Issue link: https://iconnect007.uberflip.com/i/396835
October 2014 • The PCB Design Magazine 55 Not all inventors are normal; I have been told that many times. I don't get along well with big stiff companies. I don't wear suits; I don't fit the normal engineer mold. I invent way-out ideas; I think in outer space. I would have fit in fine in a "skunk works" R&D department decades ago. " " bob and me TARzWELL'S FIRST—AND LAST—LEAN MEETING continues menting with them and continued pondering about it. Then one day I saw the difference. It was as clear as mud! It was so simple I missed it, like missing the forest for the trees. The one great panel was made with a mistake and the operator tried to fix it, hiding a problem that he created. In reality, he unknowingly created a super board! I re-created the difference with a new test panel and got the same results—another super board! Wow. I could not make these vias crack. Via cracking is one of the single biggest problems in our business, and I found a simple, easy trick to solve it. Soon, the idea was sold, and it is now on the market. Not all inventors are nor- mal; I have been told that many times. I don't get along well with big stiff companies. I don't wear suits; I don't fit the normal engineer mold. I invent way-out ideas; I think in outer space. I would have fit in fine in a "skunk works" R&D department decades ago. So what's the simple trick to making a super board, and why would I tell you now? Be- cause the fact is, it's so simple that most people won't even believe it. That's how I know my secret is safe, because you probably won't even attempt it. Here it is: The big/little secret is a physics attribute called Young's modulus. In short, Young's modulus is a measure of how hard a material expands with heat, or how hard it pushes as you heat it. We all talk about CTE, and, in fact, far too many articles work CTE to death. Yet none mention Young's modulus, its silent partner. Here's an easy way to understand how im- portant Young's modulus is. Let's do a simple experiment: Take a hunk of FR-4, tighten it in a vice, heat it up, and nothing will happen. But take a hunk of steel, tighten it in a vice, heat it up and watch it break the vice. FR-4 only ex- pands in the Z-axis with 3 million pounds per square inch (21 GPa) of force when being heat- ed, but steel expands at 30 million pounds (200 GPa) of force, pushing (or resisting) 10x harder than FR-4 as it expands. Copper pushes with 17 million pounds per square inch (17GPa), nearly six times more than FR-4. When the FR-4 board is expanding in the Z- axis, it's got just enough expanding force to rip apart the thin copper in the hole; even though copper is six times stronger, it's also thin, so it stretches and cracks. If the copper is thin enough, less than about 1 mil, it will cause cracking with thermal cycles. The expanding FR-4 in the Z-axis will cause the top of the copper via pad to tilt up as the FR-4 tries to ex- pand faster than the copper, with the copper wall resist- ing. When we put 1.8 mils of copper in the hole, it stops a lot of the upward movement of the copper hole wall and therefore reduces pad move- ment and hence the reduced cracking of the copper via. Note: This 1 mil cliff edge of failure we stand perched upon also just happens to be right dead in the middle of where we plate our holes. If the board shop hap- pens to put a bit more copper in the holes, you'll have a PCB that will survive more thermal cycles. But if they are a little skinny in the plating, your PCB will fail earlier. This is how close to the cliff edge we have been stand- ing for the 50 years I have been in PCB manufac- turing. So who decided that 1 mil of copper in the hole was a standard back in the beginning? No one said, "Hmm. How much copper do we put in the holes to give the circuit a long life?" We just plated what we could back in the '60s; there were no tests to see how much copper will survive thermal cycles. It just happened, and it continues today. A simple, little change like starting with thinner base copper and then adding 2 mils of dry film and plate-up close to the top of the dry