Battery monitoring in ultra-low-power applications

[tommys]

Hello everyone.

Has anyone tried to measure the battery level in ultra-low power devices? What should I pay attention to?

I saw that there are some battery measuring devices (fuel gauges) - would they be suitable for low power devices?

Maybe it would be better to design measuring system with operational amplifiers?

Thanks for the answers!

[BadeBhaiya]

I use nRF Power Profiler Kit. It can measure down to 100nA and record 100kS/s with a few different current ranges. It works well enough and it is 99$, much less expensive than other products in the market

https://www.nordicsemi.com/Products/Development-hardware/Power-Profiler-Kit-2

EDIT: Sorry I misread the question, you can use the power profiler to profile the device but not the battery. For profiling the battery, you may use Otii but I don't think there are many products which can profile batteries.

[radiolistener]

You're needs for a MOSFET key to enable battery voltage divider which is used to measure the battery voltage. So you can enable that key very rarely - just for battery measurement and disable it for the rest of time. In such way you can avoid fast battery discharge through battery measurement circuit.

Maybe it would be better to design measuring system with operational amplifiers?

No, that is useless and lead to additional battery discharge current for power supply operational amplifiers. Most of modern microcontrollers have 12 bit ADC, which is good enough to measure battery voltage from voltage divider with a very good precision and no needs for DC offset.

What is important is to take care to disable power supply for measurement circuit between battery measurement interval. Without that key it will lead to a pretty significant battery discharge.

For example ultra low power microcontroller in a deep sleep mode eats current which is much less than 1 nA, but if you keep voltage divider used for battery measurement powered all the time, it can easily eat 10-100 uA or even more which is 100000 times higher than microcontroller current consumption.

[Peabody]

Below is the kind of circuit @radiolistener is referring to.  The mosfet should be logic level, but something like a 2N7000 would probably work.  The circuit assumes the vattery voltage is higher than the processor's Vcc.  Anyway, the idea is that you switch on the voltage divider only for the millisecond or less that it takes to read the voltage.