Motivated by Pierre Touzelet's 2007 AX article for deriving Triode params with Microsoft Excel, I made a video showing how to use "Graph Grabber" and "Solver" to create basic, level 1 spice params for MOSFETs:
https://www.youtube.com/watch?v=2vh1-RbBJa0
The equations in the columns:
Id(sat) = (KP/2) * ((Vgs-vt)^2)*(1+Lambda*Vds)
where KP is mu * Cox * (W/L)
I figger "saturation" to be 90% of the current yielded at the maximum Vds for each set of Vgs curves -- and test for this. Unsaturated data is thus nulled so that no error term is calculated.
Pierre minimized "sum of the squared errors" for triodes. In the case of MOSFETs the saturation currents differ by orders of magnitude so "goodness of fit", "maximizing R-squareds" were calculated. Rather than simply letting "Solver" do its thing, you might want to manually iterate the approximations to derive params which best suit your application.
https://www.youtube.com/watch?v=2vh1-RbBJa0
The equations in the columns:
Id(sat) = (KP/2) * ((Vgs-vt)^2)*(1+Lambda*Vds)
where KP is mu * Cox * (W/L)
I figger "saturation" to be 90% of the current yielded at the maximum Vds for each set of Vgs curves -- and test for this. Unsaturated data is thus nulled so that no error term is calculated.
Pierre minimized "sum of the squared errors" for triodes. In the case of MOSFETs the saturation currents differ by orders of magnitude so "goodness of fit", "maximizing R-squareds" were calculated. Rather than simply letting "Solver" do its thing, you might want to manually iterate the approximations to derive params which best suit your application.
For the linear part of the curve:
Id=KP*((2(Vgs-Vto)*Vds)-Vds^2)*(1+ Lambda*Vds)
Good at first crack, but Vgs really isn't Vgs as RS gets in the way. What I've done is create an effective Vgs by multiplying the observed drain current by an approximated RS. Then let "Solver" find Vto, kappa, lambda and RS.
Now, I'm wondering if it will work!
Id=KP*((2(Vgs-Vto)*Vds)-Vds^2)*(1+ Lambda*Vds)
Good at first crack, but Vgs really isn't Vgs as RS gets in the way. What I've done is create an effective Vgs by multiplying the observed drain current by an approximated RS. Then let "Solver" find Vto, kappa, lambda and RS.
Now, I'm wondering if it will work!
First term should be: (Vgs-Vto)*Vds-(Vds^2)/2
FWIW, the datasheets and IRF's spice models don't seem to agree except for the first couple of Vgs series above Vto
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.