Mity CPU Platforms » MitySOM-335x (ARM Cortex-A8 Based Products)

We're having an issue with data going to the USB ports on our AM3352 SOMs. We're using an FTDI FT240XS USB to 8-bit FIFO on our board to make a higher throughput connection from our microprocessor which is accumulating data rapidly and only transmitting it to the FTDI which is connected to our SOM on USB1. We thought we had this working well for some time now, but with some of our units eventually (sometimes hours, sometimes weeks) the data stops coming over to the USB and the FTDI chip's buffer fills and flags that it's not ready to receive data. What can many times (but not always) repair that is a system reboot, but we don't want to lose any data in normal operation. With a specific SOM that we have, this happens a lot more frequently right after booting up, and no data comes across (although on very few occasions we have seen this unit start up and work for a few hours at a time).

We have multiple baseboards (that have the FTDI chip and MSP430 on them) that we're plugging the SOMs into, and the problem with the one unit that almost never has a good data connection seems to follow the SOMs around, not our hardware or the FTDI chip on the board. We know that the microprocessor we're using only transmits data over the FIFO, and the buffer is filling up on the the FTDI chip (we can tell because it's TXE# pin is going and staying high), while the SOM's USB interface should be reading the data and clearing the FTDI's buffer, but we're getting nothing. We even try resetting the FTDI chip, and linux re instantiates it as the next USB device, however data doesn't appear to be coming across.

We've applied the updates from here https://support.criticallink.com/redmine/projects/armc8-platforms/wiki/Using_USB_port, thinking that maybe some jitter may be causing the SOM to think the USB is disconnecting, and then going from host and into OTG device mode, but it hasn't seemed to fix the issue and the USB still reports as a host when querying it, even after the data seems to be locking up and not making into our board.

We're really not sure what is going on and we're hoping that the guys from Critical Link may have experience some similar things and can suggest somethings we can do to identify the problem further and get closer to fixing the lockups.