Temp delta limit across BGA in reflow

I'm trying to help one of our remote sites solve a problem with a warped PBGA. The corners are lifted, creating columnar joints and the central balls are sometimes shorting. I've verified that it's not popcorning and I've got samples going to the lab to verify via DSC that the molding compound is fully cured. The site was just audited last week to verify their MSL compliance, so I'm reasonably confident that they're being careful in that regard. I'm really narrowing in on the reflow profile.They're using a steady ramp profile that hits it's 217C peak at about 300 seconds. I had them place thermocouples on top of the BGA and on the board up next to the perimeter balls. These two thermocouples track within 2-3 degrees of eachother right up to the peak, but deviate by as much as 25C in the cooldown zone with the top of the BGA being cooler than the board. This morning I'll have the site try to bring these temperatures closer together, but I'm wondering if anyone has a spec they place on the difference between top and bottom of a PBGA. What's the max delta you would find acceptable?

Also, I'm assuming that the profile from ambient to the peak is OK since the top and bottom track so well in that region, but if anyone thinks the 300 seconds to peak is too fast or that I should have a soak zone I'm all ears.

reply »