chuckw wrote:The fix worked for me, for both the simple testcase and for my main code. I already had an 0805 resistor, and there was plenty of room for it.
Great! I've updated the guide to note that 0805 also fits.
chuckw wrote:Naked eyes may be adequate for young guys, by the way, but for those over 50, serious magnification is likely in order!
Yes, sure. I wasn't trying to brag with the post, more just trying to encourage people who might think SMT is only for gurus with expensive tools. My eyesight is very good but I normally still solder under a microscope or a magnifier anyhow, especially if you're doing a whole board it gets taxing! I've added a sentence to the page to clarify things a bit.
chuckw wrote:Thanks so much for your quick diagnosis of the problem!
No problems, thanks for bringing it to our attention! It's a really unfortunate combination of events, especially the fact that none of the pre-production prototypes exhibited the problem despite being identical to the factory ones (and the testing done in the factory didn't disconnect power for long enough to trigger it).
Atmel have told us they're looking into the undocumented behaviour that seems to underlie the problem, so it'll be interesting to see what they say.