• State Resolved
  • Locked Locked
  • Replies 7 replies
  • Subscribers 98 subscribers
  • Views 389 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.

Hawkboard: Where has the layout gone wrong?

Anonymous
Anonymous
Other Parts Discussed in Thread: TPS65070, AM1808, TVP5147

Hi,

I read from Hawkboard Issues - TI Annoucement that there are instability with Hawkboard design and OMAP-L13x Marketing Manager Huy Pham attributed the cause to Hawkboard's not following TI layout recommendation.

Since I am doing a new design taking EVMAM1808 as a reference, I'd like to learn some lessons from the faulty design of Hawkboard so that I could take precautions. To be more specifically, not following which layout recommendation has caused the problem? Is it

1.    DDR2

2.    Powering, TPS65070, split planes

or others?

 

And is there any fault analysis provided for viewing?

 

Zheng

 

 

 

  • 0
    TI__Mastermind 27585 points

    The DDR2 layout violates the routing rules for the interface in the datasheet. That is the critical issue which let to DDR failures on the board.

    In a week or two we can provide updated layout files which have addressed the issues with that board.

    Jeff

  • Anonymous
    0 Anonymous in reply to jc-ti

    jc-ti,

    I got it, thanks.

    Is there care needs to be taken specifically with internal power planes (and split sub-planes on each of the internal planes)?

     

    Zheng

  • 0 in reply to Anonymous
    TI__Mastermind 27585 points

    In general it is best to avoid the split power and ground planes. Hawkboard seemed to use a separate ground for almost every interface which is not good practice and wasn't done well. I would recommend using a single common ground unless you have a specific need and know what you're doing.

    The next revision of this board will not use split planes.

    Jeff

  • Anonymous
    0 Anonymous in reply to jc-ti

    jc-ti,

    jc-ti said:

    In general it is best to avoid the split power and ground planes.

    I have rarely seen split ground planes, but split power planes seems to exist on virtually all boards? 

    jc-ti said:

    The next revision of this board will not use split planes.

    Do you mean next revision of Hawkboard does not use split plane for power planes? But for AM1808/OMAP L138 which is powered by TPS65070, only TPS65070 alone has five different voltage outputs (3 DCtoDC converter, 2 LDOs). It seems unreasonable for there to be 5 power planes.

    jc-ti said:

    Hawkboard seemed to use a separate ground for almost every interface which is not good practice and wasn't done well.

    Same as above. I haven't seen split ground planes before.

    jc-ti said:

    I would recommend using a single common ground unless you have a specific need and know what you're doing.

    Agree; but what about the questions for split power plane above?

     

    Zheng

  • 0 in reply to Anonymous
    TI__Mastermind 27585 points

    Zheng, what I meant by split power planes was that for a single voltage rail (1.8V for example), some boards will power components with a filtered version of that, such as the VCC_1V8ADDR found on Hawkboard. Of course different I/O rails are required, so I wasn't referring to that.

    Some components, such as TVP5147 on Hawkboard, give the option of using analog grounds for some pins. In most cases, it is best to use the digital ground, which will avoid many of the issues that can come about if the split ground is not done properly.

    Jeff

  • 0 in reply to jc-ti
    Expert 1935 points

    I also noticed that the DDR and OMAP have separate reference voltages - should these not be tied together?  Further, can't the OMAP actually be set to *generate* this voltage, obviating the resistors and all but one bypass cap to ground?

  • 0 in reply to bowerymarc
    TI__Mastermind 27585 points

    Yes there should only be one DDR_VREF generated. For this device the OMAP cannot generate the voltage and the resistor divider is required.

    Jeff