Hi Julian,
You make some very good points and I hope that ODB++, or something very much
like it eventually replaces Gerber. Your arguments aren't specific to
Valor, so that's not an issue at all. I'd like to mention a couple of areas
where we view things a little differently.
In comparing the previous transition from film to Gerber to the imminent
transition from Gerber to an intelligent format, the analogy is not as
strong as you imply. In the transition from film to Gerber, there were
rarely any differences between the two, at least once people learned how to
take care of their photoplotters. This is not surprising, since the early
photoplotters came from the same company that wrote the standard! Even so,
the engineering groups I ran during that period only archived one format of
data. When we changed over from film to Gerber, only the Gerbers were
archived. We added the additional process step, as did many groups back
then, of approving the film plots from the board vendor. This was not to
check for improper interpretation of the Gerber data by the photoplotter, it
was to check that the Gerber data represented the native CAD data properly.
For us, these approval films were really check plots that verified the
Gerber generator in the CAD software. Many board vendors required the
customer to approve the films before fabrication anyway. If I recall
correctly, it was a number of years before affordable Gerber viewers
appeared. After using those for a while, the vendors no longer required our
sign-off on films and we were confident enough to not require it either.
The coming transition is a bit more dicey. We're talking about changing
from a simple, dumb format to a complex hierarchical one. The complexity of
the software to generate and read this new format is much higher than for
Gerber. We can therefore expect more software problems than with the
previous transition. I would like to know what are the tools many designers
currently have, that you mentioned, to compare Gerber and ODB++. If they're
affordable, I'd like to try them out. Obviously the fabricators can do this
task but their software costs more than most CAD packages we use to design
the boards, so that is a barrier.
While we could also use high-end CAM software to improve our design quality
by eliminating slivers, acid traps, etc., the situation is the same as I
mentioned before: the board vendors already do this for us and are not
offering reduced prices if we take on that part of the job. The end-product
quality is the same, it's just a question of who does the work. Likewise,
they are not offering improved turnaround times. They are not even telling
us they are considering such policies. This is a very tough sell to top
management who want to know how much it will pay back and when.
Remember that when we switched from film to Gerber, many fabrication shops
bought their own photoplotters and quickly reduced or eliminated
photoplotting charges. The company I worked for at the time was able to
junk our photoplotter and darkroom, reclaim the space they occupied and put
the operator to better use as a PCB layout person. Similarly, when we
stopped requiring sign-off on films, our turnaround times improved
instantly. If an investment of time and money today won't lower our board
prices, shorten the turnaround time or improve quality, it would be more
accurate to call it a donation.
I'm still very much in favor of adopting an intelligent data format. But it
has to be driven by economics. The return has to be at least in the
foreseeable future. The fab shops are apparently not yet able to make it
profitable for us to make the change. Hopefully, that situation will
change.
Regards,
Seth Goodman
Goodman Associates, LLC
tel 608.833.9933
fax 608.833.9966
> -----Original Message-----
> From: Julian Coates [mailto:[log in to unmask]]
> Sent: Thursday, January 31, 2002 3:25 AM
> To: [log in to unmask]; [log in to unmask]
> Subject: RE: [TN] ODB++
>
>
> Hi All,
>
> I am from Valor, I should state that up front, but I would like
> to offer two
> points for consideration:
>
> 1. In the mid-1980's the mainstream PCB manufacturers began to
> accept Gerber
> (instead of film) into their tooling departments. In order to do
> this, they
> had to make capital investments. These investments gave them a competitive
> advantage, because they were able to cut costs overall, reduce
> lot-size due
> to faster tooling cycles, and increase quality. Despite this, for many
> years, the OEMs (their customers) continued to send the design to the
> fabricator in two formats; Gerber, and film. The film was regarded as the
> reference and ultimate specification of the design, but the
> fabricator could
> use the Gerber if he so chose (those who could, did). What is going on now
> between Gerber and ODB++ is basically the same, a transition of formats
> which have uncertainties between them, and for each design somewhere down
> the line somebody has to decide which format to believe in and
> work from. In
> the case of Gerber, it took a few years for the average fabricator to have
> unconditional confidence in the data rather than the film. These
> days, they
> do not receive films at all, and they just work from the Gerber without a
> second thought of what became of the film. In the case of ODB++, many
> fabricators already have tools (as do the designers by the way) to compare
> the graphics and netlist interconnect of the ODB++ versus the Gerber in an
> automated fashion, reducing the risks substantially. The upside (as Mark
> points out) is faster throughput and higher quality. The question
> is: who is
> the beneficiary? The ultimate beneficiary of these gains is the consumer
> (the designer), via the competitive business model of the
> fabrication-outsourcing. From my observation, most fabricators do not
> recover their actual tooling costs in the charges they make to
> the customers
> anyway (they bury their tooling costs in their overheads), so it
> is probably
> unrealistic to expect PCB fabricators to give tooling price-cuts to the
> customers, since they are running their tooling operations at a loss
> already. However, there are gains for the customer, albeit
> indirect: further
> reduction in average delivery times, and higher quality over time; prices
> are dropping anyway. To turn it around, if the designers continue to send
> Gerber, the level of service received from the manufacturers could not
> improve so fast as it can with a smarter interchange format.
>
> 2. Seth makes a good point about validation of ODB++ input and output
> processors. Currently, ODB++ is not a formal standard, though we
> believe it
> will be in good time (thanks to the industry-driven NEMI
> process). When the
> formal standardization occurs, we hope that some independent
> compliance-validation service will be offered by the
> standards-body. Just as
> a side note, independent compliance-proving by an independent body never
> happened (as far as I know) with Gerber RS274D, and RS274X never was a
> standard anyway, the industry just went ahead and implemented the format
> widely, though with considerable variation, as Seth points out.
> Between now
> and formal standardisation of ODB++ or some further-developed
> version, Valor
> offers a service of supporting third parties who implement ODB++, via its
> 3rd-party cooperation program, the "Open Systems Alliance". We do our best
> to offer advice and support for interface testing to all organisations,
> competitive or non-competitive, so that the format has the best chance of
> practical implementation. Some 3rd-parties opt to stay away from this
> service due to competitive reasons, which usually (in our view) results in
> sub-standard implementation of the format, which is a pity. Those who join
> get full support.
>
> After many years of quasi-stability with Gerber, the industry is in
> transition again. I do not sure that there is any clear way of
> accelerating
> the shift to a new CAD/CAM format with price-breaks from the manufacturers
> (fabrication or assembly), since they are all bleeding cash right now
> anyway. The transition will happen steadily, and be driven
> top-down by more
> subtle, but more powerful, factors such as time and quality.
>
> I hope these comment contribute to the debate in a positive way, as
> intended.
>
> Julian Coates
> Valor
---------------------------------------------------------------------------------
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
---------------------------------------------------------------------------------
|
