Re: [Help-gnucap] how to use LMC6482A spice model with gnucap?

[Al Davis]

On Sunday 12 February 2006 22:12, John Steele Scott wrote:
> I have made a small netlist to simulate an active rectifier
> which uses National Semiconductor's LMC6482A. The model is
> available from
> <http://www.national.com/models/spice/LM/LMC6482A.MOD>.
>
> When I try to simulate my circuit, I get the following
> errors:
>
> *////////////////////////////////////////////////////////////
>////////// G4  98 16 POLY(2) 1 49 2 49 0 2.812E-8 2.812E-8
>                 ^ ? need 1 more nodes

POLY(2) is not supported.
Give me a couple of days.  I will come up with a work-around.

> .MODEL MOSFET PMOS(VTO=0 KP=1.842E-3)
>                    ^ ? no level specified, using 1
Just a warning.  Spice would give you the same warning.  You can 
ignore it, but it is better to add the parameter "level = 1" to 
the model statement.


> M2.X_U2, MOSFET
> model and device parameters are incompatible

You didn't specify a length and width.  It should substitute 
from the option DEFW and DEFL.

You found a bug, which happened when the "binning" feature was 
added.  The binning check happens before the default values are 
filled in.  Now that I know about it I will fix it.  The work 
around is to specify length and width.  The value for both 
should be 100u.

What they did is bad practice, especially with a packaged model.  
If you change the defaults, through the options command 
or .options statement, the model will be changed in a way that 
its creator did not anticipate.  Being able to change the 
defaults through the options statement is a dumb idea, but I 
included it for Spice compatibility.

> In case it matters, the netlist which I am using this circuit
> in is below. This is based on an article at
> <http://www.edn.com/article/CA6250014.html> "Precision
> full-wave signal rectifier needs no diodes".

It uses the op-amp's clipping to produce rectification.  I am 
not sure how "precise" that is.