Minimum load current from charger IC

In my current project setup I have a Li-Ion charger IC (the BQ25890 from TI) which manages the battery charging and discharging process. The charger IC has to power an ATmega328PB. The ATmega in turn controls a MOSFET to a high-current load in the 1-2A area. But most of the time, the ATmega sleeps in power-down mode, drawing only ~200nA of current. Even when it's active, it should not draw more than 0.1mA (+ the 1-2A of course) in average. To sum it up, a charger IC has to power a device which is always powered but draws minuscule current.

Now there's a problem, and this is why I ask.

I recently noticed a strange behaviour of my power bank, which cuts off current from a device with low current consumption. After reading a bit on the internet, this is not uncommon, and even a well-defined feature: The charger circuitry inside the power bank thinks it's leaking current and therefore cuts any small current off. With that in mind, I carefully checked the BQ25890's datasheet as I don't want such a feature for my MCU. However, I couldn't find nearly anything related to this besides the output voltage. These two values are suspicious:

These values are not further mentioned in the datasheet. If 30mA is the minimum current I have to draw, I would have to power the ATmega from a coin cell. Now these are my questions:

• Is my assumption correct the given values specify the minimum output current?
• Is my idea of using a coin cell generally better even if I can draw a small current out of the charger IC?

To give you a better understanding, I visualized my circuit configuration:

simulate this circuit – Schematic created using CircuitLab

• Which clock/grade?<=4MHz @ 1.8~5.5V or <10MHz @ 2.7~5.5.V – Tony Stewart EE75 Mar 16 '18 at 19:54
• 1MHz or even lower – Overblade Mar 16 '18 at 19:58
• Is there a charger side to this? I see 9.2.2 Device Power Up from Battery without Input Source If only battery is present and the voltage is above depletion threshold (V BAT_DPLZ ), the BATFET turns on and connects battery to system. The REGN LDO stays off to minimize the quiescent current – Tony Stewart EE75 Mar 16 '18 at 20:05
• The only issue I can see with original idea is if Learning Mode is enabled without started from 0% SoC. ti.com/lit/an/slua597/slua597.pdf – Tony Stewart EE75 Mar 16 '18 at 20:13
• Sometimes the battery gets recharged but I suppose the Power Path Management will power the small load. Looking at 9.1, SYS is directly connected to Vbus. – Overblade Mar 16 '18 at 20:51

The BQ24xxx/BQ25xxx family of PMICs don't have cut/off or shut/down artificial limitations. Unlike the "powerbanks" that must save the battery when no load, these ICs don't disconnect at low current, otherwise their use in mobile devices (as the primary power management and power source) would be a disaster.

I have devices based on BQ24296 and BQ25896 chips, and they don't have any cut-off limits. I am not sure what the ISYS(LOAD) is, the TI tool doesn't seem to have any description nor control bit for this, but the IBAT(LOAD) has a control bit7 in I2C register 03. When activated, the IC drains 16 mA from the connected battery. If not enabled (by default), the battery provides whatever voltage at no load.

The IC has an obvious disconnect when battery voltage drops below certain threshold, which is programmable, and set to 3 V by default.

So, it definitely looks like your concern about battery drop-off in MCU low-power mode is unfounded. The coin-cell backup might be a good idea though.

I think it would be a good idea to open a support case with TI and ask for clarification what these functions are about, or post a question on their discussion forum. At least the search for "Battery discharge load current" yielded no results other than links to the datasheets. It might be that these functions are for power bleeding in case of total system shutdown.