TUSB2036: TUSB2036 - "Port reset failed" error

Apologies, the error is "Device Descriptor Request Failed" when connected directly to a windows 11 machine. If connected indirectly through another USB hub, the error is "Port reset failed" - but this is not a valid use case in production. 

Another note - the suspend pin is always asserted except when the reset pin is pulled low.

Let me review schematic first.

Best

brian

great, thanks

Hi Chris:

  What is your real clock connection ? 6Mhz oscillator or 48Mhz clock?

For Reset: you need to replace R309 with 01.uf cap, and R302 should pullup to 3.3v.

Best

Brian

We are using the 48MHz oscillator (and per note 2 above, even though the SCH doesn't show it, the mode pin is now asserted to specify the 48MHz clock). Per note #3 above, we did already update the reset circuitry to provide a reset pulse. Unfortunately we saw no change in performance. We've also tried resetting the part manually after initial power up (by briefly tying the reset line to ground), and the results are the same. Any thoughts?

DId you try with Windows 10?

Regards

brian

No, we don't have access to a machine running windows 10 (I think microsoft stopped supporting windows 10 last year). Are you implying that this part isn't compatible with windows 11?

So the host is Raspberry PI CM4 or CM5 ? We do have issue for for CM4 or CM5 hub driver.

Best

Brian

Currently we are testing with a windows 11 machine. However the final implementation will use a raspberry pi 4 model B (long term we may move to a compute module). Will the TUSB2036 not work with a rpi4B? If not, do you have a recommendation for an alternate part that will serve a similar purpose and has been confirmed to work with a rpi?

Otherwise, do you have any recommendations for how to get the TUSB2036 working now with a windows 11 machine?

Hi Chris:

for Windows system, did TUSB2036 hub detected by windows system, Do you have USBviewer tool to check?

Can you disconnect touch display from hub downstream port2? maybe something  wrong between hub and touch display.

Maybe we can have a call to discuss this issue.

Regards

Brian

Brian-

     When the windows system is connected to the hub, we get a "Device Descriptor Request Failed" error. Nothing is connected to the downstream ports. I'm assuming that we shouldn't see a "Device Descriptor Request Failed" error when nothing is connected to the downstream ports, right?

     Also, please let me know if there is an alternate hub IC you recommend for use with RPI-based systems. In the next system revision, we can move to a different part.

     Happy to hop on a call if that makes sense after reviewing the above. Currently I don't have the hardware in front of me, so hopefully we can continue the discussion here until next week.

-Chris

Hi Chris:

   I tested TUSB2046 on Windows 11 and works fine. TUSB2036 and TUSB2046 are very similar device..

  Can you send updated schematic for review again?

Best

Brian

Apologies for the delay. Not much is different in the schematic, I believe the text describes it well, but see below for the markups. R300, R301, R303 and R302 are not populated. The MODE pin is now pulled up to 3V3 with a 10k resistor. The /RESET line is routed out to our MCU for control over the startup reset pulse. Please let me know if you see any issues.

Also, please let me know if you have an alternate part recommendation that is better suited for use with a raspberry pi.

...

Can you try  reset  circuit below?

 Since it;s linux hub driver issue for CM4  or CM5 ., You may just buy a cheap UBS2 hub to try with CM4 or CM5  to  check if the bug is fixed  or not.

Best

Brian

We can give that circuit a shot, however we have tight timing control over the reset line using a microcontroller - I'm not sure what that circuit will offer that the MCU cannot.

I don't understand your suggestion about buying a cheap USB2 hub. We are looking for a hub solution to design in to the custom PCB in a product we are designing. It is not an option to integrate a cheap USB2 hub into this product. Can you recommend a different TI USB hub IC that is compatible with linux?

Christopher Phelan said:

We can give that circuit a shot, however we have tight timing control over the reset line using a microcontroller - I'm not sure what that circuit will offer that the MCU cannot.

just use RC for reset signal, no  control  needed from MCU.

Christopher Phelan said:

I don't understand your suggestion about buying a cheap USB2 hub. We are looking for a hub solution to design in to the custom PCB in a product we are designing. It is not an option to integrate a cheap USB2 hub into this product. Can you recommend a different TI USB hub IC that is compatible with linux?

I mean the driver issue  will affect all USB2 hubs. You can try with any USB2 hub you have on CM4 or Cm5 first. If no issue, then order TUSB2036EVM or TUSB4041 EVM to try to make sure it works 

Best

Brian

Understood regarding the RC usage, but I'm not sure why that will offer better performance than using tight reset pulse control with the MCU. Regardless, we can try it.

Regarding an alternate hub IC, is there something you recommend other than the TUSB2036 for use with a linux-based system? Do we need to move to a USB3.0 hub to (more likely) avoid the compatibility issue? If so, part recommendation?