Need help determining MU and KG1

[platplaas]

I have taken a liking to Falstad's java applet for modelling circuits.

This is a simple but remarkably powerful little app. I have found it really useful (especially as a newbie) to help me understand how circuits work. For the time being it is everything I need, though I might later try some spice modelling.

My challenge is this: To model triodes in the applet, I need to input the MU and KG1 parameters of the tube. In my particular case it is for a 6w6gt / 6dg6gt tube.

I have tried to determine the parameters from the tube datasheet by using Curve Captor, but it outputs the information as a spice model and doesn't specifically identify the tube's MU and KG1 factors.

* 6w6 curve captor macro model
.subckt 6w6 curve captor P G K
koren8(0.1900542192,-0.2273964238,1.280710317,11.85586446,-17.62800097,25.56566888,3.84752409,1.378232258)
.ends 6w6 curve captor

* 6w6 curve captor Spice 3F4 model
.subckt 6w6 curve captor P G K
Bp P K I=(0.1900542192m)*uramp(V(P,K)*ln(1.0+(-0.2273964238)+exp((1.280710317)+(1.280710317)*((11 .85586446)+(-17.62800097m)*V(G,K))*V(G,K)/sqrt((25.56566888)^2+(V(P,K)-(3.84752409))^2)))/(1.280710317))^(1.378232258)
.ends 6w6 curve captor

I have also tried tracing tube parameters using Model Paint Tools. Seems simple to use - but I've spent countless hours trying to match the curves to the datasheet. I came up with three different parameter configurations that match the curves fairly closely, but the configurations have substantially different MU and KG1 values.

Is there an easy way to find MU and KG1 parameters? Or possibly some way to derive these from the spice model output from Curve Captor?

Thanks.

[ywara]

I'm going to guess that the Falstad simulator uses a two-parameter Child-Langmuir-based model for plate current. You're not going to get that from an 8-parameter modified Koren model.

You need to fit your curves with the basic 2-parameter model, which should be

Ip = KG1^-1 * (Vgk + Vpk/MU)^1.5 for (Vgk + Vpk/MU) > 0, else Ip = 0

(from Koren's site)

Curve Captor outputs a slightly different form (same model) to its SPICE model like so

Ip = (KG1^-1 * MU^-1.5) * (MU * Vgk + Vpk)^1.5 for (MU * Vgk + Vpk) > 0, else Ip = 0

if you just want to translate from Curve Captor's coefficients to KG1 and MU for the simulator.

I get far better results doing a surface fit after eyeballing the curves into a table for MATLAB than using Curve Captor, however. Either the point entry method in Curve Captor is too finicky for me or the optimization algorithm is lousy. Probably a little of both.