IN a custom design using the Kinetis K61 we have a directly connected USB interface (no ULPI), running in full-speed mode as a usb device, not host. It has protection diodes commonly used for a usb device connection on the D+ and D- lines. While testing our usb stack we connect and reconnect the cable multiple times. During this process we have managed to kill the interface on two boards in just a few days. The internal regulator seems to be shot as it only produces about a half volt from the 5V supplied. Static precautions are taken. The device descriptors are set to bus powered mode, no current requested.
Has this problem been seen before? What can we do to fix and/or prevent this?