Non-obvious power problem

[John]

Hello,

This is less a question and more of a survey. We've recently completed development on a project using the nRF51822 but have run into an odd power problem. We built and tested the first 500 or so units. Our design has the unit in System Off mode while they are in stock or not in use by the customer. However, after a couple of weeks 5 units in stock were found to have completely depleted 2 AAA batteries.

We've gone through all of the usual suspects such as floating pins, misuse of peripherals and most anything we could search for in this forum. So, I thought I would post and see if anyone had experiences with odd power bugs using the nRF51 that they would share to give me additional ideas to try.

We are using the nRF51 as our main processor on this product, and it is running the FreeRTOS port using tickless idle. When we are in our "storage mode" as we call our system off behavior while in stock, the board typically burns well under 1uA. While in stock the units are boxed and the only GPIO that could wake the device back up is a button push which should be impossible.

Thanks in advance for any ideas or just good old war stories.

Regards, John

[Stefan Birnir Sverrisson]

In system off, it is much simpler to get low current consumption, as entering system off disables all peripherals and the only way to wake up from system off is with a reset, assert on sense enabled pin, or with low power comparator. In the nRF51 Series Reference manual 3.0, section 11.1.2, it says:

If the device is in System OFF mode, entering into debug interface mode will generate a wakeup.

So if you are in system off and there is noise on SWDCLK pin, it can cause the chip to wake up and enter into debug mode with ~1mA current consumption. That is the most likely explanation I would think, as the battery is depleted in two weeks. So I would suggest to put 470ohm external pulldown resistor on SWDCLK pin

[Moritz]

Could this (entering debug mode unintentionally) also happen on a nrf52832? Looking into the NRF52-Product Spec at Debug and Trace it seams unlikely with the internal pullups/pulldowns. Any way we have experienced something quite similar in our preproduction run. 3 of 100 devices had a completely empty battery after 2 weeks. What I found out in the meantime is that we compiled the software with CONFIG_GPIO_AS_PINRESET, but have left the reset pin floating. Could this cause a higher current consumption in any way?