[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: gEDA-user: Split ground planes and zero ohm jumpers



On 4/8/2011 12:57 PM, Mark Rages wrote:
On Fri, Apr 8, 2011 at 11:39 AM, rickman<gnuarm.geda@xxxxxxxxx>  wrote:
On 4/7/2011 1:13 PM, Stephan Boettcher wrote:
rickman<gnuarm.geda@xxxxxxxxx>    writes:

I have to say I am philosophically opposed to any feature that allows
a design to pass DRC when the layout differs from the schematic.
Just to get the terminology right:

DRC has no business to care about the schematics at all.  There shall be
a tool to check if the layout implements the schematics netlist, but
that is a different issue.

PCB implements this distiction properly.  DRC checks consider coper
structures as layed out when evaluating the rules, without regard to the
netlist.

The Rat's-nest (O-key) ignores DRC rules when checking connectivity.
  Ok, if you want to be pedantic the net list is not the schematic, but if
the netlist differs from the schematic, then you have another problem.

DRC is a part of my design process which includes a verification that the
layout matches the net list.  In fact, my number 1 "design rule"  to be
checked is that  they match.  What button I push to get the tool to do my
required design rule checking is irrelevant.  It is just a tool and does not
define my process.

So my point is that adding an attribute to any copper to tell the tool to
ignore the connectivity violates my idea of design rule checking.

Rick
Rick, your idea about DRC and the pureness of connectivity is wrong.
Connectivity is insufficient for real world designs, which is the
point of all this discussion.  It's possible to have a design with
perfect connectivity that doesn't work.   Any analog or RF design, and
a lot of high-speed digital designs, require a lot more attention than
getting connectivity right.

So the desire to tag certain nets / connections as controlled
impedance, no DRC, equal trace length, etc.

Regards,
Mark
markrages@gmail

I don't know how you get from my statements to suggesting that I don't agree that tags should be used. Controlled impedance and equal trace length have nothing to do with what I said. If you have some portion of your design that you need to exclude from DRC, I would suggest that you need to adjust your design rules. It may be that the tool is not sophisticated enough to accommodate the needed design rules, but saying that it is desirable to exclude portions of a design from design rule checking is a very poor way of dealing with the limitations of a tool. I would much prefer to have the design rule violations reported and then allow the designer to sign off that these violations are not a problem. It can be too easy to make a mistake when excluding a design rule check and missing a violation that results in a problem.

Too often people confuse the functionality of a tool with the process of verifying correctness of a design. I done't want a tool dictate my design rule process. If the tool has limitations, I prefer to have it report problems that I can then manually verify are not problems than to run the risk of missing problems that cost time and money.

The problem of connectivity of nets can be handled without violating design rules. I do this using a part footprint that connects the two nets. This shows up on the schematic as the single connection point between the two nets, is reflected in the netlist as the only point of connection of the two distinct nets and can be verified in design rule checking that the nets are connected no where other than by this footprint.

Rick


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