• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Subscribers 34 subscribers
  • Views 910 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

TUSB522P: Disconnect / Termination Behavior

Chris Nicol
Prodigy 75 points
Part Number: TUSB522P
Other Parts Discussed in Thread: SN65LVPE512,

Hello,

In our prototype design, we are switching our USB system (HUB, with downstream ports) on the upstream side, between multiple connectors. Basic block diagram is shown below.

Here the switch is actually multiple switches (for USB3 and USB2), but the control signals are shared (so USB2 and USB3 signals switch at the same time).

We have the ability to turn off power to downstream devices (turn off VBUS) and reset the HUB, which we do, prior to switching between upstream connectors. We have tried a variety of timings - everything is powered off / reset in the hundreds of milliseconds (i.e. much much longer than it takes to switch). Initial USB3 / USB2 transmit compliance testing was a pass.

Initial Results / Issues:

When we switch away, some hosts appear to not detect that the HUB / devices have disconnected electrically (sometimes), while other hosts exhibit no issue (example - no issues with a couple different macbooks running macos or windows 10, no issue with our chrome book, and a couple different laptops with windows 10).

The USB2 section of the HUB (and any USB2 devices) drop / re-appear as one might expect when switching away / back (no issue with USB2 at all, on any tested host).

When we have an issue, its always USB3. What we are seeing is that after first switch away - occasionally the USB3 portion of the HUB + any USB3 devices are phantoms - they show up in explorer / USBView, but we cant navigate to them (because they are not actually connected), and do not recover when we switch back.

Our LeCroy analyzer shows no "term off" when it fails in this fashion, and host traffic continues.

Probing the SS_RX (at re-driver, upstream / host side) after switching away (failed state), we can see that the host is continuing to poll.

In the failed (+switched away) state, if I short EN_RXD on the re-driver to GND, the phantom USB3 HUB + devices disappear from host immediately, and when I switch back everything is fine.

This is our first dive into USB3 (so excuse my ignorance) - but I assume the host is continuing to transmit (and "seeing" devices that are disconnected) because terminations remain applied at the re-driver (despite being disconnected on the downstream side).

Why would this be an issue for some hosts but not others?

How is this any different than me physically disconnecting my downstream USB system (i.e if the "switch" was a connector)? For example if the re-driver was built into the host instead.

What conditions must be met in order to trigger a proper disconnect to the host?

Anyone have advice / things to try?

Thanks,

Chris

  • 0
    TI__Guru**** 179886 points
    Chris

    As an experiment, would you please one 522P with SN65LVPE512 and see if you duplicate this issue?

    Thanks
    David
  • 0 in reply to David (ASIC) Liu
    Prodigy 75 points

    I will order some and try it out.

    For the record, we tried a Diodes Inc. part already which doesn't seem to have the issue. Trying to avoid a spin though, as there are a couple pin differences (had to cut 2 traces).

    This part you are suggesting looks pin compatible though, and would be an easier swap.

    I'll let you know how it goes.

    Do you have any thoughts on why this part might not show the symptom vs. the TUSB522P ?

    The part you are suggesting is NRND. What is the expected lifetime for it considering TI recommends TUSB522P instead?

    Thanks,

    Chris

  • 0 in reply to Chris Nicol
    Prodigy 75 points
    David,

    I have received the parts as per your suggestion - but haven't had time to try it yet... Do you have any responses to the questions in my last post? I was busy with holidays and now I am stuck doing EMC testing... so I might not have results for another week or so.

    Thanks,

    Chris
  • 0 in reply to Chris Nicol
    TI__Guru**** 179886 points
    Chris

    When TUSB522P was initially released, it was tested with existing hosts on the market and we do not see any interoperability issue. I believe the issue you are seeing is a timing issue between TUSB522P and some of the latest host release. We are aware of this issue and working to fix it. SN65LVPE512 is a different design and will not impacted by this timing issue and we do not plan to EOL this device right now. So as a stopgap solution, please use SN65LVPE512 until new TUSB522P is released.

    Thanks
    David
  • 0 in reply to David (ASIC) Liu
    Prodigy 75 points

    David,

    Thanks! I finally got around to testing the SN65LVPE512. They do NOT display the issue. I will discuss the use of these in our product with the team. Our current options are this, or the Diodes Inc. part I mentioned earlier. We have confirmed the Diodes Inc. part pin incompatibility is OK (NC pins can be safely grounded). Will be considering USB3 compliance results + cost + supply chain.

    Thanks for your help! I will mark this as resolved - but feel free to let me know when you think the timing issue has been resolved and we will re-test with the new component (TUSB522P).

    Chris

  • 0 in reply to Chris Nicol
    TI__Guru**** 179886 points
    Chris

    Thanks for your update, I will let you know the schedule of updated TUSB522P as soon as possible.

    Thanks
    David