i.MX6 Plus - USB is unstable issue

[yen-polin]

Hi ,

    We've implement i.MX6 Dual plus/Quad plus and Yocto 1.8 .

    USB port of HOST/OTG are unstable at both Dual plus and Quad plus .

    If we try to do "time dd if=/dev/sda1 of=/dev/null bs=8k" , then sometimes show the message as below

         ==> "usb 1-1: reset high-speed USB device number 2 using ci_hdrc" .

         ==> "Buffer I/O error on device sda1"

         ==> "usb 1-1: device firmware changed"

    So we've measure the USB signals and power , the report as attached file looks fine .

    Also we try to reduce USB trace length (under 1200mils) , but still can't fix this issue .

    So

       (1) Does it have any power rail for internal USB logic , we need to double confirm ?

       (2) Do you have any suggestion about USB Test item (EL_xx) we need to check , if we'll measure Hi-Speed USB signals again ?

       (3) Does it have any register cause these wrong message ? Could we try to disable it ?

       (4) Other recommend ?

    Thanks a lot.

Terry

Original Attachment has been moved to: 1.04.1.02.1.rar

[yen-polin]

Hi ,

       I've change the value of address: 0x021b0400 to 0x14420000 from 0x514201f0 ,then it can fix this USB issue .

       Address : 21B_0400 is MMDC Core AXI Reordering Control Register .

              Thanks a lot.

[Yuri]

Hello,

The messages from the log may be concerned with hardware

inaccuracy, for example, in PCB design. Please double check

USB layout and 24MHz crystal accuracy, described in “Hardware
Development Guide for i.MX 6Quad, 6Dual, 6DualLite, 6Solo Families …”

< http://cache.nxp.com/files/32bit/doc/user_guide/IMX6DQ6SDLHDG.pdf >

  According to EB810 Migration Guide, there are no significant difference

between USB modules of i.MX6 D/Q and i.MX6 DP/QP.

< http://cache.nxp.com/files/32bit/doc/eng_bulletin/EB810.pdf >

  As 24 MHz accuracy : please look at the following comments of Sergii

in the Community discussion :

“At some of our devices, that now at mass production, we found that BGA and crystal was

mounted not so good, and it was additional capacitance between iMX28 package and PCB.

As a result - frequency deviation from 24MHz was more than 150ppm. I recommend you to

check, that you have frequency at crystal inside the range 24MHz ±50ppm (you can do that with

spectrum analyzer or with pulse counter. Oscilloscope usually will not have needed accuracy).”

“USB Attach Failure i.MX28 Windows CE 6.0 Memory Stick”

< https://community.freescale.com/message/618741#comment-618741 >

Also You may create request in order to check schematic.

"How to submit a new question for NXP Support"

< https://community.freescale.com/docs/DOC-329745 >

Have a great day,
Yuri

-----------------------------------------------------------------------------------------------------------------------
Note: If this post answers your question, please click the Correct Answer button. Thank you!
-----------------------------------------------------------------------------------------------------------------------

[yen-polin]

Hi ,

     Thanks for your reply .

     1. Layout check :

           (1) I've double check the layout of USB pair , it's 90ohm impedance / reference GND plane only / Length matching within 2mil

           (3) No route USB under oscillators or parallel to clock traces and/or data buses

           (2) And I've try to reduce the trace length of USB signal under 1200mil , but still can't fix this issue .

           (3) In original desing , USB Eye pattern is Pass.

     2. Crystal 24MHz check :

           Crystal : 24M +-10ppm be implemented in our project , and below is measured report , it's fine.

          

     3. Do you have other suggestion ?

     Another Question :

         In iMX6QP-SABRESDB-DESIGNFILES reference board as below :

             (1) The ball out of USB are close crystal's ball out (24M & 32.768K) in breakout area .

             (2) 32.768k crystal have one via

         Does it have any risk in reference board ? Do you have any recommend about the spacing between "USB signal" and "the Crystal's trace" ?

     Thanks a lot.