[Radio Wrangler]

The ECC83 is a fairly low current valve, and so the stray and grid capacitance of the following stage is going to introduce more phase shift than had something running lower impedance in its anode circuit been used. The phase shift puts more limitation on how much feedback you can use, or how much stability margin you get for a given amount of feedback.

The Radford ECF82 phase splitter was invented by Arthur Bailey. Quite a smart cookie. It's rather elegant and has sufficient advantage that I can't understand why anyone does it differently. Arthur Radford was an excellent transformer designer, and together the results really were synergistic.

Using fixed bias will condemn you to keeping checking quiescent currents and twiddling them. Cathode 'automatic' bias loses a little potential output power, but frees you from worrying about the amplifier so you can get on with listening to the music.

Unfortunately, some valves and some variants have been decided to have certain special properties which only true believers can discern. But those true believers assure the rest of us that the differences are actually very large and dramatic. In order to fail to notice them, the hearing of the rest of us must be dreadful or our powers of perception must have fled.

As evidence of the magnitude of these differences, the believers are prepared to pay fortunes for specific valve variants. It's definitely a case of putting their money where their mouth is, and rather a lot of money at that. Odd that the rest of us hear so little.

There are trends which surprise me: No small differences are ever described, no dead heats. You'd expect small differences to outnumber larger ones. There is a mathematical law involved that small things outnumber large things in general. Rare/expensive/hard to find/exclusive things always seem to be rated better than the mundane.

I have a couple of ECC83 in stock because I have a couple of receivers which use them. I can't tell you their makes, or the colour of their markings because I'm not bothered. They tested OK when I checked them, so they'll work just fine if the sets ever need them.

EF86 are well known that some can be very microphonic. Not specifically any manufacturer, shape of getter or whatever, just some individuals. You wind up having to try them out and tap them with a pencil. It's something to be wary about if you use them. This is a real difference. Repeatable, audible by people irrespective of belief-system, measurable even!

The cathode potential running autobias shouldn't cause you to have to up the screen voltage. It should be still within tolerable bounds.

Avoiding microphony is one reason for ECC83 input stages, but you would lose open loop gain. So maybe an ECC83 cascode? But many power amps of the era were excessively sensitive. maybe you'd be better designing for a larger input from the preamp. This would let you use an ECC82 cascode.

David