We have relative high production failure ratio with AD9388A. About 3-5% of the chips must be replaced manually on the board, because digital interface doesn't work. The symptoms are same with all failed chips: there is a 2.6 khz 2.4V pk-pk square signal added to the incoming TMDS clock lines (and sometimes to data lines) with 0V low-level DC voltage. It seems that the chips turn-off and turn-on the internal terminal resistors periodically (the incoming clock is not detected according to the registers). Replacement of the chips has solved the issue in every cases. The failure always occurs at the first test after production, the replaced (or not failed) chips are reliable in long-term.
Please let me know if anybody has an idea about the possible origin of the issue. (our first idea was ESD discharge, but verifying the production process had no result).