comparison. The output iron always won. Meanwhile Spice showed that the capacitor was better. So who do you trust, your ears or the software program?
"To reiterate, my signal-path iron allows me to use just six components from input to output to have a true 120Wrms class A amplifer. The choice came down to engaging the world of magnetism and its many associated issues of energizing multiple complex windings inside a transformer with minimal losses and cross-coupling; or dealing with the world of dielectric fields where inside a capacitor there is a conductive foil separated by a dielectric which charges and discharges constantly. Finally, using global feedback to match impedances is, in my view, the enemy of our ear/brain's decoding system. From a purely theoretical perspective meanwhile, feedback is purely beneficial when our required bandwidth is just a few tens of kHz. And I'm not saying that amplifiers with feedback can't sound good. Each solution has advantages and disadvantages. But trusting my ears, my decision was clear.
"I've been winding transformers since 1986 when one of my first applications was an SMPS. At the time there were no standardized transformers so it meant pure trial 'n' error. Then I made an output transformer for a 6L6 guitar amplifier and things continued from there. At present, this never-ending story involves a friend even crazier than I who talked me into finalizing a field-coil woofer. We have in our hands an original plus all the details on a number of other vintage field-coil bass units. We know how '50's and '60's American products were produced under license in Germany and Italy. Almost all of those companies have disappeared but their documentation hasn't. And today it is easier than ever to source astonishing magnetic materials to pursue plenty of opportunities.
"For my CanEver output transformers, I went with an EI not double C core geometry. This is built around a 96% silicon-steel core with 4% of another magnetic material. The EI layout enables the necessary sandwich. With the currents of a 120Wrms class A amplifier, the biggest issue was how to manage the physical vibrations caused by the audio signal. Thus the transformer is highly sectioned, its windings are properly interleaved and distributed and finally everything mounts inside an elastically isolated stainless-steel cowl potted with a mix of gypsum and sand to shield and damp the magnetics."
