Andreas Lydersen at Elmatica

IPC 2581 – when a standard is not enough

Standards are close to my heart. They level the playing field and ensure transparency (if they are any good). But standards aren’t always enough, actually – they seldom are.

Let`s talk about the IPC-2581 standard. A noble effort to fix a lot of what has gone wrong in the exchange of data, in the electronics industry over the last 40 years. Lots of brains, time and effort has been poured into 2581 – so why is it not a huge success? And before you say that it actually is a success, let me give you an example of the adoption of the format: Of the approximately 15K boards we get to quote on in Elmatica each year, only a handful is sent us in the 2581 format. Tragically, we convert those to the old Gerber-format to be able to get them further down the supply chain.

Let’s start with what IPC 2581 is trying to fix
A printed circuit board is exchanged throughout the supply chain (from designer to manufacturer), by a set of files packed together. The files describes the circuits, the board itself and a whole lot of other details. The format that became the de facto standard for this back in the days was “Gerber”. Being an old format, it rapidly became clear that a lot of the information needed to actually make the board manufacturable wasn’t defined in the format. This was solved by adding information to documents and supply them together with the Gerber. But the necessary extra information didn’t have any structured format or use a defined set of terms, so everybody solved their problem by explaining it with their own words in their own language. The result is a supply chain with a lot of manual labour trying to decipher.

How IPC 2581 wants to fix this

  1. Lets create a format that contains everything needed
  2. Make sure that there is a living community that revises the format to include new stuff

Why isn’t it working?
Again, the answer is simple: It’s not working because the proposed solution addresses the wrong part of the supply chain – the parts that are the least hurt by the problems of the current way of doing it. The initiators in the supply chain, the designers and product owners, do it in the way they always have – they send out gerbers. Further down in the supply chain the problems will be found and handled, and the only consequence for the product owners are some more questions and maybe a day or two lost in the ETA of the product.

Why aren’t the people further down the supply chain forcing a change – demanding new formats?

Because we are all playing a game of margins and speed – don’t ask any unnecessary questions – don’t spend too much time – make it as cheap as you can.

How do we solve it?
To me, it is clear that 2581 will not solve this problem by itself in the foreseeable future. It will be adopted slowly by the most eager players, while competing formats such as Gerber X2 will steal some of the market and the rest will keep on with the current formats, such as ODB++ (which doesn’t solve the problem either). 2581 is trying to solve a big problem without offering immediate satisfaction. Remember what the problem was? There is a gap between the information in the gerbers and what is needed to manufacture the board. So let´s fix that problem! Use the parts of the information and efforts that are put down into the 2581 format and subtract it into something standalone and usable for everybody. Let me give you an example:

When you send a Gerber to someone, you have to give them information on the solder mask needed, if any. As there is no set way of supplying this information, everybody will have to read through documents to get it. And is it called “Soldermask” or “Solder Mask”. How do you state which sides you want it on? (top=yes, bottom=yes or should we use “both sides”?). Then there is the problem of colour: should we use colour names, like white, green, red, or state it in a hexadecimal manner? Thickness? As this is not set, people will simply use there own way of specifying it, so no automation is possible.

If IPC agreed on how this should be stated if not part of an 2581 formatted file, this would give an immediate effect in the supply chain: If only two parts in it started using this way of giving that information, it instantaneously gives those two a benefit: the giver will know that it is understood correctly, and the receiver can avoid the manual labour.

Elmatica wants this to happen, and we are committing ourselves to providing both time and hard cash to see it through. We have already started stating what questions are not answered in the gerbers, and which questions usually are raised in the tooling of boards. We have hired developers to create tools for this and have booked our tickets to IPC APAX EXPO in San Diego in 2017, to gather support. Let’s join forces and solve some problems on the path to total 2581 world domination.

Andreas Lydersen, CTO at Elmatica

1 reply
  1. Chris Shaw
    Chris Shaw says:

    Hello Guro, I believe I saw you at the IPC meeting(s) at APEX. I hope you were able to take away from APEX that anyone is welcome to openly contribute to, or voice their opinion on, the IPC-2581 schema. As member of the IPC-2581 technical committee I strongly encourage you to join the committee (for free – see and take part in the biweekly meetings, to highlight any shortcomings, or put forward new ideas. However everything you say is currently missing – unambiguous definitions of thickness and soldermask side & color – is already covered in the IPC-2581 schema. Please let us work together to avoid miscommunication of what IPC-2581 can or cannot do, so that others are not discouraged from adopting it. IPC-2581 is so much more than a gerber replacement, and we share your vision of supporting areas of the supply chain that gerber does not support. I see no point in spending time and cash in creating an entirely new open standard, when IPC-2581 can meet your needs. Finally I whole-heartedly agree with your last statement – “Let’s join forces and solve some problems on the path to total 2581 world domination!”
    Chris Shaw,


Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply

Your email address will not be published. Required fields are marked *