[flipper]

On Sun, 20 May 2012 04:28:58 -0700 (PDT), Patrick Turner
wrote:

On May 20, 2:05*pm, flipper wrote:
On Wed, 16 May 2012 23:18:38 +0000, TerryG

wrote:

Hi all,

I am new to this forum but have known several members here for quite
some time. *I know not many of you will recognise me from this handle,
but you will I know.

Anyway *I have been enjoying SEP amplifiers this last year an 1/2.
Don't know why they have the reputation they do compared to SET's, but
that is another issue.

Because, all else being equal (which it seldom is), a SE pentode will
have higher output impedance (worse damping factor) and more
distortion than a SET. The saving graces are it takes less input to
drive and produces more output power.

I have designed and built several pentode single ended amplifiers at
this point. *And the one point that still alludes me is: *How from what
the datasheets say can you figure out what resistor to start with on g2?

I understand the DC voltage value and that is not difficult to figure
out, but I like pentodes best when there is an unbypass resistor on g2.

I think you answered your own question. The common purposes of a G2
resistor are to bias, which you note is easy enough to determine from
the existing data, and to kill parasitic oscillations, which are not
'static' characteristics and, so, not something for a static graph.

An unpypassed G2, of the nature you're describing, is not 'typical'
operation.

I have found this does several things, it moves the top bias lines down,
and the bottom bias lines up at the same time raising ( or lowering
depending on how you look at it) the cut off bias voltage. *Usually by
putting a unbypassed g2 resistor on g2 your current drops compared to
bypassing, and a little more wattage is possible for the same operating
point.

G2 current will be less because you are, in effect, applying negative
feedback to the screen with an unbypassed G2 resistor. Plate current
will also be less because you are, in effect, applying negative
feedback to the screen with an unbypassed G2 resistor.

I don't know how you arrive at the 'more wattage' conclusion.

As a side note, the 'opposite polarity' Rg2 derived screen signal was
used for the, so called, "self split" (Class A) PP amplifiers. It
drives the second grid so you don't need 'another tube'.

Negative feedback is also why the SET has lower impedance, lower
distortion, less power and also why it takes more signal to drive it.
Ia of a triode depends on *BOTH* G1 and plate voltages so if you stick
an impedance between plate and B+ the generated signal is negative
feedback to the plate. That is precisely what the 'fixed' screen
voltage in a pentode prevents, plate voltage feedback, hence it's
higher plate impedance, gain, and distortion. Putting an unbypassed
screen resistor on G2 means you're applying 'a little' of the feedback
a triode would see, plus lowering screen voltage.

*It seems most g2's have a certain impedance,

Clearly it has impedance or else screen voltage would be 0 with any
resistance between G2 and B+.

EL84s you can start
with 1.5K and go to above 10K with the resistor have more effect on the
signal traces. *This also lowers distortion it appears,

Negative feedback.

I've always wondered how much good is the "negative feedback" that is
obtained in this very simple manner.

Then I guess it's about time for you to "get away from the PC and go
solder up a circuit and measure it all."

On the other hand, the OP's question was where he could find G2 values
in the datasheets and no amount of bench work will answer that.

snip puffery

but there is
always a point of diminished returns for each tube.

Sure, because the larger Rg2 the lower screen voltage, lower Ia, and
less power.

The 'better' way of doing it is UL. You still get less power but at
least screen voltage isn't also reduced so you only 'sacrifice' that
which goes directly to reducing distortion. (Allegedly there's a
'sweet spot' ratio where 'maximum benefit' of both is achieved).

It sure would be nice to be able to do a few math problem when looking
at datasheets and know where to start with a tube.

The rest is usually contained in the 'typical' operating
characteristics or example circuits.

I will introduce a controversial subject with g2 later, but for now I
just want to know what information on the datasheet will tell me what I
want to know.

I imagine that, with enough perseverance, one could derive screen
current magnitude, the resulting Rg2 voltage, and the degree of
feedback but it's not 'straight on the graphs' because they don't
expect them to be operated that way.

Terry

Something that might be interesting to try would be a sort of 'poor
man's UL' with a tube like the 6CW5 that can use 250V B+ but is screen
limited to 200V. There a 'significant size' unbypassed Rg2 of 10k
would put screen right at the spec'd 200 V.

Ah, the poor man's UL! aka Free Lunch UL.

No one said a blooming thing about a 'free lunch' and a "poor man's
insert anything" always refers to an inferior 'cheap' substitute for
the otherwise preferred solution.

It doesn't work too well
because screen input resistance is fairly low, and where you have RG2
high, then VG2 becomes high, but screen FB applied this way is a non-
linear way of applying FB.

That's the second time you made an arbitrary declaration of
"non-linear," after 'always wondering' about it, and with no
explanation of what the heck that means. Non-linear how? Like Ia is
non-linear?

I don't have any data one way or the other but it was apparently
considered 'linear enough', whatever that criteria might have been,
for self split PP amps. And while that's not entirely equivalent it's
a heck of a lot more substantive than you declaring it 'sort of works
like' Rk and then never again speaking of screen feedback in the
previous rambling.

The conventional UL OPT supplies the screen
input power, ie, Vg2 x Ig2 input from a relatively low source
resistance.

If you are externally *driving* G2 then you need a low source
impedance but a large value Rg2 is not 'driving' G2, the voltage
produced is the *result* of G2 current. G2 does the 'driving'.

That's akin to bitching that Rl on a triode is 'too high an impedance'
to 'drive the plate'.

snip


Patrick Turner.