nPM1100 - Current consumption is higher than expected

[dossalab]

Hi all,

I'm working on a custom design based on nPM1100 and nRF52832. It's a battery powered device that's intended to be on at all time, so I'm taking some care about current consumption.
I've assembled a simplified version of the board with just PMIC, MCU and required passive components.

The chip is running an empty loop with WFE instruction, on a standalone board I'm measuring about 30 uA with that setup and I'm quite happy with it.
However, on my new board I'm measuring about 2200 uA of current, and that's very off.

The only guess I have is VSYS decoupling, which is just missed on this board (since I'm not using VSYS at all). But I'm not sure if that can lead to such issue. It somewhat makes sense - currently there is no battery-side DC-DC decoupling at all and I guess this can lead to improper DC-DC operation.

This is the relevant part of the schematics and a part of the PCB with the chip.

Can you confirm if my initial guess is correct or there is anything else that I can check?
Thanks!

[Kenneth]

Hello,

I can't think of anything else either. Likely the regulator is not stable here.

Kenneth

[dossalab]

Since this board does not allow me to install that capacitor I've temporary switched to another board with slightly inferior layout but easier access to traces. So that's what I have so far:



All caps are X5R 10uF 6.3V (C1608X5R0J106M080AB). R12 is charge current selection, R1 is NTC stub. L1 is the main coil (DFE201210U)

The consumption is still around 2200 uA...

I'm not really sure where to look next. The chip is hand-soldered but I doubt something is shorted. I got pretty stable results with both lead-free and reballed leaded solder (with stable output voltage and working charger and indications)

There are no other components on this board besides the battery.

[dossalab]

I tried that and it doesn't seem to make a difference

[dossalab]

one more thing that I observed is that consumption seems to depend on battery level. I charged the battery overnight and now the chip is driving more (2.8 ma)

[Kenneth]

Do you have access to an nPM1100-EK so you can measure on and try for comparison?
https://www.nordicsemi.com/Products/Development-hardware/nPM1100-EK 

Unfortunately I am not able to find anything that can explain your behaviour, can you provide a measurement table with some voltage levels and the currents you measure?

Kenneth

[dossalab]

Sorry for the late response - I tried to run some tests in the meantime and I think I might've found something.

I was measuring the current with a regular handheld multimeter, which (as it turns out) adds a very high shunt resistance while connected (something about 10 - 100 ohms, not sure exactly). This in turn seems to confuse the chip in various ways (for example it never starts the charger when the meter is in the circuit). That makes sense since that kind of shunt will create a rather large voltage drop which will make it impossible for the chip to monitor the charging process.

I used that setup because I already tried it before on simpler chips (i.e DC-DC converters) and it seemed to work fine.

So I need to find some better gear to continue measurements. I unfortunately do not have DK on my hands, so I can't confirm if it behaves the same in this setup.

I wonder what method of measuring power consumption is recommended? I know about PPK but I don't have it either. It was not obvious for me that a regular meter won't make a sensible measurement at such low currents.

[Kenneth]

Hello,

I suggest to order an PPK2 yes:
https://www.digikey.com/en/products/detail/nordic-semiconductor-asa/NRF-PPK2/13557476

Typically the in-rush current can be both rather high and short, because it can operate in mode that only turn on the regulator for a short periods of time (to save current), I can imagine that a resistor value of 10-100ohm can be problematic in that case (due to the voltage drop).

Kenneth

[dossalab]

I did some more measurements - this time using a separate 10 ohm shunt. It's a pretty rough measurement, but I finally see a proper consumption in the range of 10 uA without the load and 30-40 uA with the powered MCU.

So yes, using a regular multimeter is definitely not an option with such designs. Even with a proper setup that's still an interesting thing to consider - especially with lower-tier batteries with high internal resistance.

Thanks for your help and assistance!