Well, what I might do is set the pin high or low and see what the I vs V 
curve is at different temps.  That static data ignores transient effects 
like PIC pin internal capacitances but I think the dominant factor here 
is just this profile of how much current it's able to drive into the 
capacitance as voltage rises.

However I'm not sure what the best way is to translate that into a SPICE 
model.  It's easy to find a SPICE model of most transistors.  If we know 
that during a high-to-low transition when Vout=4v Iout=20ma sink, 
Vout=3v Iout=15ma sink, Vout=2v Iout=7ma sink, Vout=1v Iout=5ma sink 
then SPICE will be able to calculate switching time and loss for that 
transistor.  I'm sure there's a piecewise model in SPICE that'll do 
voltage vs current like that.

Indeed, temp is a good question since it has a strong effect on drive.  
Now with a MOSFET, when it's cold the threshold is and gain is low but 
if it's experiencing high switching losses as a result it sure won't 
stay cold for long, making the situation self-correcting.  However a 
driving PIC pin would generally not heat itself up substantially even as 
the external transistor overheats due to overly long transition times.  
This still lacks information on lot variations though, and what 
Microchip actually guarantees for performance.  Ideally we want to have 
info on various conditions and lots and simulate if the system will 
still meet minimal performance in the worst case scenario.

I wouldn't want to measure individual FETs.  Exact measurements are very 
hard to come by and kinda pointless since I would need to be able to 
predict performance beforehand when looking though say 30x potential 
candidates at Mouser.

I searched through the Microchip site, they have SPICE models of their 
op-amps but not PICs.  I asked around on the Microchip forums and nobody 
had a lead on a SPICE model.

Danny

Sean Breheny wrote:
> Hi Danny,
>
> Seems like you have a very good handle on what affects the PIC pin
> switching speed. Have you considered doing some tests? For example,
> you could write some code which just toggled the pin (being careful
> not to run into read-modify-write problems), put a worst-case
> capacitive load on the pin, and then scope the voltage on the pin to
> measure the rise time. You could do this to several pins on the PIC
> and also to several different PICs. You could also vary the
> temperature using a hair-dryer and possibly some freeze-spray to see
> how much temperature affects it. This would give you a nice little
> dataset which may be adequate for your needs.
>
> If you use the actual FET instead of a fixed capacitor on the pin,
> just watch out that the gate charge will depend on what the drain is
> attached to (which you seem to realize already). You would probably
> have to actually put the full current through the external FET to be
> sure it was presenting a realistic load to the PIC pin.
>
> Microchip may or may not have the data you are interested in. Have you
> tried contacting your local rep or microchip headquarters?
>
> Sean
>   

-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist