Power consumption in Standby mode a little too high

ramses946 · ‎2023-07-09

Hello,

Following setup: STM32L431 MCU with 64 pins TQFP

The board I'm using is a minimally soldered version because I could not reach the stated figures in standby mode from the datasheet.

Necessary capacitors are soldered + 1x 22uF 0805 + a 32kHz crystal with capacitors
VCC: 1.84V
ambient temperature: 27.5°C

The datasheets claims a typical consumption of 27.7nA in standby mode at 25°C without RTC + 173nA if SRAM2 is retained. So that's a total of 201 nA.

I'm measuring a stable 277 nA (after 20 minutes so that any capacitor leakage currents are minimized).

Where does the excess current of ~75 nA come from??? First I had 5 pins pulled up in standby mode and was measuring 280nA. I suspected that maybe some GPIO leakage currents are contributing to the high figures. So now I have all pins in their default state during standby and am still measuring 277nA. There is nothing attached except the power supply.

1) So it's not GPIO leakage.

2) It's not 22uF capacitor leakage. I know that because the first board I assembled had more components soldered (among them 3x 22uF) and measured 278nA. So the number of capacitors does not seem to have a big contribution after several minutes.

My guesses:
Is it because I'm 2.5°C above the ambient temperature? I'd understand an increased consumption of ~30nA if I plot the temperature figures between 25°C and 55°C linearly. They are rather exponential than linear, so actually the overcurrent should be less than 30nA.

Could it be some leakage on the VBAT pin? It's directly connected to VDD but VBAT is not used on the board.

Do you have any suggestions?

ramses946 · ‎2023-07-09

I've assembled another board with the same MCU.

Standby with RTC in low drive mode (281 nA) + SRAM2 (173 nA) = 454 nA expected
Auto-wakeup every 2 minutes.

I'm measuring 590 nA. That's more than 130 nA too much.

Aime · ‎2023-07-09

Hi @ramses946 ,

For the consumption, the typical values are based on one sample and we know that there are a variable between samples from the same family. This is why we provide the maximum values which are guaranteed by characterization and take account all the possible variations.

The maximum consumption value for the Standby mode at 25°C 1.8V + all SRAM2 retained is 368nA.
The 277nA is still lower than the maximum value I think your setup is clean and has no leakage.
I agree with you the temperature has a effect on consumption mostly on Standby mode where most part of the system are powered off.

Best regards,
Aime

ramses946 · ‎2023-07-09

Without RTC the figures would be within the limits, yes.

But with RTC (datasheet figures):
Standby with LSE low drive: 281 nA (no max value specified)
SRAM2 max consumption: 249 nA

That's 530nA in total. I'm measuring between 590 and 600 nA.

Aime · ‎2023-07-09

Hi @ramses946 ,

We can not guarantee the maximum values on LSE tests because of the hardware constraints in production.
This is why we don't have in datasheet the maximum values for the Standby with LSE in low drive mode.

I think the over 60nA - 70nA that you are measuring are due to sample variation.

Best regards,
Aime

ramses946 · ‎2023-07-09

Hi @Aime

I guess you're right. I've measured both MCUs now from a batch, one measuring 580 nA, the other 590 nA. I suppose the whole batch will give figures like that.

I also had the idea that maybe the 32kHz load capacitors are oversized. But measuring the LSE frequency on a scope showed 32768.0 Hz. So the capacitors (10pF each, very close to MCU) seem to be fine.

RhSilicon · ‎2023-07-09

@ramses946 wrote:
Hi @Aime
I guess you're right. I've measured both MCUs now from a batch, one measuring 580 nA, the other 590 nA. I suppose the whole batch will give figures like that.
I also had the idea that maybe the 32kHz load capacitors are oversized. But measuring the LSE frequency on a scope showed 32768.0 Hz. So the capacitors (10pF each, very close to MCU) seem to be fine.

About frequency measurement, it might be interesting to see this project, maybe using a TCXO of 0.1 PPM is better than common crystals:

https://www.instructables.com/High-Resolution-Frequency-Counter/