> Are all limiters with a 6386...
It is possible to use the 6386 for other things. Could be an OK mike-amp or line-out tube. BUT there are a bazillion other tubes to do such chores, so in general if you see a 6386 it is doing variable-gain work.
> considered vari-mu?
Pet Peeve: The Mu may vary but that is NOT the key feature of an audio variable-gain tube.
Mu of 6386 drops about 1:3, maybe 1:4, but you can force the gain down over 1:100 and still get useful output.
I do not know of any audio gain-control scheme where shift of Mu (Amplification factor) is the primary effect.
As John says: "You don't have to use a tube with a remote cut-off grid", nor one with a variable Mu. In fact ANY tube "can" be used for electronic audio gain control (a limiter is such a gain-control plus level detector). The basic idea in all of these isn't vari-Mu, but Variable Transconductance (vari-Gm). I don't know why we call them "vari-Mu"; maybe "moo" or "meu" sounds more mystical than "gmm".
Any useful amplifying tube has Transconductance (Gm): change of grid voltage causes change of plate-cathode current. And Gm always reduces when tube current is reduced. In a conventional voltage-amplifier stage, gain is total load impedance times Gm. If Gm is reduced, gain is reduced.
But tube current normally falls faster than Gm. For a limiter, where maximum input wants minimum gain, sooner or later the current will be too small to handle the output level, and distortion gets gross. Using "ordinary" tubes like 12AU7, with a post-amplifier, you can get 15dB-20dB of gain reduction before it all falls apart.
That's where the tapered grid winding that John mentions comes in. This trick allows Gm to fall almost as fast as current. We still eventually run into gross clipping, but in audio use (push-pull plus a post-amp) we can easily top 30dB of gain reduction with 6386 or 6BA6 or a host of others (6ES8 is another good type, though it was one of the last TV-set tubes and can be hard to find).
In any case, the output of a vari-Gm stage in deep gain reduction is low. (The original intent was to drive 100K+ tuned circuits, not wide-band audio loads.) Most vari-Gm limiters follow the vari-Gm stage with an amplifier. Fairchild went a bit nuts: massively parallel 6386 with beefy iron can just-barely make recording studio line levels, and avoid the flavor of a post-amp. I think much of the sound of the Fairchild 660 is the super-simple yet highly-tuned audio path.
And in any case: a lot of the "working sound" of a limiter is the sidechain and the time constants. Fairchild has a massive sidechain and fine time control. Things like the 175 have "good enough" sidechains that will chip transients.
There are other ways to control gain electronically. LDRs are light sensitive resistors. Simple, cheap, no trims, but not precise. You can sharpen them up but then it makes as much sense to go vari-Gm. PWM has been used, but it isn't easy. Instead of using a tube as a gain-stage, you can use it as a cathode follower: reduce current enough, output falls. Since cathode impedance is just 1/Gm, this can be seen as another kind of vari-Gm, but the rate of Gm drop with grid voltage is very different. There is another form using plate resistance, either as passive load or as was done in the GE BA-6.....
> GE "uni-level amplifiers."
The BIG broadcast GE limiters use a 6386 front end, plus a post-amp (twin 6V6 to drive WATTS into long transmitter lines), plus a whacky feedback scheme that only happens at VERY high levels where the 6386 loses control. I think it can be pumped 35dB above nominal level and still not splatt your signal. There is no good reason a broadcaster would drop +35dB into a limiter, but stuff happens. Network feed accidentally gets patched through a mike-amp, etc. A lot of the old broadcast stuff was run with nominal level 30dB or 40dB below clipping to get low THD, like 0.1%. So it is possible to mis-patch too hot, wind up barely clipping at 3% THD but 30dB over nominal level, and then the BA-6 will whomp that down to 0dB and deliver a lightly-flavored loud an dlegal signal instead of splatting clip-garbage all over the radio band.
> I guess like matching 6L6's
6L6 has a bent-enough Gm curve that you could build an effective vari-Gm stage around a pair. You may not be able to get 30dB clean reduction without heroics; anyway two 6L6 is pretty heroic already (idle ~200V ~180mA). But the tubes are readily available on friday night.