Ian Thompson-Bell wrote

Perhaps we are more inclined now to use valves outside of
the niche applications for which they originally
competed. I note the 6AU6 was intended for radio. Knowing
nothing about radio, I can't guess why low current in
triode mode is not given as a typical condition.


The 6AU6 is a bit odd - it is labelled as an RF pentode -
but it seems to have been used a lot in triode mode in low
level audio applications. There are examples in RDH4 and
there's even a short paper on using the screen as the
anode in triode mode to get better screening of the
heaters.

I vaguely remember from previous discussions that screen
current is unusually high as a proportion of cathode
current.

My RF knowledge is sadly lacking so I have no idea if the
anode currents in the data sheet are in fact normal for RF
applications. It is quite possible I am barking up
entirely the wrong tree.

You might expect the published curves to cater for the
published typical operating conditions.

Ian, consider the output impedance of your stage, and
make sure your loadline is for actual AC operating
conditions, and not just whatever load resistor you are
using. It may become obvious that you need more
current.

ra is about 10K and the anode resistor is 39K so the
output impedance is less than 10K. It feeds a load of
470K.

OK. I assumed you were in the region where grid lines are
bunched and skewed clockwise, right at the bottom of the
graph, where ra rises considerably.

Sort of. It is hard to tell how skewed the grid lines are
from the data sheet but at present Ia is just under 4mA.

At which point ra will be a bit higher than 10k by the
looks.

Also suspected you may
have been using it as a driver into a lower load.

But, if the valve is a 6AU6, 5mA gets you out of that
region, and your load allows a decent voltage swing


That's the thing. In a preamp, two tube factors contribute
to noise - gm and Ia so it is quite typical to run the
first stage at a very low current - say 1mA or less - and
the curves are unreadable in that region.

But not typical to use a 6AU6. Curves for an ECC83, for
example, are scaled appropriately for your application.
Compare with those for an ECC82, where scale is similar to
what you have for the 6AU6. That's why I assumed you where
talking about the driver end of the pre-amp.

Anode characteristics charts I find are on a scale that
puts 5mA well up the left axis for pentode operation, but
only about one fifth of the way up for triode connection.

Precisely what I an talking about.

Similarly, typical operating points extend down to 5mA
for pentode, but not for triode, where about 12mA is
suggested. So in each case, the graphs are scaled around
the suggested operating points.

All the same, resolution is reasonable for triode 5mA.
Clear enough to confirm your 10k using a ruler on-screen:

http://www.mif.pg.gda.pl/homepages/f...93/6/6AU6A.pdf

That's the best data sheet I have found so far for the
triode curves except perhaps for a Mazda one in French. I
have used a graphics program to blow up the region of
interest and I have printed it out on an A4 sheet. Looks
quite good even at 1mA.


To my eye, the triode looks quite lumpy at low currents.

One thing I wonder, which relates to all of this, is why
a sharp cut-off valve should be typically operated
nowhere near cut-off? I could do with a lesson on what
sharp cut-off valves were for and why.


I am not sure why you would want to operate it near cut
off. AIUI sharp cut off is simply a result of an evenly
wound grid - and that's how all grids were to start with
as it was the obvious and easy way to make them. As the
grid is evenly wound the grid field is uniform so there is
just one -ve voltage at which the grid stops ALL the
electrons reaching the anode. The other factor is that the
closer the turns on the grid the higher the gm. I
understand the first need was for a variable gain stage
for AGC in radio and someone thought of varying the grid
turns spacing to vary gm. As the grid gets more -ve only
the most closely packed grid turns stop electrons reaching
the anode at first, so gm falls, then as it gets more -ve,
more widely spaced turns stop electron flow and so on
until eventually full blown cut off is reached. This is a
remote cut off tube.

OK, that's similar to last time I asked what sharp cut-off
valves are for, and why. Consequently I know what they are
and how they work, but I still don't know what they are for,
or why...although the variable gain application is useful to
know, as I have been wondering if I can do my
expanding-feedback headphone amp with valves, instead of
this chip I got for a wah-wah pedal. Actually, comparing
with common small triodes eg ecc82, I can't see how or where
the sharpness manifests itself in the anode characteristics
for triode operation.

On the face of it, one might think that cut-off
characteristics only matter when you are likely to be
operating close to cut-off, but then perhaps that's when a
remote cut-off valve would be better because clipping would
be softer.

John probably knows all this stuff.

Are you sure you will have enough headroom at such low
current? Are you using global nfb?

cheers, Ian