RE: [LTspice] Screen-grid (G2) current problem at ECF82 penthode (Koren_Tubes.cir)

Hi Cezar

Norman Koren’s models do NOT accurately model G2 current.

Norman Koren’s model improves the Plate (Anode) current modelling of pentodes over Scott Reynolds. But like Reynolds does not attempt to accurately model screen current except at one arbitrary point.

The answer: An approach that borrows from W. Marshall Leach Jr. Fortunately Ispacescharge is approximately equal to a constant

Ispacecharge = Iplate + Iscreen; 

Iscreen = Ispacecharge - Iplate;

Curiously both Scott Reynalds and Norman Koren almost get there as their Iscreen formula. But somehow manage not to get there. Fortunately the Iscreen formula can be adapted to give space charge.

Why haven’t I done it? Simple because some data sheet information is very patchy. I would have to get real samples and do new original data.


That said No model is entirely accurate under all conditions. Especially when the Mu Gm and Rplate varies with Vcontrol and Vscreen. Another thing that has to be said is that the published data is not always as accurate as you might think. They are stylised. A prime example is diodes that claim that no current flows when no voltage when NO voltage is present.

It should also be said that Koren's models work better for pentodes than Beam Tetrodes with secondary emission.

As far as I know no models for valves (Tubes) attempts to model control grid space charge accurately. They also assume a simple diode resistor characteristic for positive control grid voltages. When in Triodes the actual current depends on control grid voltage, plate voltage and amplification factor.


Best regards

Suusi Malcolm-Brown