Re: gEDA-user: KJWaves - new release (a r)

[Dan McMahill <dan@xxxxxxxxxxxx>]

al davis wrote:
> On Monday 07 January 2008, a r wrote:
>> Say I have a MOS transistor instance on schematic. When I
>> open the properties dialog of this instance, I expect to see
>> a drop-list of available model types/levels, along with a
>> list of text fields where I can define parameters that are
>> relevant to the selected model type/level. All parameters
>> should accept expressions so that I could automate editing
>> and verification of the design a bit.
> 
> That's really an unreasonable request.  
> 
> Consider this variant:
> 
>> Say I have a klystron instance on schematic. When I
>> open the properties dialog of this instance, I expect to see
>> a drop-list of available model types/levels, along with a
>> list of text fields where I can define parameters that are
>> relevant to the selected model type/level. All parameters
>> should accept expressions so that I could automate editing
>> and verification of the design a bit.
> 
> Some readers might ask: "what's a klystron".  That's the point.  
> The symbol library cannot know about all variants of all 
> symbols, all of the parameters, etc.

I agree completely with Al that the full list here is not even 
desirable.  For example

bsim3v3 is a very popular model formulation for MOS transistors.  A 
model set has almost 300 parameters (things like mobility and oxide 
thickness).  In addition, there are another 20+ instance parameters 
(things like width and length).

Now consider that in different simulators that support bsim3v3, that 
these 300 model parameters + 20 instance parameters do not all exist and 
don't always have the same name.

Now consider that bsim4 has closer to 360 model parameters and 30 
instance parameters.

EKV is pure simplicity, only 100 some model parameters.

You get the idea.  You'd have to tell the system about a thousand or two 
  model parameters just for MOS devices alone and it surely would only 
work with one particular simulator.

Now it wouldn't be that unreasonable to have a drop down that looked in 
a database and picked up all the right settings (which probably don't 
mean model parameters but rather a footprint and what model to call) for 
a known list of discrete MOS transistors.  I'd argue that there is not 
that much difference between that and a heavy library though.

-Dan




_______________________________________________
geda-user mailing list
geda-user@xxxxxxxxxxxxxx
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user