STM32L433 STOP2 current draw higher than expected

[Andre_k]

We are working on a project which requires us to put our STM32L433 MCU's into a low power mode with memory retention(STOP2 mode). When measuring the IDD current draw on our NUCLEO-L433RC-P  in STOP2 mode, we found that at 1.8V VDD the current consumption is as expected (<2uA). If we raise the voltage to 3.3V VDD, while still beeing in STOP2 mode, the current jumps up to ~400uA(should be <2.3uA).

All supply pins are connected together(except 1.2VDD), SMPS is disconnected from 1.2VDD, debugger is disconnected using jumpers. No peripherals have been explicitly enabled. All GPIO's except the wakeup interrupt pin are configured as analog inputs.

What can cause this jump(non-linear with voltage) in current draw when increasing the supply voltage from 1.8V to 3.3V?

Section in Datasheet:

IDD current draw(1.8V on the left, 3.3V on the right):

 

Jumper settings:

[Simon V.]

Hello,

Please take care of not back powering U7 LDO component via SB13 solder bridge or JP6 jumper on the Nucleo board.

It may explain the extra current consumption you are measuring. Otherwise, could you please share your solder bridge and jumper HW configuration (open/close)

Regards,

Simon

[Andre_k · Accepted solution]

Found the culprit! There are two level shifters on the upper half of the Nucleo U5 & U6. Removing these solved the issue. Now the current draw in STOP2 mode is 1.3uA at 3.3V VDD_MCU, just as expected.

Following changes were made on a fresh out of the box Nucleo:

• SB46 -> opened
• U5 & U6 -> removed
• SB10 & SB13 -> connected
• JP4 & JP8 -> removed
• JPs on CN2 -> removed
• R23 -> place 100R resistor

VDD_MCU rail powered externally over pin 2 of JP7.

Best regards,

André

[Andrew Neil]

Unfortunately, ST's boards (Nucleo, Discovery, et al) are notorious for having hidden leakage paths that will bite you when trying to explore the limits of really low power...
:face_with_rolling_eyes::frowning_face:>:(

(Well, perhaps I should say that ST's documentation is notorious omitting to mention leakage paths that will bite you when trying to explore the limits of really low power)

[Andre_k]

Yes, one would think that Nucleos with an STM32L series MCU are built for testing low power applications ¯\_(ツ)_/¯.

[Andrew Neil]

You'd have thought ...  :face_with_rolling_eyes:

EDIT:

Yes, I speak from bitter experience - with the B-L072Z-LRWAN1:

https://community.st.com/t5/wireless-mcu/excess-current-consumption-running-b-l072z-lrwan1-from-battery/m-p/322961

>:(