UniGaBuf, a follower cut out to be a leader

[Bigun]

I'm always on the look out for a current buffer that I can put a triode VAS in front of. Why would I pick this options over a number of alternatives (I'm not trying to be combative, just want to learn why) ?

 

[Elvee]

Why would I pick this options over a number of alternatives (I'm not trying to be combative, just want to learn why) ?

I certainly wouldn't push any unwilling individual towards this option.
It is undoubtedly an interesting one: it has the ability to produce a result that transcends the crap material it is made of, but the beast isn't easy to tame:

I've had some problems, one totally unrelated to the topology, a very unusual defect of one of the transistors and the other was a high frequency oscillation (several MHz).
I could trace the origin of this oscillation: it was caused by the wirewound 0.33 ohm emitter resistor.
I made a quick and dirty fix, a 100n ceramic in parallel. Probably not optimal but sufficient to go on with the tests.

This could be incompatible with:

If the machine produces tranquility it's right. If it disturbs you it's wrong until either the machine or your mind is changed.

You have to decide for yourself.....

 

[CBS240]

I'm always on the look out for a current buffer that I can put a triode VAS in front of. Why would I pick this options over a number of alternatives (I'm not trying to be combative, just want to learn why) ?

Uh, er... HEC mosfet follower with paralelled output devices and a beefed up bootstrapped driver stage.🙂

I currently am devising something like that using a low gain ECC88 bias at a constant plate current as a balanced VAS, driving 2 HEC follower stages for a balanced output. The EC circuit has enough emitter follower buffers that it has a very high AC input Z, so as not to load the plate.

 

[Elvee]

J. Broskie also has some nice, proven solutions. They are a safe option, but of course not half as interesting, challenging and effective as an UFO like Unigabuf 😀

 

[Bigun]

but the beast isn't easy to tame

Yeah, it's always difficult not to get attracted to these new ideas, they look like a lot of fun but you are right that I wouldn't necessarily go in this direction if I'm after something relatively foolproof. And at the moment I'm leaning more towards the latter, but I always keep an eye out for the new.

I've not considered HEC. I've not seen it used extensively and am not convinced from what I've read that it offers a real benefit over many other negative feedback topologies. Perhaps it's an idea worthy of renewed interest.