Re: gEDA-user: Gnetlist -g PCB

[John Doty <jpd@xxxxxxxxxxxxx>]

On Jun 29, 2008, at 8:38 PM, al davis wrote:

> On Sunday 29 June 2008, Kai-Martin Knaak wrote:
>> A check for pin type is even more worthless for analog
>> applications.   There are perfectly reasonable and common
>> opamp circuits that connect the pin 3 (input) directly to pin
>> 6 (output).
>
> Actually, pin type is important for modern analog design, and is
> used by modern analog simulators.
>
> Using Verilog terminology, there is "direction"
> and "discipline".
>
> The "direction" can be one of "input", "output", or "inout".  I
> think everyone here so far has forgotten about "inout".  Spice
> supports only "inout".  If you think this is unimportant, you
> are assuming everything is "inout".  All pins on passives like
> resistors are "inout".
>
> The "discipline" is something like "electrical".  Spice supports
> only "electrical".  In a more advanced sense there can be high
> voltage, low noise, high current, and other variants.  There
> can also be voltage only and current only disciplines.
> Non-electrical systems have a whole bunch of others.
>
> As far as matching and design rules...  Obviously it is ok to
> connect an "output" to an "input".  You can't tie an "output"
> to another "output".

You're assuming voltage interfaces. Current interfaces turn that on  
its head. And then there's the trick of grounding an opamp "output"  
and getting output current from the power pins: this illustrates that  
the direction and discipline depend on the application, not on the part.

>
> If everything on a net is "input" it
> isn't very useful, so that too would be a violation.
>
> To determine the legality of connecting disciplines together,
> you need to set the rules somehow.  In Verilog-AMS,
> a "connectrules" block defines what you can do and how it is
> resolved.

I read the book you recommended on Verilog-AMS. It misses the  
critical idea that mixed signal design is applied physics, not  
computer programming. A computer can usually evaluate a function that  
you can logically describe, but circuitry is more restricted,  
especially if speed, noise, and power are significant considerations.  
The top-down design methodology the authors advocate is almost  
guaranteed to demand parts made of unobtainium if applied to  
challenging requirements.

>
> It looks to me that geda has mixed the concepts of discipline
> and direction.

Yes, but the real problem is the mixing of such clerical concepts  
into what is really a set of applied physics issues.

>   It probably has not been a problem yet but
> could become an issue in the future.  It certainly limits scope
> of designs it is useful for.
>
>
>
> _______________________________________________
> geda-user mailing list
> geda-user@xxxxxxxxxxxxxx
> http://www.seul.org/cgi-bin/mailman/listinfo/geda-user

John Doty              Noqsi Aerospace, Ltd.
http://www.noqsi.com/
jpd@xxxxxxxxx




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