A while back I wrote an article for Electronic Design entitled "Ten Reasons Why I Love The BGA".
After that article, I received a lot of hate mail, mostly from mechanical designers and production test engineers. All had good points. This one was the best. Bill Bowhers of Teradyne, Inc. hits on a number of excellent issues in his note, issues that I am sure are of concern to many engineers. I have little to add, except to say that I can't wait to hear what he has to say about flip-chip packaging...
By the way, I still love the BGA.
There are pros and cons with everything in life. Understanding the down side of a technology - even a winning technology like BGAs - is important to successful application and improvement. What are the reasons we hate the BGA?
- Testability - the ball spacing of most packages is tighter than available probe technology.
- Routability - Even with power and ground at the center of the package, it's damn near impossible to get traces into the fourth row of pads.
- Design Verification - It takes hours to probe up a single timing measurement. The pad escape vias are tiny (also expensive since only one board can be drilled at a time) which force us to work with #30AWG wire for probing.
- Reworkability - If your board has more than one expensive component you would like to fix it whenever you find a bad part. BGAs are difficult to remove without damaging the PC. They are most likely destroyed when removed (making error cause removal impossible).
- Reworkability II - Not even the smallest design blunders can be patched with PC rework. Design verification can be stalled for a week while a quick-turn PC is spun.
That's all I can think of. It would be great to see folks working on these issues but they can't if they're not aware of the down side.
This feedback comes from the Automatic Test Equipment industry. With 15x20 in PCs [19 BGAs, 24 160p PQFPs and memory], we understand that we are outside the norms of the PC industry, but these are real issues that we face.