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.

TUSB8020B: USB3 hub enumeration fail on downstream TUSB9261; Enumerates as USB2 device if AC caps removed.

Part Number: TUSB8020B
Other Parts Discussed in Thread: TUSB8041, TUSB9261,

Tool/software:

When connected to a Raspberry Pi CM5 via the SOC's USB3-0 channel, the USB2 segment of the TUSB8020 (pg 6, U19, PID8027) appears under lsusb, but downstream chipsets (U18, U20) do not appear in the list, and USB3 segment of the hub (PID8025) doesn't appear at all.

Removing AC capacitors C101, C106, C113, C116, C119, C120, C121, & C122 allows the downstream TUSB9261s to populate.

Similarly applied hub TUSB8041 (pg 10, U28) on USB3-1 does not have any issue enumerating the same downstream chipset (pg 11, U14) as a USB3 device with AC capacitors intact.

Both implementations of USB3 hubs share the same AC capacitor scheme. Are USB_TX caps C117 & C118 undersized just for that chipset?

See attached schematic for page/part reference. Full design files can be found here: github.com/.../X3

T1-T0ast.X3.pdf

  • Hi Kurt:

      AC capacitors C101, C106, C113, C116, C119, C120, C121, & C122 should be needed  for USB3 to work, do you mean if these AC caps replaced by 0 ohm resistor, TUSb9261 USB3 will show up with lsusb? but you mentioend before hub (PID8025) is not show up with lsusb.

    Best

    Brian

  • Hi Brian.

    What I mean is that if I remove all of those caps and leave all USB3 lanes floating, then the 9261s enumerate as USB2 devices, but if those capacitors are left in place, then they neither populated as USB2 nor USB3.

    What I think is happening is that the USB3 hub segment of the 8020 is internally active and able to communicate downstream to the 9261s, and the 9261s switch to broadcast their ID as USB3 devices, but since that portion of the hub isn't communicating correctly with the Pi, it doesn't appear in the device tree at all. Does that logic track?

    In the documentation for the CM5, the IO/dev kit lacks AC caps on both Rx & Tx between the module and the IO boards USB-A ports. When I connect a discrete TUSB8020B-EVM into either of the USB-A ports, both PIDs appear in the device tree. That is why I included AC caps between the hub Tx and CM's RX, but not vice-versa in my design.

    I've seen further design notes recommending optional AC caps to be placed on some chipsets' Rx lines as well. Since the 8041 populates fully without those caps, I didnt think that those are necessary.

  • Hi Kurt:

       Let's getting hub work first. Both TUSB8020 and TUSB8041 should  be connected for both USB3 and USB2 with lsusb at CM5.

      You need add AC cap 100nf on pin 31/32 if there is no AC cap on CM5 side for USB-0-TX

     .

    Regards

    Brian

  • Noted. I will add that to my design, but actualizing the hardware is going to take more time, as my current PCBAs dont have the requisite pads to place them.

    Still, it's peculiar that the 8041 has better functionality than the 8020 with the same means of connection.

  • did you see USB3 hub for TUSB8041 and TUSb8020B from Cm5?

    Best

    Brian

  • Yes, I saw both PIDs (8140 & 8142), and when I plugged a USB3 loopback from Passmark into one of the downstream USB3-A ports from that hub, it displayed a speed of 5Gbps. The 8020B only showed up as PID 8027

    Fat-fingered that response and accidentally hit solved. To be clear, issue isn't resolved yet

  • for TUSB8020B, since hub USB3 did not work then TUSB9261 USB3 will not work either., but USB2 should work, maybe something else was wrong  for TUSB8020B.  let me check TUSB8020B schematic as well.

    Best

    Brian