66AK2H12BAAWA2 - FAILS AFTER X-RAY

Hi,

Do you mean that the 66Ak2H12 SoC stops working after going through x-ray, or you refer to some other chip on your board?

How are you trying to boot your board?

Best Regards,
Yordan

Good Afternoon,

Here is the response I received from Engineering:

When the Keystone is initially progammed, a bootloader is flashed into on-chip NOR. Upon execution of this bootloader the first time, a second stage uBoot image is then flashed to off-chip memory which is used as the startup image from then on. Once the kernel is booted, one of the dsp core images is loaded via software. In some cases after the Keystone “failure” it is not possible to reprogram the NOR memory or even connect via JTAG.

Thanks,
Danette

Good Morning,

Is there any update or response to the last message. We just need to know how to proceed.

Hi,

I've started a discussion with the factory team. Their feedback will be posted here.

Best Regards,
Yordan

Hi Danette,

The 66AK2H12 does not have any programmable on-chip NOR so I'm assuming that you are referring to some external memory device installed on your board. Does the 66AK2H12 attempt to access the external memory and fail to read the device? 

Regards,

Bill

Danette,

X-ray for rework purposes should not damage the chip or board.  Are the boards repaired after the 66AK2H12 is replaced?  Why are they going through an x-ray if they are fully functional?  How many boards exhibit this behavior?  Is this 100% of the boards through the x-ray?

Tom

Thank you.

Thank you. I have forwarded this to the engineering group for a response.

I have forwarded this to the engineering group for a response.

Hello Bill,

This is what I got back from the engineer:

When the Keystone is initially programmed, a bootloader is flashed into on-chip memory. Upon execution of this bootloader the first time, a second stage uBoot image is then flashed to off-chip memory which is used as the startup image from then on. Once the kernel is booted, one of the dsp core images is loaded via software. In some cases after the Keystone “failure” it is not possible to reprogram the on-chip memory or even connect via JTAG.

Hello Bill,

Per the engineer:

We x-ray before test to confirm that process results are as expected.
We don’t know the operational status of the board at the time of x-ray inspection.

The removal process does not require repair, no the boards are not repaired.

Eight of twenty were found to be non-functional. We are not certain that X-ray is the cause of the component failure.
We know that replacing the part restores function.

Danette,

OK, so this is an early board bring-up scenario.  The team needs to establish robust operation through JTAG.  You are loading an intermediate boot loader into FLASH and then seeing problems.  Note that problems with the bootloader code can make ARM and DSP cores unresponsive, even through JTAG.  You might validate your IBL on the EVM to separate board debug from software debug.  Alternately, you should test your hardware with minimal programs that are loaded and executed from JTAG to make sure that you have robust operation of the process and access to the memories on the board.

Have you been able to recover a failed board with any method other than replacing the 66AK2H12 device?  I recommend that you change the bootmode to a bootmode that waits for JTAG to connect to see if this recovers a 'failed' board.  You should also verify that the power, clock and reset signals all meet the requirements and follow the mandatory sequence shown in the Data Manual.

Tom

Thank you, Tom. I am passing this on to Engineering.

Good Morning,

I have not seen anything since the last response. Are there any updates or additional information?

Danette,

We are waiting for a response from your team.

Tom

Thank you. Somehow I only was 2 out of 3 of the responses from 8/15. I will pass this on.

Hello. Per the engineer: "Many of the parts work fine. So far, for those that don’t, we have been able to get boards to run by changing them.
I can’t speak for the technical aspects of the boot loader and startup sequence.
But, it seems unlikely that the customer has not considered this."

I mentioned in my original post that we have samples that we can send back for analysis. Can we have the parts sent back rather than troubleshooting without the benefit of analyzing the actual parts?

Good Morning,

Would it be possible to send samples back for analyzing per my last response?

Danette,

The debug analysis has not established anything wrong with the 66AK2H12 devices.  The only results we have is that replacing the devices on your boards results in correct functionality.  The 66AK2H12 devices are being produced and shipped in high volume without yield issues.  I do not see value in returning devices for Failure Analysis until you have established that this was not something a simple as an assembly error.

Tom

Hello,

I have been instructed to ask if we can get a quote for a go/no go verification test on these parts. We do agree that there ma not be an issue with the parts specifically but a component test would be a good first step.