I think others have already stepped on these stones, but anyway....
> My gut feeling is that tubes make sense for higher-output cartridges (3.5mV at 5cm/sec and up) but they're marginal for cartridges with outputs lower than that.
No: phono pickups evolved so that the noise of the tube would not be a problem.
The moving-magnet design is not the easiest way to get a smooth high-end, but does allow lots of turns for high output.
Purists who went moving-coil for smooth high-end got very low voltage, down in the tube noise; but also low-Z so a transformer would bring the voltage up above tube noise. Hey, it works.
> if the noise from the mechanical tracing the vinyl is much higher than the thermal noise of the pickup coil (is it?)
For standard Shure/Pickering/Grado needles on vinyl, for any non-crappy conventional preamp, the surface noise is higher than the electronic noise.
BUT: what you really need to look at for phono-preamp noise is the noise below 1KHz. RIAA EQ for magnetic cartridges is a mild bump on a 6dB/Oct slope. There's a LOT of bass boost and treble cut. Phono preamps don't so much hiss as rumble randomly. So the conventional flat-amp concepts can lead you astray. 1/f noise will often be a bigger problem than straight thermal noise. Device specs often omit this data. And I have a feeling that the TV Tuner tubes, kings of Hi-Gm, were not tested for 1/f noise because it does not affect 50MHz+ operation. What little I know of thermionic 1/f noise is that you may have to sort several batches of tubes to get a few really good ones.
Surface noise is semi-flat, though to the ear it is mostly highs. So comparing surface noise hiss to amplifier noise rumble is apples to pears.
In a no-holds-barred design, current is a free variable. Since Gm rises with current (but not so fast), you might look at BIG current. 8417 working at 50mA-100mA in triode has Gm near 20,000uMho, Rk 50 ohms, noise resistance 125-150 ohms. I think as triode the Mu is around 18, so you can get some voltage gain. (However, grid-current noise on this beast may be non-negligible on a 47K-loading pickup.)
Conventional pickups work fine with 6SL7 or 12AX7 at 0.5mA-1mA (Dyna PAS), noise resistance of several K ohms. They were made that way. I think of "conventional" as "5mV"; if your 3.5mV is measured the same way, I'm sure a conventional tube preamp is fine. If you are fussy, you may reject 20% of tubes instead of 5%; a good preamp has another socket where you can use noise-reject first-stage tubes.
If your 3.5mV is Max, like a conventional needle gives 30mV-100mV Max, then tubes won't compete with transistors.
> a tube computer, where losing a flip flop somewhere could bring computations to a grinding halt.
Ah, but today we have Windows!
I had a new-build PC, loaded WinXP, surfed the Web, a few stray small oddnesses. The one thing it would not do is finish installing MS Office. After a day reformatting, I ran MS RAM Diagnostic. This found hundreds of RAM errors in 5 minutes: bad memory (flopped flip-flops). Yet Windows ran, and nearly normally.
How much pointless code is in there, that massive memory errors never caused a grinding halt, or even a "You see that?!!!" obvious screwup?
> Audio Research..... SP11, was a hybrid design.
So was my ~1958 ArKay. Looked like classic 6V6 mono amp, but had one transistor hidden in the phono stage.
> My gut feeling is that tubes make sense for higher-output cartridges (3.5mV at 5cm/sec and up) but they're marginal for cartridges with outputs lower than that.
No: phono pickups evolved so that the noise of the tube would not be a problem.
The moving-magnet design is not the easiest way to get a smooth high-end, but does allow lots of turns for high output.
Purists who went moving-coil for smooth high-end got very low voltage, down in the tube noise; but also low-Z so a transformer would bring the voltage up above tube noise. Hey, it works.
> if the noise from the mechanical tracing the vinyl is much higher than the thermal noise of the pickup coil (is it?)
For standard Shure/Pickering/Grado needles on vinyl, for any non-crappy conventional preamp, the surface noise is higher than the electronic noise.
BUT: what you really need to look at for phono-preamp noise is the noise below 1KHz. RIAA EQ for magnetic cartridges is a mild bump on a 6dB/Oct slope. There's a LOT of bass boost and treble cut. Phono preamps don't so much hiss as rumble randomly. So the conventional flat-amp concepts can lead you astray. 1/f noise will often be a bigger problem than straight thermal noise. Device specs often omit this data. And I have a feeling that the TV Tuner tubes, kings of Hi-Gm, were not tested for 1/f noise because it does not affect 50MHz+ operation. What little I know of thermionic 1/f noise is that you may have to sort several batches of tubes to get a few really good ones.
Surface noise is semi-flat, though to the ear it is mostly highs. So comparing surface noise hiss to amplifier noise rumble is apples to pears.
In a no-holds-barred design, current is a free variable. Since Gm rises with current (but not so fast), you might look at BIG current. 8417 working at 50mA-100mA in triode has Gm near 20,000uMho, Rk 50 ohms, noise resistance 125-150 ohms. I think as triode the Mu is around 18, so you can get some voltage gain. (However, grid-current noise on this beast may be non-negligible on a 47K-loading pickup.)
Conventional pickups work fine with 6SL7 or 12AX7 at 0.5mA-1mA (Dyna PAS), noise resistance of several K ohms. They were made that way. I think of "conventional" as "5mV"; if your 3.5mV is measured the same way, I'm sure a conventional tube preamp is fine. If you are fussy, you may reject 20% of tubes instead of 5%; a good preamp has another socket where you can use noise-reject first-stage tubes.
If your 3.5mV is Max, like a conventional needle gives 30mV-100mV Max, then tubes won't compete with transistors.
> a tube computer, where losing a flip flop somewhere could bring computations to a grinding halt.
Ah, but today we have Windows!
I had a new-build PC, loaded WinXP, surfed the Web, a few stray small oddnesses. The one thing it would not do is finish installing MS Office. After a day reformatting, I ran MS RAM Diagnostic. This found hundreds of RAM errors in 5 minutes: bad memory (flopped flip-flops). Yet Windows ran, and nearly normally.
How much pointless code is in there, that massive memory errors never caused a grinding halt, or even a "You see that?!!!" obvious screwup?
> Audio Research..... SP11, was a hybrid design.
So was my ~1958 ArKay. Looked like classic 6V6 mono amp, but had one transistor hidden in the phono stage.
