Use a MOSFET source follower instead? No wasted heater power, no heater lift required, much lower internal resistance, takes less space, costs about the same as the tube socket. It's just a follower after all, it's not changing the signal.
The traditional solution, sure. 🙂
But what do you do when there are other tubes (with cathodes near zero volts) that need to run off the same heater power supply? Or when both the common-cathode stage, and the CF stage, are in the same 12AX7 bottle?
In the latter case, warmed up and running, one cathode in the pair is at +1.5V, the other at +200V. At turn-on, the common-cathode stage has its cathode at zero volts before the tube warms up and starts conducting. It's anode is at B+ voltage, so the CF stage's cathode also rises to the full B+ voltage (say 300V). One cathode at zero volts, the other at B+ voltage. 😱
There is no happy heater elevation voltage when one cathode is at 0V and the other at +300 V. Nor when one cathode is at 1.5V and the other at +200V. Vhk is just too big. 🙁
The obvious workaround is to use a separate half-12AX7 for the CF stage, and elevate only it's heater supply, maybe to 150 volts DC or something. But that approach comes with several overheads - an extra 9-pin socket, an extra 12AX7, and an extra triode you can't use for anything other than a CF, because it's partner has its cathode at +200 V!
By contrast, generating a low-current negative supply rail only needs one rectifier diode, a couple of caps, and a resistor. It can tack on to the existing power transformer. Relatively simple and easy to do today, just unconventional. So I thought, why not consider it?
With the added negative rail, and using the circuit I posted last, quiescent CF cathode voltage can be adjusted to be just a few volts DC. Worst case, at turn-on with cold heaters and full B+, the CF grid rises to halfway between the negative rail and the positive rail; with -150 and +250 rails, say, that puts the CF grid at +50 volts. That's well within the heater-cathode insulation's voltage rating, so it shouldn't reduce tube life.
Part of my motivation is that I bought a stereo "tube Hi-Fi preamp" board off Ebay a while ago. Forget the Hi-Fi part, the board will accept two 12AX7s, and has large tracks and generous track spacing, so it's easy to hack the circuit into a (tube) guitar preamp instead. Basically, I viewed it as four triode stages, waiting to be joined up into whatever I want. That opens up a *lot* of possibilities!
One thing I've never built is a guitar preamp with the Bassman-style DC coupled gain stage/ cathode follower, so I started thinking about that. If I could solve the excess heater-cathode voltage problem, I could have both stages in one bottle, meaning I could build a clone of a Marshall 2204 preamp or something similar.
-Gnobuddy
That's surely what Leo intended, but fortunately for rock guitar, he (or his tech) screwed-up in a big way, and he quite accidentally created an extremely nonlinear stage that very gently and very progressively squashes the signal peaks. Lots of "rawk", relatively low harshness!
What happened is that the anode of the common-cathode stage is at quite a high voltage, something like 180V - 200V DC. So the direct-coupled CF stage has its cathode at close to this voltage, and with a 100k cathode load, it has to flow something close to 2 milliamps of cathode current. At the same time, the CF anode is only at some +300 volts, so there's only maybe 100 - 120 volts between anode and cathode of the CF stage.
If you look at the 12AX7 curves, and try to make one flow 2 mA with only 100 volts Vak, you'll find you have to bias it very hot indeed - so hot that a little grid current has already started to flow.
And so, the CF struggles to produce positive output half-cycles - it can't deliver enough current, so it flat-tops them. Meantime, the common-cathode stage driving it will flat-top (flat-bottom?) negative half cycles if driven hard enough, so the signal coming out of the CF stage progressively distorts, first with the positive half-cycles getting flattened, and then as overdrive is increased, the negative cycles too.
Leonidas wanted a clean buffer to drive the tone-stack in his *bass* guitar amplifier. Instead, he created a bad bass amplifier that turned out, quite by accident, to be one of the earliest good rock-and-blues (tenor) guitar amps, with heaping spoonfuls of tasty distortion. Everything Leonidas didn't want, and didn't expect. 😀
Merlin Blencowe spells it all out beautifully in his preamp design book. He has an abbreviated version on his website: The Valve Wizard
I may be worrying too much about the whole "The cathode is at 200 volts, woe is me!" thing, but the Bassman and some of its later clones and derivatives apparently have a reputation for eating preamp tubes. Blencowe suggests heater elevation, but the more I thought about it, the less convinced I was that it would ameliorate the situation sufficiently.
And so I came up with the negative supply rail idea. It seems to look good on paper. Mebbe I should build it and see what happens.
-Gnobuddy