Strange hardware failure with looping output audio

Hi Peter,

We understand that you are facing issue in SPI slave booting in ADSP-21569. Please confirm whether our understanding is correct.

Please provide your comments on below questions to narrow down the issues

1. Please let us know the silicon revision of both working and non-working board.
2. How many boards have you tested and how many of them shows this issue?
3. Can you please send us the fault ratio?
4. Please share more information about the Host.
5. Are you using INIT code while creating the loader file?
6. Is it possible for you to test with simple LED blink application and check whether the issue persist?
7. Did you already aware of the application note(EE-418) which talks about "ADSP-2156x Board Design Guidelines for Dynamic Memory Controller". If not please refer the below link and check whether it helps.
https://www.analog.com/media/en/technical-documentation/application-notes/ee418v02.pdf

Regards,
Nandini C

Hello Nandini, thank you for you fast reply

Answers to your questions:

1. Please let us know the silicon revision of both working and non-working board.
The part markings seem identical see the attached photos

2. How many boards have you tested and how many of them shows this issue?
We have built 33 boards, and 3 so far have failed with this issue.
Note the 3 that failed initially are working ok, and failed after some use.

3. Can you please send us the fault ratio?
3/33 = 9% failure rate in 3 months of use.

4. Please share more information about the Host.
The host is a TI ARM CPU running Linux and communicating to the '569 over SPI.

5. Are you using INIT code while creating the loader file?
Yes, we are using the 2156x_init project.

6. Is it possible for you to test with simple LED blink application and check whether the issue persist?
The simple LED blinking example application seems to run normally, it is only our audio application using SigmaStudio+ that seems to have the issue on these 3 boards.
By the way, is there available any kind of DDR3 validation example app for '569?
Something that similar to memtester in Linux that would test the full DDR chip for read/write operations.

7. Did you already aware of the application note(EE-418) which talks about "ADSP-2156x Board Design Guidelines for Dynamic Memory Controller". If not please refer the below link and check whether it helps.
www.analog.com/.../ee418v02.pdf

Unfortunately, the engineer who designed the DDR I/F for this board is no longer at the company, however a quick look at the design and it seems to follow the guidelines well.
The DDR3L is a common Micron part MT41K64M16TW-107:J run at 1.39V
There is a 0.69V VTT termination LDO power supply using TPS51206.
All address and control lines are terminated there through 47 ohm resistor.
Trace lengths seem short and well matched.

Anyway, each of the boards did work for a while but then failed, so seems unlikely it is a design issue on the DDR.

Thanks!!

Hi Peter,

Can you please refer the DDR sweep test, which is in POST example of 21569. We suggest you to take this code as a reference and modify it according to your external memory configuration.

Please find the below link for the ADSP-2156x Board support package.
download.analog.com/.../ADI_EV-2156x_EZ-KIT-Rel3.0.0.exe

Once Installed, you can find the example from the below mentioned path.
[Installation path]: \Analog Devices\EV-2156x_EZ-KIT-Rel3.0.0\EV-2156x_EZ-KIT\Examples\Power_On_Self_Test

Regards,
Nandini C

In the end, seems we are able to resolve this issue by reflowing solder on the DSP.

So seems it is some kind of cracked solder issue.  If there is any special application note on soldering of the BGA package for this part, let me know.  But we can mark this issue as resolved.