Modelling JFETs

[ripson]

Modeling JFETs

Hi guys, is there any piece of software/code/equations that helps to generate JFETs spice models for specific transistors?


I need to create/modify models for popular Toshiba JFETs 2sk170/2sj74 (matched at specific Idss) and 2sk246/2sj103.

UPDATE: I have created a simple model generator, template where I can specify Idss (2.6-20mA) and the progam changes beta and vto based on Idss. I have found and used some 2sk170 models specified as gr, bl and v, measured their Idss and interpolated the beta and Vto points (parabola equation for three points). It's not very scientific but roughly works:



>>>>>>>>>>>>>> link


It would be great to know how for particular JFET, Vto and Beta changes as a function of desired Idss. Also if necessary to amend other params.

 

[ripson]

According to this:

there is a linear dependency between Idss and Vto.


Vto for (2sk170) = -0.0389 * idss - 0.261
Idss in mA Vto in V

now what is the relation between beta (transconductance coefficient) and Idss... an we are home..

 

[Mark Johnson]

Since the graph is plotted using a log scale, I think it might be more correct to claim/assert

there is a linear dependency between log₁₀(Idss) and Vto.

 

[ripson]

Well, good point 🙂
So for these points from the plot:

P1 = (1, -0.3)
P2 = (10,-0.65)

..if I am not mistaken:

 function logBase(x, base) {
     return Math.log(x) / Math.log(base);
 }
Vto(Idss) = logBase(Idss / 0.138949549437, 0.00138949549437)

but I am not sure if Vgs(off) is really what LTSpice interprets as Vto

 

[ripson]

but here both axis are in log scale, so log₁₀y=a*log₁₀x+b

 

[ripson]

almost: log10y=a*log10x+log10b

Log–log plot - Wikipedia


Thank you very much for pointing that out.

Having two points in log-log plot that make a straight line we can calculate any Y for given X by:

function getVto(idss, p1, p2) {
   var m = Math.log10(p2.y / p1.y) / Math.log10(p2.x / p1.x);
   var k = p1.y / Math.pow(p1.x, m);
   return  k * Math.pow(idss, m);
}

but Vgs(off) from the 2sk170 spec is rather too big for Vto. I would need to probably adjust these values.

Still don't know how to calculate beta and how it is related to gm.