Filament Autobias

[kenpeter]

Does this make any sense, or just a new way to kill a tube?

Note: Total of cathode and fixed bias here at -17V.

Prolly shoulda thrown 330 or more at Q1 base too.

[kenpeter]

Hmmm... 0.9A emitter current could require as much as 20mA base current.
May have to rethink if this can be done in a single BJT? Darlington no prob.

I wonder if modulation of filament temperature can affect quiescent current
in the way I am hoping, or temperature only sets an upper limit for emission?

[Merlinb]

There is such a long delay between a change in heater current and a change in anode current, that I wouldn't be surprised if the whole thing oscillated at ultra-low frequency.

[kenpeter]

I won't debate that there is a huge time constant, but would have
to prove over 120 degrees phase shift and loop gain over unity.
We just need to make sure that R3 C2 corners far from the
same frequency as the filament time constant. So gain is down
before we have to deal with more than 90 degree phase shift.
Same concept as "Sluggin the dominant pole", er whatever...

I don't know the time constant of the filament, I don't think it is
anywhere near the same length as a full warmup. R3 C2 corners
at 4Hz as drawn, so we hope filament gotta be slower than that?
We could probably get away with 100uF and corner at 20Hz.

I'm not counting cathode impedance or base leakage into my RC
calculation. It might not be as low as 4Hz like I'm thinking. This
transistor really needs upgrade to Darlington or Sziklai for reason
of insufficient beta (20mA base current as drawn). Impedance of
Darlingon base would increase sufficiently to put it out the RC,
cathode impedance might still be a factor of some significance.

[Wavebourn]

Originally Posted by kenpeter

or temperature only sets an upper limit for emission?

Almost.

[Arnulf]

Originally Posted by kenpeter

Does this make any sense, or just a new way to kill a tube?

The words that springs to mind is "Why ?". At startup you're going to subject the heater to the better part of 12V and multiply thermal shock to it as if its low R at room temperature wasn't enough of a cause for concern, and when it warms the cathode up it just might oscillate, as you have already pointed out.

Why not opt for separate CCS heater supply instead ?

[kenpeter]

Reason I had thrown the 1ohm at it, 12A max.
Valid point. Deserves much safer current limit...

Why? Throw 8 mismatched 6L6 at an AB amp
and know all plates are biased just under 30W
at all times and conditions. Even as tubes age.
Run filament colder when new, hotter when old.

My first theory was auto minimum idle bias for
smooth AB crossing. Then later realize just run
plates exactly hot as they are rated at all times
probably crosses no worse.

It will auto itself to class A if AB isn't needed.
Could auto itself to class B, if stressed? Maybe
emission limit already prevents that situation?

----

What I worry most: Filament may only cap the
maximum emission, little or no effect 60mA idle?
Just have to try it and see... I havn't found any
good paper on what really happens.

[kenpeter]

3.4A max inrush assuming dead short filament.
4mA max base current assuming worst beta.

There has to be a simple way to limit inrush
down to an amp-point-two-er-so? Thenagin'
a filament isn't a dead short even when cold.
Well, it ain't suppsed to be anywho. Where'd
my darn multimeter go? If I knew that, prolly
wouldn't know where I stashed my Pentodes.

Still wasting my time if only an emission cap.
Need to resolve that truth or fictionage first.

[kenpeter]

Nevermind... I can't see how this achieves anything
that individually bypassed cathode biases would not.
If I am content to run all 6l6 at max plate and let
the resulting class drift as it may, whats difference?

Nother fine excorcize in pure stOOpid...

[Wavebourn]

Since you have such a tricky feedback by current with 2 long time constants you can get better result more directly: CCS in cathode shunted by a capacitor that gives the same time constant without stability concerns.

However, if you want to prolong tube's life increasing emission when it ages, you still can add such a through-filament feedback, but sensing cathode voltage instead of voltage drop on a resistor.