This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TUSB8043A: The USB 3.0 Hub is Unplugging and drops out of Windows

Part Number: TUSB8043A
Other Parts Discussed in Thread: TUSB1002A, , TUSB8044AEVM

Often while operating in our instrument, the USB 3 HUB (PN: TUSB8043AIRGCR) drops out of Windows as shown in the attached file.

The USB 2 section of the hub is working fine and does not drop out.

We do have some noise on the VBUS line as shown.  However, as I understand it the USB HUB has debouncing logic on the VBUS line.  Why does just the USB 3 section of the hub drop out?

Does TI have any recommendations to improve the robustness of the attached design?

USB 3 Hub Dropping.pdfUSB 3 HUB.pdf

  • Hi,

    The most common cause of USB 3 link drop out would be due to signal integrity degradation. I would recommend implementing redriver like TUSB1002A into your design on the USB 3 lines to help improve the signal integrity. There could also be improvements to the layout that could help here as well. Have you investigated this aspect of your design? 

    To clarify, the measured Vbus in the waveform is on the downstream or upstream port of TUSB8043A? Debouncing of Vbus is done on VBUS_DET for the upstream port connection is done for ~1ms but only for low to high transition.  

  • This also happens when using the TI reference design (PN: TUSB8044AEVM) so this indicates a problem with the hub chip.  See Attached.

    Also, I have recreated the problem on the test bench where there is no VBUS noise that I can measure or trigger on.USB 3 Hub Dropping Forum.pdf

  • Do you happen to have a USB protocol analyzer to decode the packets in the USB3 link? This will tell use exactly what is happening here? Do you have details on how the UBS link is stressed when the "Stress CPU" option is selected? 

    The log shared also indicated that other USB devices are disconnected at the same time and TUS8044AEVM. Has the USB Host disabled the USB port when under stress testing? 

  • Hi Malik,

    As you can see below, it is looking more and more to be a PC bandwidth issue.  Please let me know if you have any other ideas or suggestions.

    Thanks for your help.

    Regards,

    Paul

    Test run while running a program called CPU-Z in “Stress CPU” mode. 

    After a minute or two, there is an Unplug event:

    The USB3 Protocol Analyzer shows this information:

    Unknown /3 PHY Error occurred. According to the Total Phase web site “PHY errors also occur due to electrical inconsistencies. For your setup, we suggest inserting a self-powered SuperSpeed hub between your Beagle USB 5000 v2 analyzer and your target device and the host. This could resolve the problem, if electrical inconsistencies are the cause.”.

    So, I tried a powered USB 3 Beacon Hub and got a similar issue:

    Conclusions:

    • The Unplug events occur with the Beacon USB 3 Hub too. I tested with the USB 3 Hub between the board and the Protocol Analyzer and with the hub between the protocol analyzer and the PC with basically the same result.  This indicates that the problem is not specific to the USB3 DMD board design.
    • Unfortunately, the USB 3 Protocol Analyzer did not shed much light on this issue.
    • This appears to be a PC Bandwidth issue.
  • Hi Paul,

     This is certainly possible that this is a PC bandwidth issue. A direct eye diagram measurement on the PC would help shed light on this. If the USB eye is marginal then signal integrity can be impacted during long  burst read and write test such as what you are testing.