TECHNET Archives

January 2002

TechNet@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:
Seth Goodman <[log in to unmask]>
Reply To:
Date:
Thu, 31 Jan 2002 00:24:00 -0600
Content-Type:
text/plain
Parts/Attachments:
text/plain (91 lines)
Hi Mark,

> I assume you are comfortable with your system outputting
> gerber, it's ODB that has to "prove" it can duplicate the
> gerber.

Thanks for bringing this up.  I'm not suggesting that ODB++ needs to prove
itself, as its' market share among fab shops tells me that it does the job.
Given the choice, I would also opt for ODB++ as it can carry a lot more than
just bare board information.  The problem I have is in archiving any data
object in more than one format for the reasons explained below.


> Here's what I have in mind, an experiment. You asked the
> question "The problem is how do you insure that both sets
> of output data are identical in every way?" Maybe one way
> to test that is to test the input and output capabilities
> of Genesis/Enterprise.

This will test, for one particular data set, the ability of the Valor tool
to translate back and forth between Gerber and ODB++ without accumulating
error.  It really shows that their forward and reverse translators are
accurate inverses of each other.  It doesn't show that the two data files
produce the same board.  Even if we made a test to show that these two data
files were comparable, not very many designers use Valor tools to make their
output files.  Instead, we typically use the data file generators built into
our CAD systems.  If the CAD system doesn't produce ODB++ directly, and most
don't, then we use a third party tool (I use CAM350) to translate into
ODB++, either from the native CAD file or from Gerbers.

Even if the original CAD tool produced both output formats, the problem
remains.  Producing an output file, either Gerber or ODB++, from a native
CAD database is a translation process.  It is more difficult, though
fundamentally similar to translating Gerber into ODB++.  In both cases, some
program looks at a series of data objects and translates them into another
series of data objects of a different format.  That program is written by
human beings and therefore it has bugs.  I have yet to own a piece of
bug-free software.  The software that produces ODB++ is not the same as the
software that produces Gerber.  It most likely has different bugs.  My point
is that the two output files will not be exactly the same.

I know from experience that there are flaws in the Gerber data generators in
most CAD programs.  The nastier ones are history but some subtle ones
remain.  Similarly, different Gerber viewers can display the same data file
differently.  Occasionally, I have to spend a lot of time working with a fab
shop because their Gerber viewers and my Gerber viewers show different
results from the same data.  This is an unfortunate waste of both my time
and theirs, but it has to be done.  Because we don't use the same brand and
version of software, it is unavoidable.

What I want to avoid is dealing with problems in ODB++ data in addition to
Gerber data.  Also, consider that if ODB++ becomes the Gerber replacement,
there will be companies other than Valor writing software for it.  That
means there will be some differences between how a Valor program and someone
else's program interprets the same ODB++ data.  This may not come up very
often, but it will come up, especially when other vendors first start using
ODB++.  Unless Valor achieves a 100% market share, we will all have to deal
with some subtle differences between vendors' software.  This problem could
be lessened if an independent industry group produced a validation suite
that a program would have to pass to call itself ODB++ compliant.

I would be happy to support ODB++ as the single output format for any given
board.  In fact, I would prefer almost any intelligent data format to
Gerber, which is truly a rotten old standard.  But until my customers'
purchasing departments will accept ODB++ only, the only "universal" option
today is Gerber.  Until ODB++ becomes as universally accepted by fab shops
as Gerber, or the fab shops start to give a big enough discount to offset
the extra costs of supporting two data formats, IMO most designers will opt
to stay with Gerber.  If using ODB++ really saves the fab shops money, and
I'm sure it does, all they have to do is pass some of those savings on to
their customers who use it and it will become the de facto standard before
you know it.

Regards,

Seth Goodman
Goodman Associates, LLC
tel 608.833.9933
fax 608.833.9966

---------------------------------------------------------------------------------
Technet Mail List provided as a free service by IPC using LISTSERV 1.8d
To unsubscribe, send a message to [log in to unmask] with following text in
the BODY (NOT the subject field): SIGNOFF Technet
To temporarily halt delivery of Technet send e-mail to [log in to unmask]: SET Technet NOMAIL
To receive ONE mailing per day of all the posts: send e-mail to [log in to unmask]: SET Technet Digest
Search previous postings at: www.ipc.org > On-Line Resources & Databases > E-mail Archives
Please visit IPC web site (http://www.ipc.org/html/forum.htm) for additional
information, or contact Keach Sasamori at [log in to unmask] or 847-509-9700 ext.5315
---------------------------------------------------------------------------------

ATOM RSS1 RSS2