Low Power E5 Development Kit (How to get uA level STOP2 consumption)?

[LiquidPlasmas]

Hi All,

I can’t for the life of me figure out how to get the Lora E5 Dev Kit board to achieve the low power consumption I’m expecting when sending the chip into low power STOP2 mode.

I am developing a custom STM32Cube application for the devkit board based on the example code and wiki instructions and powering the dev kit board via the USB-C port.

Instead of expected 10s of uA I am always seeing a minimum current consumption of about 4-5 mA while in STOP2, even when using the highly simplified LowPower example project or a new project with no chip configuration other than going into STOP2 at boot.

One clue I have noticed, consumption is a bit lower on my meter when connecting the board via a USB 2.0 micro-B cable than when connected via a USB 3.1 C cable.

At first I thought maybe the built in radio is not being disabled in STOP2 (the SX126X) but even when asserting LOW on PA4 and PA5 RF control pins there is no change. To minimize consumption I have removed the 5V and 3.3V enable jumpers to ensure there is no power to the onboard RS485 transceiver and IIC temperature sensor (though this had no measurable effect on consumption). Connecting or disconnecting the ST-Link debugger has a small effect, a few 10s of uA but still board consumption is over 4mA overall.

Maybe the CP2102N USB->TTL chip is consuming this power somehow even though there is no active serial connection (I am seeing this even when connecting via a cable with no D+/D- data lines). If so, it’s confusing because the low power example in the Wiki shows the E5 Mini board connected via USB through a meter achieving <1mA power levels, and that board also uses the CP2102N.

Any ideas? Thanks for any help!

[LiquidPlasmas]

Quick update, this does appear to somehow have to do with the current consumption of the CP2102N chip, even when D+/D- are disconnected and UART is not initialized (pins are high-Z).

When connecting to the battery terminal J5, power consumption while in STOP2 is less than 10uA. ~125mA when transmitting and ~4-5mA when receiving, exactly as expected!

From the CP2102N datasheet, it seems this chip should consume about 195uA when in USB suspend but may consume up to 9mA at 115200 baud or up to 14mA at 3 Mbaud, fairly close to what I am seeing. Not sure why the chip wouldn’t be in USB suspend state when no data lines are present, but will need to dig into the datasheet more. This still doesn’t explain how the E5 mini wiki example can achieve <1mA power consumption when powering through the CP2102N chip.

[LiquidPlasmas]

Sure enough if I put the host PC to sleep so that the suspend signal sequence is sent on the USB bus the board power consumption drops to about 350uA, which is pretty close to what would be expected when the CP2102N is in SUSPEND state rather than normal operation.