Brown-out and POR not 100% failsafe?

[M6]

I've noticed that our custom board sometimes does not work (e.g. execute the code) when the VCC drops to about 0.3V- 0.5V and comes back on again to 3.3V.

For testing I've made a simple program: It makes one GPIO an output and then toggles this GPIO in a while loop. This program is flashed at address 0, the soft device is not used.

I measure the VCC and toggling of the GPIO with a scope. See attached graph. Upper trace is the toggling of the GPIO. Lower trace is the VCC. After the second VCC dropout the GPIO isn't toggled anymore.

Nothing is attached on the GPIO and SWDIO/NRESET pins. The SWCLK is pulled down with a 12K resistor. Grounding SWCLK did not make a difference.

Device used is a NRF51822. I've tested this with the C0, G2 and H0 revisions.

Is this known behaviour or am I missing something?

[kchuva]

I have the same problem and the same behaviour on my nRF51822 H0 revision. S110 SoftDevice is enabled. When the VCC drops to about 0.5 V and slow comes back to 3.0 V, the brown-out reset is not happenning. So, I can't establish connection to BLE device.

If VCC return to 3.0 V rapidly, than brown-out reset device successfully and connection to BLE device is established.

I don't think when this happens the flash is corrupted, because after full power reset the firmware is started correctly.

Does brown-out must always reset the MCU when VCC drops below 1.7 V?

Are there any known problems with reset?

Can the slew rate of the voltage or the time spent at low levels affect on the reset system?