Deep sleep power consumption - has something been changed in the latest ESP_IDF release?

[pcbreflux]

I am looking at the 9mA mesurement but don't see the connected ESP32? only some kind of resistor? Missing some thing? What voltage are the (same?) batteries?

[pcbreflux]

Similar setup with different multimeter:
Source https://github.com/pcbreflux/espressif/ ... leep_timer

LiFePo4 3.4V

[andrew_p]

Blue NOA board is disconnected on the picture with 9uA measurement. ESP32 is just not fully visible on the photo, because it's a side view - ESP32 mounted on a small PCB with battery handlers and powered by 3.2v LiFePO4.

[pcbreflux]

Some additions to the mesurements with the Fluke 17B+:
losses: Long leads, battery holder leads,
crocodile clips, breadbord + wires + pcb adapter
dmm accuracy at mA 1.5% +-3 Digit
uA Range ESP32 dont start
burden Voltage?
shunt ?

[ESP_igrr]

I can confirm the 35uA current consumption in deep sleep mode in the latest ESP-IDF (that's the ESP32, flash chip, and pullup resistors current combined). This may be caused by a regression in RTC clock setup code, we're investigating.

[WiFive]

I can confirm the 35uA current consumption in deep sleep mode in the latest ESP-IDF (that's the ESP32, flash chip, and pullup resistors current combined). This may be caused by a regression in RTC clock setup code, we're investigating.

When librtc going to be open sourced?

[ESP_igrr]

Turned out to be unrelated to RTC clocks.

You can add these two lines at the end of bootloader_random_disable function in components/bootloader_support/src/bootloader_random.c:

Code: Select all

    /* Disable pull supply voltage to SAR ADC */
    CLEAR_PERI_REG_MASK(RTC_CNTL_TEST_MUX_REG, RTC_CNTL_ENT_RTC);
    SET_PERI_REG_BITS(RTC_CNTL_TEST_MUX_REG, RTC_CNTL_DTEST_RTC, 0, RTC_CNTL_DTEST_RTC_S);

We'll have this change in master soon.

With this change, supply current is back to 5.5 uA in my test.

Edit: here's some more detailed info:

Code: Select all

  RTC_PERIPH    RTC_SLOW_MEM    RTC_FAST_MEM     Current, uA
      On              On              On              6.3
      Off             On              On              5.3
      Off             Off             On              4.9
      Off             Off             Off             4.5
 

Current value is the total of all supply rails. Flash is powered from internal regulator.

[pcbreflux]

New Setup with new esp-idf: looks fine

commit 13dfb5568dcc30aa43363c49f7a4505d04036424
Date: Fri Feb 24 22:23:06 2017 +0800

Module: ESP-WROOM-32
Adapter: only Reset with 10k Pullup and GPIO 0 Button, no other components assembled
LiFePo4: Soshine 18650 3.2V 1800mA

Teststeps:

Step1: Flash deep sleep Test firmware: (https://github.com/pcbreflux/espressif/ ... leep_timer), GPIO_DEEP_SLEEP_DURATION=1000
Step 2: USB to UART converter removed
Step 3: to start Fluke 15B+ is bridged/shorted (assumption: DMM shunt/voltage drop in uA range to high, Volt on Fluke 17B+ ~1.8V)
Step 4: when ESP32 enter deep sleep bridge is removed

[WiFive]

New Setup with new esp-idf: looks fine

Hey, better than before, right? 9.46uA -> 5.1uA? Nice.

[Maddesh]

I got the same problem with the current consumption in deep sleep. Can you make the Firmware available again (Link is not working)? Or should it work with the current status of firmware?

Best regards,
Matthias