This thread is about Sonic's products and not general theory, but some of this might be very interesting to his customers and potential customers. If somebody is interested you can start a new thread (please link it to me in a private message) or just ask me in a private message. I'm happy to elaborate.
When did someone first hit on the idea to use FET's as source followers instead of tube cathode followers
Same afternoon as the first FET was made. Now in terms of putting FETs into "classic audio circuits", that would be the DIYAudio guys back when the forum was just new. The general idea is nothing new, the specific applications (what spot to use it in, and for what purpose) is the point.
Grid drivers were pioneered and popularized by Tubelab George (again, what a nice and knowledgeable person, truely a great person in the DIY scene), the gyrator plate loads by Wavebourne (same goes for him, a true legend in the scene by now).
I put these elements together in a balanced format, and made the output a parafeed type. Thus far parafeed was seen as an SE only thing, and output section gyrator loads were out of the question.
Also the idea of combining gyrator plate loads and CCS tails was out of the question for a long time. I had many a 10 page long arguments on this subject, people thought the whole thing wouldn't even produce audio output!
Spoiler, it does. And if you look at the theory behind, it obviously should work. But that subject is too large for this thread.
FET grid drivers and the open loop are 2 important distinguishing features of this amp's hybrid design.
If I had to say the most important, it would be open loop. The rest of the choices kind of flow from that point forward. The grid drive is very important, yes, forcing the balance is maybe more important and the horizontal load lines. These assure linearity despite no gNFB.
To get open loops AND low distortion is the clever part, as is no gNFB AND low output impedance.
Yes. Low output impedance comes from driving the OT primary (remember the OT is an "impedance lens") from a relatively low, few tens of ohms of impedance: the plate load gyrators!
The OT primary 'sees' two source followers driving it. That's a good spot to be in, you've got all the current you can eat right there.
So the fast transient speed of the amp mentioned in the reviews can be explained by reference to these design features, and also the open accurate soundstage which has also been mentioned. In rejecting the traditional way of achieving low distortion and low impedance by the use of gNFB feedback loops the amp manages to achieve that real, live sound that has also been mentioned which could not have been achieved had gNFB been used because of the extra loop that the signal must travel, which inevitably has a negative impact on sound as in SS amps.
Yes.
The gNFB loop lowers overall distortion to very very low numbers, but it changes the harmonics distributions. Distortion is made up of harmonics, they are multiples of the original signal. Say you input a 1kHz signal, then the amp adds some 2 kHz (this is 2H distortion), some 3kHz (this is 3H distortion) and so on.
8H and 9H are very very unnatural, these are ones that cause the sound to be "hard" or "bleak" or just unrealistic sounding. Get these OUT of the harmonic spectrum, and you get lively sound.
High order harmonics come in as a part of the gNFB tradeoff; the tradeoff is "lower THD" traded to "but tiny amounts of 7H, 8H and 9H appear". This is the "SS sound". It's caused by gNFB, not by the FETs or transistors in themselves.
Some of these design features go to explain why the amp could not be driven to distort as Maxx found in his review.
The amp does hard clip, oh yes. But not that easily with headphones at least.
At least my hearing is still logical.