MSP430F5437A: Chip reliability due to design error - VCore connected to DVCC and AVCC

Is there a reasonable risk that the silicon of the MSP430 could become damaged from shorting VCore to the DVCC/AVCC pins and allowing operation?

Yes

You are exceeding the maximum specs.

Stresses beyond those listed under Absolute Maximum Ratings may cause permanent damage to the device.

Exposure to absolute-maximum-rated conditions for extended periods may affect device reliability.

Thank you, Peter, for your input.  Here is my concern:

The Absolute Maximum Ratings do not speak specifically to this issue, at least not in the typical measures for absolute max range limits. There is a footnote to the "Voltage applied to any pin (excluding Vcore): -0.3V to VCC+0.3V" line that states VCore should not be externally loaded or have voltage applied.  However, it does not state at which voltage or current loading permanent damage may actually occur to the device. It's just a footnote to the general pin under/over voltage limits.

The simple and convenient answer is to infer that there may be permanent damage at any level of DC loading or external voltage to VCore.  However, when it comes to determining the management of fielded products (e.g. recalls), it helps to be pretty confident that there is an actual risk before deciding on the solution.  For example, if the most likely risk was the the occasional malfunction of the supervisor/BOR circuits, then that's a lot different situation than the risk of hard IC failures.

I think the only way to really answer this specific question is with inside information on the device operation. As a first time poster on E2E, I'm not sure if this is the right venue to ask something like this.

Hi Brent

On second thought, your "unreliable startup/POR/BOR" may be related to the 1.8V supply.  

Try using 2.2V if possible

Brent, 

If using 2.2V to power supply, it is better to disconnect the Vcore from DVCC because of the operating specification of the Vcore. 

Thanks, 

Lixin 

Thank you, Lixin, for the analysis!

To resolve this design mistake, we are considering cutting the PCB trace that connects VCore to the 1.8V VCC. The end effect would be that the VCore pin was 'floating'. This would keep VCore from being driven by the 1.8V supply but it then leaves VCore without the recommended capacitance (470nF or 10:1 Vcc_cap/VCore_cap ratio).  I've tested 5 assemblies at room temperature with the VCore trace cut with no VCore external capacitance, and they boot 100%.  It appears far more reliable that way than with VCore shorted to Vcc but is still a significant risk for stable performance over temperature and other operating conditions.

We are currently investigating ways to rework the assemblies to add the proper capacitance on VCore.  Of course, the specific solution to that is outside the scope of this forum.

Peter - I think using 1.8V VCC should be fine in my design as long as VCore is connected properly.  Bumping to 2.2V Vcc would likely allow for higher clock rates but isn't viable given other design constraints on the existing assembly.  But thank you for providing some ideas to consider for a future respin of the design.

Hi, Brent, 

I am glad to see the test result on the 5 reworked boards shows far more reliable performance. I think it is right direction to cut off the PCB trace that connects Vcore to 1.8V VCC and try to add proper capacitor on Vcore trace.

if you think the issue is resolved, please press the resolved button to close this thread. If you have more questions, please let us know or you also can open a new thread in case you want to know something else. 

Best regards, 

Lixin