MPC8536 operated DDR3 wrong by CW memory test

[joezhang]

Hi，

Our customer use Codewarrior to debug MPC8536(with 4 pieces DDR3), they found there were 3 pieces DDR were normal, but 1 piece DDR's low 16 bits data value always was 00. Customer wrote data in this DDR's any address, then there were only high 16 bits changed, the low 16 bits still were 00. Customer has 2 pieces PCBA and they are the same status. And the circuit design is from a old design which was normal.

In the picture, the "0000BEEF" were wrong data, and wrote these data by a value which low 16 bits was Non-zero again, found the low 16 bits still were 00.

Please help the check it, thanks a lot!

[yipingwang]

Hello Joe Zhang,

Have you configured DDR controller in the CW initialization file according to the specified type DDR?

Are these 4 pieces DDR3 the same type?

Have a great day,
Yiping

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

[joezhang]

Dear Yiping，

Yes, Customer configured DDR controller in the CW initialization file according to the specified type DDR, and the 4 pieces of DDR were same.

In customer’s old projects, they use MPC8536 with 4 pieces of the same type DDR, and system could work normal.

This time, the circuit design for MPC8536 is same with those old projects, except the PCB design.

If need more information, I can mail to customer for the more detail.

Thanks a lot!

Best Regards!

Joe Zhang（张秋瑞）

Mob: +86-18611045020

Email: joezhang@comtech.com.cn<mailto:joezhang@comtech.com.cn>

发件人: Yiping Wang

发送时间: 2016年11月17日 13:58

收件人: Joe Zhang <joezhang@comtech.com.cn>

主题: You have been mentioned by Yiping Wang in Re: MPC8536 operated DDR3 wrong by CW memory test in NXP Community

NXP Community <https://community.freescale.com/resources/statics/1000/35400-NXP-Community-Email-banner-600x75.jpg>

You have been mentioned

by Yiping Wang<https://community.nxp.com/people/yipingwang?et=notification.mention> in Re: MPC8536 operated DDR3 wrong by CW memory test in NXP Community - View Yiping Wang's reference to you<https://community.nxp.com/message/852189?commentID=852189&et=notification.mention#comment-852189>

[yipingwang]

Hello Joe Zhang,

I suggest you use CodeWarrior to do more hardware memory test, you could test "walking 1's" provided in Hardware Diagnostics panel.

Walking Ones:

Address Line: The board or chip address lines are shorting or stuck at 0 or 1. Either condition could result in errors when the hardware reads and writes to the memory location. Because this error occurs on an address line, the data may end up in the wrong location on a write operation, or the hardware may access the wrong data on a read operation.
• Data Line: The board or chip data lines are shorting or stuck at 0 or 1. Either condition could result in corrupted values as the hardware transfers data to or from memory.
• Retention: The contents of a memory location change over time. The effect is that the memory fails to retain its contents over time.

This problem should be caused by the hardware problem on the target board, you could create a service request from Sales and Support|NXP  to let the hardware support team to do more detection for you.

Have a great day,
Yiping

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