I'm not firing on anything like as many cylinders as Joe implies just at the moment. I've just tested positive for that thing on the go at the moment. Big headache and heavy cold etc. Yuk!
Anyway, the 12AT7/ECC81 was designed as an RF valve with useful gain to much higher frequencies then the ECC82 and ECC83.
Read this as meaning that it can get up to mischief at much higher frequencies. As a consequence the circuit design techniques and construction techniques have to move up to a higher league. The fashionistas and tube-rollers looking for the glamour of more exotic devices all seem to be bothered with audio and haven't really built up the necessary experience in RF design needed to keep RF devices tame. No matter what frequency the intended operating frequency range is, if it oscillates outside this range, then it will still affect things in the wanted range.
Designing a feedback loop is a sort of race. The aim is to get things to behave gracefully at higher frequencies, for the gain around the loop to be rolled off before the phase shift builds up to a point which either causes oscillation or bad overshoots. Usually this trouble is at ultrasonic frequencies. No-one wants this. Audiophools think there is no limitation on the discrimination of their hearing so oscillation at any frequency is offensive. Mathematical engineers do believe their hearing has limits, but understand matters like RF intermodulation distortion, spurious oscillations consuming dynamic range etc etc.
The snag is that the business of analysing or designing loop stability is fairly mathematical and uses concepts somewhat beyond A-level in our educational system. To compound this, it is taught maths first, and only then do they tell the students which didn't run away how useful and necessary it is... and show them the short-cuts!
This knowledge is needed to not only design a good loop, but also to assess one someone else has published. This is highly ironic. The only people who can look at a schematic and say whether it's worth building are those people who could design a good one themselves. Nature can seem very cruel.
Everyone else winds up building things to find out. This can waste a lot of their time and money, but worse, it damages their confidence and that is precious.
Another irony is that the maths to analyse/design feedback loops isn't actually hard. It involves some very weird concepts which scare people off. Some folk have to do it because they need it. I think I may have been too thick to have been scared off.
Anyway, computers allow us to plot charts which allow us to easily visualise these things, adn it comes down to playing chess, moving gain factors, roll-up and roll down frequency breakpoints around to bend curves so that the gain is rolled away before the phase can circle and bite. (very bad pun, no apologies, I've been wanting to use that one for some time).
There are several people around this forum who are comfortable with these concepts, and occasionally post some clues.
Changing the subject, somewhat. Valves look good. Shiny, warm and with a glow like a welcoming log fire in winter (Christmas decorations and Bing Crosby are optional) They are usually mounted up top on full display so everyone can appreciate their charm.
On the other hand, transistors got ahead of them many years ago,
especially for the sorts of jobs found in preamps. In power amps, valves have some useful characteristics, especially if overdriven, which transistors don't do so well. But in preamps this isn't going to do anything. So, if you see a preamp or a CD player with a token valve stuck out the top of it, have a good laugh. It's a handy indication of a piece of crap set as bait for the wallets of the gullible. It's like the sort of guy who swears his car is faster now he's stuck go-faster stripes on it. I don't suppose we should laugh, it's a bit sad really.
So, if you want a good preamp, you can do a better job with semiconductors. Over 40 years ago the NE5534 opamp came out. It was spectacularly good for audio, reliable and easy to use. There have been newer parts introduced, with advantages in some specs, but usually at the expense of others. Also the '5534 is cheap. (NE5532 is the dual version) So do a preamp around '5534s. Chances are every record/CD you have that was done in the last 40 years has had its signal passed through lots of these parts. If 5534s are not good enough for you, you'll have to throw all those records away
Decca pressed some gorgeous recordings back in the 50s and 60s, so the 5534 class opamps have no monopoly on good audio, but there is nothing wrong with the 5534 either - it just makes life a lot easier.
My amplifier setup was designed and built 40 years ago. It uses NE5534s in the preamp. I've seen no need to change it in the intervening period, I've just got on with listening to music.
David