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.

AM3352: DDR3 initalization

Part Number: AM3352

Hello,

I have a very strange problem with a custom board. The board is very similar to beagle black  with AM3352 and a single ddr3 memory. I have two identical verions of the board: one with micron MT41J128M16-125 and one with samsung K4B4G1646D-BCK0. I can't initialize the ddr3 correctly in both boards. After initialization, if I write 0xFFFFFFFF I read 0xDFFFDFFF and if I write 0xAA55AA55 I get 0x8A558A55. I have tried all the possible consigurations in u-boot SPL and in starterware in UART boot mode. It seems that the same bit in both byte lanes have problem. How is this possible? Does it look like a timming configuration problem, or a pcb trace routing problem?

I have read all the relative TI's support documents and I have used all the spreadsheet tools for calculating parameters.

Can anybody help me with this problem?

Thanks.

  • There seems to be something wrong with the DRAM data line D14.

    Check it.

  • - Did you also perform SW leveling in addition to calculating the timing registers? If so, can you share all the relevant details?
    - Did you follow all the routing guidelines as in the data sheet? Can you check for any compliance issues vs. data sheet?
    - Can you report your trace length for DDR?
    - Do you have schematics that you can share? Did you do any bit swapping? If so, is the swapping within the related byte?
    - Is this occurring on multiple boards or a single board? Did you check if there are any issues in the layout/PCB?
  • First of all, thanks for the reply. I will post tomorrow all the necessary information (trace lengths, schematics and PCB layout). Software leveling I tried to do with a tool from chinaTI, because the board doesn't have jtag. I think that the tool failed to do leveling because all the results were zeros. I didn't do any bit swapping. How can I do it? Exactly the same happens in two boards, one with micron and one with Samsung memory.

  • Hi to all. Please ignore the previous posts. The problem was only in bit D13. I checked the board with the microscope and I found that there is no hole in D13 via at the processor side!!!. Probably the pcb manufacturer didn't do the job right. Does anybody knows any way to rework this board?

    Thanks