DESIGNERCOUNCIL Archives

July 2000

DesignerCouncil@IPC.ORG

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
"Olson, Jack" <[log in to unmask]>
Reply To:
DesignerCouncil E-Mail Forum.
Date:
Tue, 18 Jul 2000 12:03:49 -0500
Content-Type:
text/plain
Parts/Attachments:
text/plain (58 lines)
I would like to thank all of you who took the time to answer, I really appreciate all the responses. I was intrigued by how many different "philosophies" you all are using to make your libraries! I would have been happy to summarize the answers if it would have helped anyone else, but there is really no consensus.

What we decided to do was created the silkscreen at NOMINAL, using a standard library (we chose SMTPLUS), figuring that there was at least some security in the "safety in numbers" theory. If hundreds of companies are using the same footprint dimensions....

....but that doesn't solve the placement issue. Fortunately for us, in addition to the silkscreen layer, our software (Veribest98/VB2000) has an additional assembly layer and component outline layer. The component outline will be created at MAXIMUM component dimensions for error checking (this outline even stores the maximum height if desired). VB can make sure all the component outlines are at least 30mils apart, or something like that.

The last remaining problem is that most DFM specs list different spacings for different component types, like maybe a BGA needs 100mil clearance, where discretes only need 30, or IC to discrete is 40, all different. I can't think of any way to make a DRC checker flexible enough to check all the variations, unless we just add a "fudge factor" to the component outline to make the rule apply correctly (like adding 70mils to the BGA to make sure when we use a 30mil rule, we still get our 100mil preferred clearance around BGAs) Kind of tedious if the rules change, eh?
And in the case where components are made to be butted up together I guess we could shrink the outline by 15 all around so we don't get false errors with our generic 30mil rule, but that seems kinda lame.

If anyone has a more elegant solution I would love to hear it.
Thanks again for all your help,

Jack (the new guy)       <grin>


                -----Original Message-----
                From:   Abd ul-Rahman Lomax [mailto:[log in to unmask]]
                Sent:   Thursday, July 13, 2000 4:07 PM
                To:     [log in to unmask]
                Subject:        Re: [DC] CAD library philosophy?

                At 04:40 PM 7/13/00 -0500, Olson, Jack wrote:

                >I seem to have gotten myself into a big raging discussion here about
                >whether library footprints should be created (and placed) using the
                >nominal dimensions or the maximum dimensions. [...]

                >Now some are saying we should be using the MAXIMUM size, which I don't
                >think buys too much and creates other headaches (different vendors list
                >different tolerances, etc) so I would like to know how YOUR library is created.
                >NOMINAL or MAXIMUM?

                I've usually used nominal. It is simply easier.

                There is something to be said for using MMC, though it is rather
                complicated to actually realize this. To draw true MMC outline, not only
                will there be the dimensional tolerances for the component dimensions to
                consider, which, as noted, may vary from manufacturer to manufacturer, but
                also placement tolerances as well, which may vary with the assembler.

                So, in my opinion, the most practical idea is to draw the parts at nominal,
                and allow sufficient clearance in the design rules to take care of
                component and placement tolerances.

                However, for visual identification of potential problems, it might make
                sense to draw two outlines: one at nominal and one at MMC. If I were to do
                this, I would put the nominal on the silkscreen and the MMC on a special
                assembly layer, used just to check clearances.

                I'll still stay with nominal alone and appropriate design rules. It's much
                simpler.


                [log in to unmask]
                Abdulrahman Lomax
                P.O. Box 690
                El Verano, CA 95433

ATOM RSS1 RSS2