During the Design Forum at IPC APEX EXPO 2018, Jan Pedersen, senior technical advisor for the PCB broker Elmatica, gave a presentation on CircuitData. The language is designed to help facilitate other design data transfer formats such as Gerber, ODB++, and IPC-2581. Jan spoke with Managing Editor Andy Shaughnessy and Contributing Technical Editor Happy Holden about how this open language works with the existing data formats, as well as the need to eliminate paper documents from design process, and how the industry can help shape this open-source language.
Andy Shaughnessy: Jan, I enjoyed your presentation. During your talk, you discussed the CircuitData tool. Tell us a little bit about that and how it works.
Jan Pedersen: Yes. CircuitData is a language. It's a digital language to communicate an article specification. And the corporate profiles, like EMS profiles, have requirements that are added normally to PCB specification as a PDF file or as a paper. We want to make these digital. We want to make it easy and we want to make it open-sourced so that it's not a limitation, or competition for any company. There should not be any competition between different companies. CircuitData is primarily a language to communicate the article specification.
Happy Holden: Can you step back and define the difference between a language and a PDF data file? Right now, information is in a PDF data file, so what is the difference if those words are in this language?
Pedersen: The language is a digital language that the computer can read. A computer cannot read a PDF.
Holden: What do you mean by "read"?
Pedersen: You can read it through your own ERP system. You probably have a template yourself where you have the article specification in your ERP. You can read this file directly or you can export it from your own ERP system. It's not a new a software. It’s just a language, and a way to communicate an article specification digitally.
Holden: Is this a language that allows it to be a schema for a database?
Pedersen: It is a schema. It's very similar to IPC-2581.
Holden: Is it a database that you can then search on? Whereas with a PDF you can't do a search and it doesn't know what that field means that has a particular word in it?
Shaughnessy: You say it works with ODB++, Gerber, and IPC-2581. What are some of the challenges of working with those formats that your language helps overcome?
Pedersen: How it is today is that we receive simple Gerber files, external Gerber files, all of these things, but it doesn't say anything about the stack-up, the material, the color of the solder mask or the final finish? It says nothing.
What we're trying to do is to make this article specification that comes as read-me file, a PDF or Excel or whatever it is, so you get an export file besides the Gerber. Today, IPC-2581 does that, but that's very comprehensive. We want to make it simple, open-sourced and free of charge.
Shaughnessy: It sounds like you're automating it more, and making it more user-friendly.
Pedersen: Yes, absolutely. But what we do today is this: We receive all of these papers, or documents that are like paper. And we need to translate them. We type it into our system, and when it comes to the board shop, they retype it. They type it into their system so they can create their own job cards so they can do feasibility studies. They need to know "What do we have here? How do we describe this board that we are receiving? Is it an RFQ or an order?"
Now, this is the description of the board, the article specification that starts with the designer. Then it comes to the EMS company, they have some other requirements so they can update the file. And then they have one file that they send to a broker like us, or directly to the PCB shop. So, for the PCB shop, the ultimate thing is to receive one file that they can read into their own computer, in their own language.
Shaughnessy: So, when someone is sending 2581, Gerber, or ODB++ files to the CAM guy, are your files a part of that? How does it work?
Pedersen: It's additional to the Gerber. The Gerber is fabrication data. That's how you make the board. That's the image of the board, more or less.
Holden: The output of a CAD system, or even IPC-2581, is still not sufficient to build a board. There are all of these specifications, notes, drawing—all kinds of additional paperwork, files, etc. And then you try to incorporate that data into a tooling scheme so they output it for a digitally-run factory. But until it's all data in the computer, you cannot output it.
Pedersen: You always have notes that you want to incorporate. There are notes and drawings, corporate requirements that come in as a note.
Holden: The worst part is the revision files. The OEMs, if the product is out there and it's on model D, model E, and model F, and a part becomes obsolete or bad quality and they replace it with this other part, they change the land pattern, but everything else is the same and eventually it'll reach obsolescence. And even in obsolescence the OEM has to manage that. They're shooting out files to fabricators and to assemblers but they’ve still got this genealogy of this product until it's totally dead. And some of these products have to live 20−30 years, by law.
Pedersen: Let’s say you design a board and you want gold. You call it "flash gold." I don't know that is in America but in China that's electrolytic gold. It's used as an etching resist. Some people use the word "flash gold," but they mean ENIG.
Holden: That's not what flash gold means [laughter].
Pedersen: But they're using different languages, and different words. That's a good example. If you're following one language, everybody understands what you want. And that language is digital, so when you read the file you can have it pre-translated to simplified Chinese.
Holden: It's never easy to translate into Chinese [laughter].
Pedersen: But it reads in as “green oil,” you know what I mean? Because in Chinese, they call solder mask green oil.
Holden: It has to be a data dictionary. So that they can take this data and the two characters for it may be green oil.
Pedersen: We are using T-50, I think it's called, the IPC standard terms, as far as it's defined. If there are new words, then we need to develop that.
Holden: We have 12 different Chinese sets of characters for mouse bites [laughter]. Because it's the rodent that runs in the forest eating on the bamboo tree, that's what the characters actually mean for mouse bites. Because in the English, German, or Russian, you can translate mouse bites into those languages. In Chinese, you have to use specific characters and you can't invent a new character like you can in any other language.
Pedersen: So we're trying to simplify the understanding and the communication of the article specification.
Shaughnessy: So what's been the response of the fabricators to CircuitData?
Pedersen: It's very good. Those that we have been talking with have said "Yes, we need it." We get a million different kinds of documents in, and if we can do everything digitally, like you say, then they are very happy with it. As I said, I support 2581, and I'm part of the consortium, but we only receive a handful every year.
Holden: What you need are some leaders like Ericsson or Nokia.
Pedersen: We have former Ericsson with us today, and this system would be beautiful for those kinds of companies. If they can implement this, they can communicate internally, accurately and everybody knows what you mean.
Shaughnessy: If people want to get involved, where can people go to find out more?
Pedersen: You go to www.circuitdata.org. Everybody can join. And we are also searching for people for the CircuitData board.
Shaughnessy: And it's an open source, so they can help create it?
Pedersen: Absolutely. It's completely open-sourced. It's free of charge. And to program this into your own ERP system, it should take between two and five days. That's affordable. At Elmatica, we say that we are successful if we have one customer and one supplier with us. We have already gained a lot. That is how much you actually win by using it.
Shaughnessy: Anything else you want to add?
Pedersen: The one thing I maybe would want to add is regarding the laminate database that we have launched. I think this is the first digital database that allows you to search for the laminate you want. You can search either by parameter, like DK, or whatever you want…CTI, volume, etc.
Holden: Can you send us a copy of that?
Pedersen: Sure, of course.
Holden: One of the things you need, if that's really an extensive database, is an output format for the stack-up software in Polar Instruments.
Pedersen: Yes, we will use it in the stack-up in CircuitData.
Holden: Because Polar needs the material characteristics. They have their own schema, but if you could translate from your database to their database, then Polar can import that.
Pedersen: I'm sure we can talk with Polar about that. And that's the first thing we'll do is talk with people like Polar.
Shaughnessy: I appreciate your time, Jan.
Pedersen: Thank you both.
Holden: Thank you.