After going several years between updates, IPC has released IPC-2221B, the industry standard for designing printed circuit boards. PCD&F editor in chief Mike Buetow spoke in January with Task Group chairman Gary Ferrari about the critical changes to the spec.
[Disclosure: Buetow worked as an IPC staff member under Ferrari on the original IPC-2221.]
MB: It’s been a long road since the IPC-2221A was released. What’s new, and what’s changed?
GF: IPC-2221A didn’t cover anything on lead-free. We wanted to make sure that in the appropriate sections we included warnings or advice on as much of lead-free as we could – the materials, decomposition, and so on – that’s a big factor in the materials set. Then there are all the new surface finishes. For tin-lead, everyone used ENIG and HASL. Now we have nine or 10 finishes: ENEPIG, immersion silver, immersion tin, electroless gold, it goes on and on. IPC-2221B includes a chart that lists all the surface finishes, and what they are recommended for – e.g., keyboards, keypads, gold wirebond – what’s good for RF, what’s not good for RF. It’s all shown in a Consumer Reports type of layout.
We realized that there was no point for [the IPC-2221 task group] to try to re-describe all these finishes; that’s what the IPC Surface Finishes Task Group is better suited to do. So we went to [Surface Finishes Task Group chairmen] George Milad and Gerard O’Brien and asked for descriptive summaries for each of the most popular surface finishes. Information such as typical thicknesses, what applications they are good, and not good for, and shelf life, to name a few.
Then we conducted a big research project on the press-fit connectors. That one was tough. Phil Henault of Raytheon was in charge of that effort. He did a lot of research. At the end of the day, he had all this data. Each connector had its own hole size; you couldn’t really come up with a recommendation that was common to all. We finally came up with a tutorial about press-fit pins and what to look for, but could not come up with a table for recommended hole sizes because it was all over the map.
We expanded the sections on thermal management and metal core designs. This wasn’t really influenced by LEDs; 80% to 90% of the task group members come from the aerospace sector. We tried to glean information from wherever we could.
The biggest controversy was the development of a new quality conformance coupon set. We spun off a task group, chaired by Tim Estes of CAT to look at the coupons. In the past, we would develop all the coupons, and the IPC-6012 Task Group would pick them up and apply the appropriate testing, but the IPC-2221 had to invoke them and show the design in the design standard. This time, we spun it into an Appendix, which can be updated on a regular basis instead of waiting for a major revision. The instructions we gave Tim were simple: Look at the coupons and find out whether people were using them. If not, why not? And if they don’t apply anymore, get them out of the standard.
As an example, the registration coupon came from IBM years ago to measure annual ring and measure breakout using an x-ray. With all the copper planes, you couldn’t see anything. We scuttled that, and developed a new registration coupon that is mathematically developed. We modified or added new coupons – there are nine new coupon designs now.
This was a really big one. What we were afraid of is that, when a coupon says you need to take the worst-case land size and worst-case component, there’s no guarantee the board shop will use the right one. There’s no advice about which blind and buried vias to use, and today’s designs are usually so dense, the plating of holes is affected by the hole density in an area. So we now have a coupon that reflects the density we see today. The concern we had was, we wanted to develop a software package that could be distributed to the fabricators that would automatically extract from the native CAD files what data was needed to create the coupons. There would be no human intervention. We should have a software validator for the extraction program and generator. You can’t just go with Gerber; it needs to be the IPC-2581 [the data transfer standard]. That’s the primary route we’re going to go with.”
[continues on next page]
What we wound up doing, since we know the industry won’t just move to something new, is to keep the legacy coupons in an Appendix B, as is, they won’t be supported moving forward. So the aerospace guys won’t have to go back and revise their drawings. But in the next revision, we’ll make the swap and say, “This is the one you must have.”
We also fine-tuned a bunch of other sections. Laminate materials; we added info about high-pot testing. We recommended values and ranges to test at. For plating, we had to include several of the new finishes. We addressed hole fill, adding a table for clarity.
IPC-4761 talks about the three ways vias are protected: partial fill, total fill and tenting. We took the IPC-4761 chart showing when to use and how to use each one, and added design criteria. For certain via protection, we said, “You just don’t do it.” Then we added limits for minimum and maximum hole size and tolerances for via protection.
We also extracted anything of value from the now defunct IPC-2224, the PCMCIA standard, and made sure it was included in our document. We went through the thin core spec (IPC-2224) because everyone is using those materials.
Finally, we addressed problems with microvias. The conventional definition said a microvia hole is 0.006" and below. The problem is the hole wall plating, which is typically 0.001" in the hole. Microvias plate very quickly and generally very small. So the spec allowed the hole wall plating to be half the amount a through hole wall plating would be. Companies are mechanically drilling holes at 0.006", and since the hole fits the definition of 0.006" and below, they are providing half the plating and indicating that it meets the specification. So we redefined the microvia based on an aspect ratio, so someone drilling a hole at 0.005" would have to still follow drilled-hole plating specs if they were above the defined aspect ratio.
MB: Going back to when I worked on the standard there's been the question of how much tutorial to leave in. Given that the IPC-D-330 Design Guide is 20 years old, did you talk about moving the “how-to” to the Design Guide and leaving just the minimums and maximums in IPC-2221?
GF: The discussion came up, but we left the tutorial in, thinking that if the standard just gives a number, designers won’t know how it was derived. Besides IPC-D-330 was nothing more than excerpts from the released standards and is in dire need for an update.
MB: You’ve been working on the industry design standards for 30 years. What did you do differently this time around?
GF: I think what we did very differently was to use sub-task groups, such as for press-fit components … we took out all the test coupons and made that effort a dedicated task group … going out to the Plating Subcommittee. That was a departure from the past; being able to get a group offline to put together what they think and let them present to us and kick the tires.
