P0.25 and P0.26 more likely than other pins to oscillate if left floating?

[epower]

I have two products which are very similar. They differ in which pins are mapped to which functions. Let's call the Product A and Product B.

I made a mistake and left two of the pins as floating inputs. This led to increased current consumption in System OFF mode on about 5% of the units of Product A, but none of Product B.

On product A the floating pins were P0.25 and P0.26. On Product B they were P0.1 and P0.3.

The correlation between current consumption and pins 25 and 26 led me to wonder if the issue in errata 138 (https://infocenter.nordicsemi.com/topic/errata_nRF52832_Rev2/ERR/nRF52832/Rev2/latest/anomaly_832_138.html#anomaly_832_138) makes these pins more susceptible to oscillation than other pins if they are floating.

When designing the board I overlooked Errata 138 so I do not have capacitors on P0.25 or P0.26.

The issue has been corrected and I'm no longer leaving any pins as floating inputs, but I am curious. Could the difference I saw between Product A and Product B be due to the issue behind Errata 138?

[mathiaso]

Hi, Evan.

No, Errata 138 should not lead to increased current consumption in System OFF.

Best regards,
Mathias

[epower]

Thanks Mathias,

Because I did see these pins in particular cause increased current draw in System OFF mode when they were floating, do you think I should be concerned about the integrity of these particular parts?

[mathiaso]

What numbers kind of numbers are we talking about for the increased current draw?

And now you don't see the increased current consumption in any of the devices anymore, right?

[epower]

That's correct. I'm not seeing increased current draw anymore now that I'm not leaving any pins floating. The amount of current varied from device to device, but was as high as 50uA in the worst cases.

[mathiaso]

Ok. That was quite an increase. And that only was present on 5 % of the devices with P0.25 and P0.26 floating? Did you try resoldering any of those?

I'm happy to hear it is solved now, at least.