nRF5832 Continuous reset in hardware

[nhazzard]

Hi,

We are using the nRF52832 in a battery such that we have a 3V3 LDO configured to turn off under a certain battery voltage.  Now during testing we have discovered that the voltage supervisor circuit which enables or disables the LDO to turn ON or OFF can oscillate back and forth with very small voltage changes.  Basically, when the LDO is switched off, the battery voltage can recover slightly causing the LDO to continuously oscillate ON and OFF.

Ideally the circuit turns OFF once and has sufficient hysteresis, but I am wondering if this harmful for the nRF.  Is there is a situation where it can lock up or cause adverse effects that are not normally recoverable if the voltage rises to 3v3 or under and then immediately drops to 0V?  Is there a min voltage it must rise to before dropping back down to 0V?  We may not be able to control the timing of the ON or OFF duration in this scenario.

[Kenneth]

Hello,

Do you have any traces to share? So I can see a bit on a timescale what fluctuations you are talking about? Does it happen during normal operation or only on startup?

In general it's not a problem, other than you can see sporadic resets, and also, depending on when the reset occurs relative to potential flash erase/write operations, this may cause the application to not be very "happy" on startup if it was an partial erase/write.

Kenneth

[nhazzard]

Hi, this happens continuously only when the battery voltage is very low.  I don't have an image to share but the Vdd rail is basically cycling between 0v and 3v3 every 250ms.  So it looks like a square wave.  The rise and fall times are under 10ms.

Do you see any issue with this constant power cycling?  I assume 250ms is plenty for the firmware to do any initialization.  And the rise/fall times of Vdd are within spec.

[Kenneth]

I don't have further details than I have already provided. If your application just blinking an LED, then this will run as expected every time (maybe that answers your question) as long as it falls to 0V between each try, my main concern is related what the application may be doing, in specific if it is possible that you end up in a situation with partial flash operations. One might argue this can always be the case also when users replace batteries, and that is true, but you are doing it (seemingly) with very precise timing for multiple products, so that make the question a bit more relevant. If it happens only when battery is low, maybe you can measure the power supply voltage on startup and go right to sleep if the battery is low.

Kenneth