We use the CY7C68013A in the QFN56 package on our products. We have received a couple of boards back from the field that have worked and then aren't enumerated after being plugged into a computer. On closer inspection with a scope the I2C bus load executes as expected so the 8051 side seems to be functioning fine, however one board has a constant 1V on both D+/D-, and the other board has 500mV on D+/D-. There is no visible damage to the package, also after inspecting them under a microscope.
Does this sound like the USB port has taken a hit somehow, or could there be another explanation for this behaviour?
To answer my own question in case anyone looks here with a similar issue...
I sent two devices to Cypress for failure analysis and both devices had suffered an Electrical Overstress event (EOS) on the USB data pins. It's impossible post mortem to say what caused the error (though it was not an ESD type event), suffice to say that the devices are beyond repair.