• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 96 subscribers
  • Views 506 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.

PROCESSOR-SDK-AM335X: USB device disconnect timing

Ian Holmes
Prodigy 60 points

Part Number: PROCESSOR-SDK-AM335X

I'm looking at an issue with the BBB in a system which is being tested against resilience to fast transients.

At the hardware level I see the processor release control of the DP and DN lines and at the software level (Linux, recent build) I see a device disconnect. I see this with various different devices, FULL SPEED and LOW SPEED plugged into the BBB HOST PORT. I also see "invalid OTG device" messages (strange as this port can only be a HOST due to USB1_ID being tied low.

Using a high speed multichannel scope and taking noise precautions I have been able to record the timing of the transients and verify SE0 events on the line, however none of these meet the USB 2.0 FULL/LOW SPEED specification timing of SE0> 2us<2ms (see USB 20.pdf spec para 7.1.7.3 available from USB org for details).

I see an apparent response to a disconnect to unstable 200ns signals.

My question is this : Where is the 2us SE0 timer implemented in the AM333X - is it supposed to be in software in response to the disconnect interrupt, or is this in the PHY (possibly with a settable timer). I've not been able to locate this information myself so I would appreciate any help.

This might also be a HIGH SPEED disconnect signal (which shouldn't be enabled as we have only a Low or Full speed device attached) where the differential voltage is sampled instead.

thanks for any pointers

rgds

Ian

  • 0
    TI__Guru**** 393215 points
    The factory team have been notified. They will respond here.
  • 0
    TI__Mastermind 27890 points
    Ian,
    This disconnect timing is handled by in hardware by the USB PHY, which notifies the controller via the UTMI+ bus, which then notifies the CPU via an interrupt. Although the controller is immediately aware that there has been a disconnect event, what happens next (and when) is a function of the OS in use and the loading state of the CPU.

    Disconnect signalling is handled differently for LF/FS and HS. LS/FS connections use the UTMI+ LineState[1:0] signals to report the single-ended values of the DP/DM pair, while the UTMI+ signal HostDisconnect is used for HS connections. Both LineState and HostDisconnect only have context in their respective connection types, so you are correct in that only one method is enabled at any time.
  • 0 in reply to -DK-
    Prodigy 60 points
    Perfect - thankyou