"robert casey" wrote in message

"critical" in that I have to create a tuned circuit, or
just that more R and more C is better, and that I need
to select R to get a reasonable screen voltage, and then
C to get better effective ripple filtering?

There is a conflict of requirements. Your arrangement hopes to
improve on the option of simply adding extra LC filtering for the
whole stage. Instead, you use an extra LC filter on the screen
alone. Because the screen draws much less current, you can use a
larger value resistor, and therefore a smaller value capacitor, for
the filter.

Taking this to extremes, you could reduce the ripple at the screen
at 100Hz hum frequency to 100th with a 1 meg resistor and a cap of
150nF. But for a screen current of say 1mA, this arrangement would
drop 1kV. At the other extreme, 10k would drop just 10V, but you
would need a cap of 15uF for the same attenuation.

The triode will behave as if its anode were at screen potential, so
if you are matching AC characteristics for cancellation of
harmonics, you should look at the triode characteristics at the
screen voltage you end up with.

All of the above could be solved by using a current source for the
screen, as I see Patrick suggests. But then there other
matters...(and equally you could use a voltage regulator for the
whole supply if your going to involve SS).

Normally in triode mode you can use a lowish value screen resistor,
or none at all. The non-linear effects of screen current are then
insignificant or non-existent.

One common reason for regulating screen supplies to pentodes is to
minimise the effect of screen current. A low impedance supply
reduces the voltage distortion caused by the current. Your choice of
cap value and resistor should take this into account.

Also, the mechanism for producing the even harmonic distortion that
you consider desirable in the top valve, also produces a DC offset
that will not be passed however big the cap is. This offset depends
on the amount of 2H signal current drawn by the screen multiplied by
the value of the screen resistor. Onset and recovery time depends on
the cap size. Maybe none of this is significant because any sensible
values of cap and resistor will make it vanishingly small?

Many amps using pentodes for voltage amplification don't seem to
care much about these things. I suspect they take the screen
impedance at
the operating point and divide it by10 to get the required
resistance of the screen supply, and then choose a decoupling cap to
suit, with cost in mind.

The screen impedance used for calculation should be the minimum it
reaches with the amp at full power. That could be considerably lower
than at the operating point, so you need a lower impedance supply,
and a correspondingly bigger decoupling cap.

All this is equally true of your grounded-screen triode-mode
pentode. Also, because it is on the top of a totem-pole, you have
the high voltage to consider.

Assuming you don't want electrolytics in series, you will be using a
polyprop cap, probably 600V. It could be cheaper to regulate the
voltage to the anode, and use the normal triode connection.

cheers, Ian