SPICE models for power MOSFETs

It would bear noting that the model dependent on Vgd is essentially bogus. I've never seen a MOSFET like that. VDMOS is dependent on Vds for both Cgd and Cds. Seems unintuitive, but that's how it is!

Here's an implementation from one of my models,

• Diode junction capacitance model, dependent on 3rd terminal.

• Use for G, D, S as MOSFET Crss. .SUBCKT C_CJO2 1 2 4 PARAMS: CJO=10p M=0.5 VJ=0.75 E1 3 2 1 2 1.0 C1 3 2 {CJO} BI 2 1 I={I(E1) * EXP(LIMIT(LN((ABS(V(2,4)) / VJ) + 1) * M * -1, -40, 40))} .ENDS

Note that it isn't limited on M as some simulators are (why?!). Values from

3 to 20 (more than 20 gives diminishing returns) fit SJ devices pretty well.

I also haven't seen a device that shows nearly as pronounced of a kink as the datasheet suggests. Here's a real measurement:

I have no idea how or why everyone shows a kink. It's not actually there.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Design 
Website: https://www.seventransistorlabs.com/ 

"Winfield Hill"  wrote in message  
news:q7mdl3022ad@drn.newsguy.com... 
> Design of circuits using power MOSFETs is an 
> important part of what we do.  It's a favorite 
> topic, and we talked about it plenty in AoE 3rd. 
> Our upcoming x-Chapter book, that's filled with 
> advanced material, has about 130 pages on the 
> subject.  See Dropbox for a mini-section draft 
> about MOSFET SPICE models and capacitance.  It 
> goes after a lengthy section discussing specific 
> MOSFET part aspects, and implications. You could 
> make a printout to better follow the details. 
> 
> https://www.dropbox.com/sh/9nhm5fxm8zwsei3/AAB0lYLZmwXkOl-FeG8fuDfea?dl=0 
> 
> Let me know what you think. 
> 
> 
> --  
> Thanks, 
>    - Win

In footnote 3, I think you want to use "prescribed" rather than "proscribed".

Yes. I don't even know how to measure Vgd with Vgs values other than zero. Once you would start the MOSFET conducting, C-V meters can't deal with it. Maybe with a linear source-follower circuit, but that'll have a modest Vgs, certainly not variable.

--
 Thanks, 
    - Win

You can tell the difference at low Vds, particularly in switching operation where Vgs is rising and Vds is falling (or vice versa).

Right, C meters are useless for many checks, it has to be measured dynamically.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Design 
Website: https://www.seventransistorlabs.com/