Is this still an open issue, or was it addressed in another discussion?
1 of 1 people found this helpful
It appears that this was caused by inadequate grounding in the layout. In my original design the ground pads were connected by fairly thin traces and it's possible there could be voltage differences between them. In my new design all ground pads are connected by a pour, and the problem seems to have gone away.
BTW this bizarre behavior was witnessed by a broadcom field engineer who was supporting me, and in that instance the poor grounding actually caused the chip to self-destruct by drawing a large current.
I definitely recommend to anyone using the chip to ensure adequate grounding, as in the layout now published on this site. (Unfortunately when I did my original design there were no sample layouts published)
Thanks for the clarification Lewis as this should be useful to others.