MSP430F5522 XT2 Problem

Looks liek you're not the only one with this kind of problems.
There have been more than a couple of other threads with similar symptoms.

We too have lots of devices (about 10% of our production) which won't start their crystals. This is so for about two years now. We didn't change the layout but older batches didn't have this problem (<<1% failure rate).

We suspected the crystals (which we solder together with all the other through-hole parts), despite the fact that out distributor lists them as same type as before. We suspected the capacitors (which are machine-soldered together with all other SMDs by the PCB manufacturer), the firmware (which hasn't changed in this section, but since old parts don't fail with the newest firmware, this is not the reason) and the quality of our own soldering work.

But until recently we didn't suspect the MSPs, as it happens for MSP430F1232 and MSP430F1611 as well. And both didn't undergo a silicon revision when the problem appeared first.

Now it looks like the MSPs of newer production seem to have changed somehow and without having officially changed at all.
It is really disturbing.

Your test with heating the chip was an interesting idea. We will repeat this test as soon as we build the next batch and have some test devices. (usually, changing the crystal and/or the capacitors made the failing devices work, so we don't have one at hand right now).
If we can confirm the same with our devices, this will harden the suspicion of the MSPs being the problem.

I hope we'll get a solution soon (or at least a workign workaround). If not, it can well be that out next generation of metering devices is based on a different processor series. (Which would effectively remove me from this forum. Hint, hint!)

Please keep us informed if you get more insights on this topic.