How important is the quality of a triode used as a cathode follower

[ray_moth]

There are triodes that are not considered to be very good for 'normal' use. I wonder how important tube quality is for cathode follower duty?

[Eli Duttman]

Ray,

IMO, the important parameter is gm. Since the O/P impedance is approx. 1/gm, high gm is what I look for.

Poor linearity is taken care of by the 100% local NFB. While not particularly satisfactory as a SE voltage amp, the 12AT7 makes an excellent cathode follower. Even variable mu tubes, like the 6ES8, usually regarded as unsuited to HIFI can make acceptable cathode followers.

[Dave Cigna]

hmmm.... If the output impedance is 1/gm, and gm varies over signal swing, then the output impedance varies too. That can't be good!

It seems to me that a good CF requires a tube with parameters that are constant over the entire operating range.

-- Dave

[agent.5]

It seems to me that a good CF requires a tube with parameters that are constant over the entire operating range.

How about using a pentode instead?

[EC8010]

Originally posted by Dave Cigna
It seems to me that a good CF requires a tube with parameters that are constant over the entire operating range.

Sadly, all active devices have gm that varies with current. As Eli says, you want as much gm as you can get. Further, you want the signal swing to be a small proportion of quiescent current (to minimise the variation of gm).

[Dave Cigna]

Originally posted by EC8010
Sadly, all active devices have gm that varies with current.

Of course, but that doesn't imply that all devices are equal. Some show much less variation than others.

Originally posted by agent.5
How about using a pentode instead?

It's been said that they make a better CF because of their potentially high gm. But I have to admit that I find the problem of keeping the screen at a constant voltage WRT the cathode enough of a complication that I have not tried them.

-- Dave

[agent.5]

keeping the screen at a constant voltage WRT the cathode enough of a complication that I have not tried them.

Maybe use a separate power supply for the screen

separate transformer, rectifier, choke, capacitors, etc.

[EC8010]

Originally posted by Dave Cigna
It's been said that they make a better CF because of their potentially high gm. But I have to admit that I find the problem of keeping the screen at a constant voltage WRT the cathode enough of a complication that I have not tried them.

There's no more gm than if you strap that same pentode as a triode (usually very slightly less). I assume you're thinking of some of the final valves such as E810F and E55L. As you say, keeping Vsk constant is a pain, and in the few experiments I've made, not really worth the effort beyond novelty value.

[Johan Potgieter]

I think that is making things unnecessary complicated. As said, we "tolerate" gm variation in normal voltage amplifiers, and a CF is just the same with the input signal connected differently. Because of the inherent NFB is should basically be that much cleaner than the equivalent voltage amplifier, i.e. not the weakest link in the circuit.

I am also worried about the heater-cathode feed-through, though perhaps small. Rf tubes like the ECC81 is not designed for particular good h-k insulation, as in audio standard. I felt more comfortable raising the heaters to above ground potential; more positive than any cathode if possible. I infrequently had a small problem of this nature with r.f. pentode-triodes (e.g. ECF80).

[frank754]

I have four brand-new-in-box JAN-Sylvania 12AT7WC's I'd be willing to sell for $25 incl shipping anywhere in the US (dated 1974), since I have several of these.