PCBDESIGN007 - An IConnect007 Publication
Tweet This ArticleTweet    Send Us Feedback About This ArticleFeedback    E-Mail This Article To A FriendE-Mail    Print This ArticlePrint
Letters, Part 3: Let's Have a Data Transfer Format Discussion
Wednesday, May 23, 2012 | Julian Coates, Mentor Graphics

Dear Mr. Shaughnessy,

In response to Jack Olson’s letter (for which I thank him – it is good to openly debate this matter!), I should start by making clear that no implications were intended about “dumbness” – in fact the “dumb” theme was kicked off in your original article about the roundtable discussion at SMTA Atlanta!

To the essence, I respectfully take issue with Jack’s suggestion that I am “twisting the truth.” Please consider the following points and suggestions: 

1.  Reverse engineering. In my experience, every CAM system at a PCB fabrication house reads all of those multiple files (Gerber/Excellon/Netlist for example) and integrates the data together in the database of the CAM system before starting on the generation of the actual tooling data for manufacturing. This data re-integration in the CAM system takes time and has to be verified in some way. This is the wasted effort I am referring to when I speak about reverse-engineering. Perhaps it would be better to speak about “CAM re-integration” of the data instead, because no bona fide manufacturer is interested in reverse-engineering the design data for any other purpose. This data re-integration happens in every CAM system I know. Why pay for that when it can be avoided? Why do it when you have the alternative of delivering ODB++ data in one file with all of the required data (note: neither more nor less) already integrated?

2.  Data content. The question of what data content to send to a manufacturer and whether it should be in separate files or one integrated data container are really two separate questions. For sure the designer should only send the data content to a manufacturer that enables him to sleep at night, yet is everything (and not more than) that the manufacturer needs to do his job. The intelligent formats such as ODB++ support the selection or deselection of content by the designer, as required, while at the same time avoiding the wasteful re-integration step I refer to in the previous point.

3.  Level of awareness. On the subject of the survey, I must say that I was as surprised by the results as Jack was. But the results cannot be denied – the majority of designers are relatively unaware of the opportunity to move away from Gerber and use ODB++. Jack and I find it surprising, because we are well aware of the advantages of formats such as ODB++, but the fact is there are a lot of designers out there who believe that Gerber is still best practice! I would suggest that we need to continue the communications efforts, industrywide.

As an open invitation to Jack and any other interested parties who want to take cost, time and quality-risk out of the design-manufacturing process, I would say: Let’s go together to your manufacturers, see how they handle your manufacturing data today and see whether changes can be made that will increase efficiency without changing the total data content involved. Nothing to lose, everything to gain, I suggest.

Let’s give it a try and see what happens! I would like to learn the realities here also – as a shared discovery process with Jack and other experienced members of the PCB designer community.

Yours sincerely,

Julian Coates
ODB++ Solutions Alliance Manager


To sign up for our newsletters, click here.

Tweet This ArticleTweet    Send Us Feedback About This ArticleFeedback    E-Mail This Article To A FriendE-Mail    Print This ArticlePrint


Related Articles:
  • Karel Tavernier: Gerber Definitely Not "Dumb" Data
  • Andy Kowalewski: Time for a Single Data File Format
  • Letter to the Editor: Coates' Letter Draws Comments
  • Letter to the Editor: Gerber's "Dumb" Data is no Advantage
  • The Shaughnessy Report: Designers' Roundtable Wrapup

  • MOST READ
    MOST EMAILED