Other Parts Discussed in Thread: TUSB8041,
We designed a custom board with TUSB8041 connected to AM5728 CPU to provide USB host port expansion. During hardware bringup, we noticed that the downstream ports of TUSB8041 were not working in some of the power-cycles. The AM5728 CPU is running on Linux. We use the 'lsusb' command to list the USB devices connected to AM5728. When TUSB8041 is not working, the result of 'lsusb' command shows that no USB device is connected to the AM5728 CPU, i.e. the AM5728 CPU did not recognize the existence of the TUSB8041 hub. The issue appears about once in 10 power-ups. During the power-cycles in which the AM5728 CPU successfully finds the TUSB8041 hub, the connection is very stable thoughout the power-cycle, no connection drop-out or transmission error at all.
We tried repeatedly resetting the TUSB8041 hub during one power-cycle. The same issue came up about the same frequency (1/10) as from cold start. The AM5728 CPU could not find the TUSB8041 hub after some reset release. Based on this observation, we assume the issue is unlikely to be caused by wrong power-up and reset sequence.
We also tried:
1. Disconnecting the USB 3.0 connection between the AM5728 CPU and the TUSB8041 hub by removing the AC-coupling capacitor, effectively using TUSB8041 as an USB 2.0 hub;
2. Cut the USB 2.0 connection between the AM5728 CPU and the TUSB8041, effectively using TUSB8041 as an USB 3.0-only hub.
The repeated resetting tests with the above hardware modifications show the exact same issue with very similar frequency.... So signal integrity issue is also unlikely.
I'm completely new to USB debugging. Any pointer on what I could try next is welcomed!
Thanks