Re: gEDA-user: Schematic Level DRC DIscussion

[Steve Meier <smeier@xxxxxxxxxxxxxxxxxxx>]

On Tue, 2007-11-06 at 14:31 -0700, John Doty wrote:
> On Nov 6, 2007, at 1:04 PM, Steve Meier wrote:
> 
> > U1
> >
> > pin  1   "D1"  pas
> > pin  2   "NC"  pas
> > pin  3   "D3"  pas
> > pin  4   "S3"  pas
> > pin  5   "S4"  pas
> > pin  6   "D4"  pas
> > pin  7   "NC"  pas
> > pin  8   "D2"  pas
> > pin  9   "S2"  pas
> > pin 10   "IN2"  in
> > pin 11   "V+"  pwr
> > pin 12   "VL"  pwr
> > pin 13   "GND" pwr
> > pin 14   "V-"  pwr
> > pin 15   "IN1"  in
> > pin 16   "S1"  pas
> >
> >
> > J1  pins are   pas
> 
> Yes. But does that really make sense?
> 
> >
> > nets:
> >
> > U1.1 U1.3 U1.4 U1.5 U1.6 U1.8 J1.A4
> > U1.9 U1.10 U1.11 U1.12 U1.13 U1.14 U1.15 J1.B3
> > U1.16 J1.A4
> >
> >
> > First:  J1.A4 is used twice in effect reducing the nets to two nets
> 
> Yes. Is that a real error? I think it is right now, but should  
> gnetlist treat these as separate nets? It doesn't for a ground  
> symbol, for example.
> 

I think, that there should at least be a warning. Otherwise you could
have unintentionally shorted nets.


> >
> > U1.1 U1.3 U1.4 U1.5 U1.6 U1.8 U1.16 J1.A4           All pins pas
> > U1.9 U1.10 U1.11 U1.12 U1.13 U1.14 U1.15 J1.B3      pins include types
> > such as pas, in and pwr
> >
> >
> > For the second net arcording to the drc2 matrix
> >
> > 1) pas may connect to pas, in or pwr
> > 2) in may connect to in or pwr
> >
> > There for DRC2 thinks that there are no connection errors.
> 
> Yes. Those rules are almost useless in mixed signal circuits.
> 
> >
> >> From my reading of DRC2 it tests for
> >
> > 1) Non numbered parts
> > 2) Duplicate references
> 
> But, of course, some of us often use multiple symbols for one part.
> 

I do as well, I have added to my code an attribute that tells the
netlister to not include this sybol in the bom. In essence, I am saying
that this symbol is part of a collection of symbols and that this
particular symbol isn't the one that provides the bom info. So it could
also be a clue to a drc that it shouldn't be judged to be a duplicate
part.


> > 3) One connection nets
> 
> That's somewhat useful. But it also complains of unconnected pins,  
> which are normal, not usually errors.
> 
> > 4) net pin types
> 
> Using a classification nearly irrelevant to anything except pure  
> digital design.
> 
> > 5) slots
> > 6) duplicated slots
> > 7) un-used slots
> >
> > I would expect complaints about duplicate slots and un-used slots.
> 
> Unused connector pins are extremely common in error-free designs. And  
> I think it adds clarity to treat connectors as multi-slot rather than  
> having a lot of named lines converging on a big block.

Mater of personal preference that I woun't argue about. But then we
should have a way of surpressing drc warnings or errors for this type of
slotted device. no_unused_slot_drc=true # opt out not opt in

> 
> > Not
> > about the way the pins are hooked together. One obvious weakness is  
> > that
> > the test schematic connects, V+, V-, VL and GND together. These are  
> > all
> > type pwr and thus according to the matrix they may be connected. I  
> > think
> > this is a weakness of pins not being heavier.
> 
> Yes, but how to fix. Making the pins heavier will make symbol  
> construction even harder for new users. Pin classification is already  
> confusing.
> 

I am in favior of heavier pins. I think that devices should have a way
of saying. Hey a Vcc power pin for this device should only be connected
to nets that have 3.3V. This would have possibly prevented one flipped
power level for a subcircuit on a recent board. Pop goes the
tantalims ;)

One reason pin classification is so hard is that there is no attempt to
limit attributes to any type of object and that the value of an
attribute can be anything.

A smarter attribute system that says net segments can have this  list of
attributes and each of these attributes gets a list of legal values.
Then the selction method would restrict users to reasonable values.

> >
> > Steve Meier
> >
> > On Tue, 2007-11-06 at 09:36 -0700, John Doty wrote:
> >> On Nov 6, 2007, at 7:27 AM, Steve Meier wrote:
> >>
> >>> What do we expect the schematic DRC to catch?
> >>>
> >>> 1) Detect duplicate use of reference designators, but don't
> >>> complain if
> >>> the usage is for different slots.
> >>> 2) For a multi-symbol device check to make sure that the pins arn't
> >>> used
> >>> multiple times?
> >>
> >> Is that an error?
> >>
> >>>
> >>> But what else and what are the implications for heavy/light symbols?
> >>
> >> Here's a simple schematic, with a silly useless circuit, many
> >> problems. It provokes *numerous* drc2 errors and warnings, but only
> >> two of them are real mistakes. And arguably, one should be able to
> >> duplicate connector pins for clarity within the schematic, although
> >> right now such usage doesn't netlist correctly. But drc2 misses all
> >> of the intentional errors.
> >>
> >>>
> >>> Steve Meier
> >>>
> >>>
> >>> _______________________________________________
> >>> 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
> >
> >
> >
> > _______________________________________________
> > 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



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