brownout / watchdog reset on voltage drop

[rgrr2]

Hi,

as described here https://devzone.nordicsemi.com/f/nordic-q-a/37056/supervision-timeout-and-unplanned-disconnect our device has a hard current limit of 2mA.

Problem occurs if there is a BLE disconnect due to BLE master getting out of range.  This leads to a retry burst which exceeds the current limit.  To be tolerant against short peaks there is a capacitor bank on the device.  Retry bursts happens until supervision timeout hits (or Vcc drops ;-)).

If these retry bursts last long enough, Vcc begins to drop.

This unfortunately brings the device nRF52832 into an undefined state: watchdog does not occur and even worse, brownout reset also does not do its job.  Not sure, if brownout reset does not occur or if the reset logic does not reset the device to a low power state.  Measurement showed that Vcc is at appr. 1.5V.

Actually only powercycling helps out of this situation which of course is not acceptable.

Any help / suggestion is welcome.

Hardy

[haakonsh]

When the brown/out detection is triggered the SoC will short VDD to GND in order to drain the battery. It is a sort of kill switch that prevents spurious operation whenever a coin cell battery has been discharged. 

I suggest you increase the capacitance of the 4.7µF supply capcitor. 

Do you have any scope of VDD to share?

[rgrr2]

Is this "kill switch" somewhere documented?  Do I understand it correctly, that the kill switch shortens VDD to Gnd until a certain voltage level is reached?

I will do my best to provide a VDD plot tomorrow.

[haakonsh]

Brown-out reset.

The current consumption is probably equal to the CPU run current. 

The Brown-out detection is a hard limit in any design, you just need to make sure you do not dip below 1.7V and that your dip is not more than 300mV OR you have a slow enough rise time to adhere to:

"A step increase in supply voltage of 300 mV or more, with rise time of 300 ms or less, within the valid supply range, may result in a system reset" as per Power-on reset spec.

[mrono]

haakonsh said:

it holds the system in a reset state until power cycled.

So if you ever do get a brown-out event, how do you recover? How is the power cycling going to happen without user intervention?

[haakonsh]

It's designed to require user intervention, like the change of a coin cell battery. 

Do you mind sharing some more information on your power supply and your requirements? 

[rgrr2]

You are not serious about that?  User intervention!?  Aarrrggghhhh!

So our current limit provides enough power to keep the brownout detection alive and the nRF is stuck.

I doubt that brownout is implemented that way in e.g. MSP430.

Could someone please give me a pointer where this behaviour is described?

Thanks

Hardy

[mrono]

Note that I'm not the original poster, but I think this is an issue that is going go surprise a lot of people if it truly is designed the way you describe.

My requirements are simply that the device needs to recover *without user intervention* from a brown-out situation. Assuming of course that the cause of the brown-out goes away when the MCU resets, and Vdd rises back to an acceptable level.

[cbd]

You're really pushing the limits if your maximum power draw is 2mA. As noted in the specification and you've seen in practice you require much more than that when BLE is active, hence your capacitor bank.

The system will hold in reset until the internals have stabilised, but I would suggest that you look into a means of holding off powering up your 52832 until after the power bank caps have been fully charged.

Are you only enabling the SoftDevice once you've ascertained that your power supply is of sufficient voltage? With a direct coin cell connection I'd be wary of enabling it if the supply voltage was 2.5V or less as the dip in voltage when BLE transmits can reset the unit. You'll need to calculate/simulate your power cycles with your capacitor bank included.

Have you been using the power fail comparator to help detect insufficient power and disable the BLE?

It strikes me that your support circuitry will need to provide the "user intervention" in the case of brownout reset. Sadly such circuitry will eat into your power budget.

[haakonsh]

CBD is correct, the power supply does not fit the use-case. There are a few ways to mitigate the issue, like CBD has already mentioned.

I suggest you take a look at TI's TPS3898 and similar power supply supervisors/reset chips and calculate the required size of your capacitor bank.