Good morning
I wanted to create a thread to share my experience with Model Generator and compare it with others.
I’ve used MOSFET generator to create a model
Preliminaries:
First thing I’ve noticed was that ctrl+v (paste) shortcut doesn’t work here, however mouse leftclick+paste does. Stangely ctrl+x (cut) does work.
Second thing - I think it would be more intuitive to put some kind of border (a thick line for example) between section that is purely informational and one that has actual impact on model behaviour. I know this information is visible in Help Docs, but still.
The digitize feature:
The digitizer feature is very usefull and easy to use (once accustomed). Remember to configure the plot first (set limits, logarythmic axes) BEFORE selecting any particular points from the datasheet plot. Already selected points won’t recalculate automatically if you change plot settings.
.
The thing I feel like is missing here is more direct marking of the points (placing a marker - just like in simulation waveform plot - picture attached)
and ability to manually enter the values instead of using digitizer (or to currect values acquired via digitizer).
Also I was getting some kind of redering errors throughout the process (which would appear randomly) and could be fixed by switching between tabs (picture attached - don’t mind the red annotations)
Hi Farel, thanks for your detailed notes. We’ve not formally launched the Model Generator yet, we’re still working on materials for it, we’ll go over what you’ve provided. Best, Tim
Is the model generator also accounting for Coss, Crss, Ciss?
Upon reading the model generator, there is no any way to insert these capacitance models.
Is it enough to infer these capacitance only from gate charge plot?
from my understanding, model parameters Cgs, Cgdmin, Cgdmax (and Cjo in its body diode) are fundamental elements of Coss, Crss and Ciss. Gate Charge characteristic actually the behavior correlates to that.
Thanks, Kelvin. Here are some notes Mike made on this:
The charge model is that of a power MOSFET: fixed Cgs, a highly nonlinear Cgd, and an output capacitance that of the body diode. While ideal for power MOSFETs, it is usually good enough for design purposes since it will accurately tell you whether or not you can turn that FET on or not. It’s a charge model that I invented and has been copied in other SPICE programs since most SPICE programs can’t simulate a power MOSFET.
Coss, Crss, and Ciss are basically capacitances of the channel. While QSPICE supports both Meyer and Yang-Chatterjee charge models for the channel of a VDMOS transistor, those are left zero by the Model Generation tool because almost all discrete MOSFETs are vertical and the charge model above is either better or at least sufficient. The exception is the concern for modeling soft switching. I might extend the device equations to model the output capacitance that changes more abruptly than a simple body diode junction capacitance.
