M5PaperMono Power Demo

English | 简体中文

Overview

This repository contains independent ESP-IDF firmware projects for measuring the power consumption of production M5PaperMono hardware. Each test is built and flashed separately, starts automatically after boot, and uses shared board-support modules.

The project is based on ESP-IDF 5.5.5. Build, flash, serial validation, and current measurement are manual operations.

Power Test Results

Power consumption is represented by the average whole-board input current. Unless noted in the linked report, each value is the selected WINDOW Avg from an approximately 60-second measurement window. Display refresh and single-transmission entries use the measurement field described in their report.

Device states are abbreviated in this table; use the details links for complete configurations.

For multi-mode rows, slash-separated values follow the mode order documented in the linked report. The order is low_power / reference for IMU wake, fastest / fast / quality for display refresh, and 0% / 50% / 100% for frontlight.

Test scenario	Device / operating state	Avg. current at 3.7 V	Avg. current at 4.2 V	Details
L0 shutdown	PM1: L0 shutdown ESP32-S3: not initialized	17.81 uA	20.75 uA	Details
L1 standby	PM1: L1 standby BMI270: low-power state; NFC rail off	26.07 uA	29.09 uA	Details
L2 deep sleep	ESP32-S3: Deep Sleep PM1/IOE1: L2 state	129.08 uA	81.06 uA	Details
IMU wake from deep sleep	ESP32-S3: Deep Sleep BMI270: any-motion wake	726.19 uA (low_power) 1.38 mA (reference)	717.64 uA (low_power) 1.36 mA (reference)	Details
IMU wake from shutdown	PM1: shutdown BMI270: any-motion wake	42.72 uA (low_power) 692.80 uA (reference)	45.82 uA (low_power) 696.20 uA (reference)	Details
Button wake from deep sleep	ESP32-S3: Deep Sleep Key1/Key2: EXT1 wake	101.95 uA	86.89 uA	Details
RTC Alarm wake from deep sleep	ESP32-S3: Deep Sleep RX8130CE: Alarm armed	76.05 uA	63.15 uA	Details
RTC Alarm wake from shutdown	PM1: shutdown RX8130CE: Alarm armed	18.26 uA	21.17 uA	Details
RTC Timer wake from deep sleep	ESP32-S3: Deep Sleep RX8130CE: 120 s Timer armed	661.29 uA	671.19 uA	Details
RTC Timer wake from shutdown	PM1: shutdown RX8130CE: 120 s Timer armed	18.26 uA	21.27 uA	Details
Normal operation	ESP32-S3: active SSD1677: quality refresh every 5 s; 50% frontlight	76.80 mA	72.34 mA	Details
Display workload	ESP32-S3: active SSD1677: static frame, 100% frontlight	89.90 mA	85.30 mA	Details
Display refresh modes	SSD1677: complete refresh Modes: fastest, fast, quality	110.39 / 110.13 / 98.60 mA	105.21 / 104.92 / 93.78 mA	Details
Frontlight levels	SSD1677: static frame Frontlight: 0%, 50% or 100%	31.55 / 61.28 / 89.98 mA	29.13 / 58.23 / 85.35 mA	Details
IMU sampling	ESP32-S3: active BMI270: accelerometer + gyroscope	31.63 mA	28.99 mA	Details
Full load	ESP32-S3: active SSD1677, SX1262, NFC, Wi-Fi and frontlight	187.32 mA	176.60 mA	Details
NFC power down	ST25R3916: power-down state	34.35 mA	32.39 mA	Details
NFC-A polling	ST25R3916: NFC-A polling	109.31 mA	103.21 mA	Details
LoRa periodic transmit	SX1262: periodic TX	61.25 mA	59.50 mA	Details
LoRa single transmission	SX1262: single TX	96.55 mA	93.52 mA	Details
LoRa RX idle	SX1262: continuous RX	42.02 mA	39.89 mA	Details
LoRa sleep	SX1262: sleep state	31.71 mA	29.68 mA	Details
Wi-Fi connected	ESP32-S3: station connected Power save: disabled	102.18 mA	93.97 mA	Details
Wi-Fi Light Sleep	ESP32-S3: automatic Light Sleep Wi-Fi: maximum modem power save	4.28 mA	5.31 mA	Details
Wi-Fi periodic traffic	ESP32-S3: station connected ICMP periodic traffic	102.08 mA	94.21 mA	Details

Measurement Graphs

All published captures are indexed in Power Measurement Graphs. The index contains the published images for each voltage and test scenario.

Repository Layout

components/  Fetched third-party components
modules/     Shared M5PaperMono board and test-support components
tests/       Independent ESP-IDF power-test projects
docs/        Published measurement evidence

Build and Flash

Install ESP-IDF 5.5.5 and activate its export environment before running any idf.py command. Python 3 and Git are required by the dependency script. From the repository root, verify the locked component destinations and then fetch missing components:

python fetch_repos.py check

If the check reports missing components, run python fetch_repos.py fetch --skip-existing to install only missing destinations. If it reports a version mismatch, use python fetch_repos.py fetch --replace after confirming that the component tree should be regenerated. The script creates _component_backups/ before replacement. fetch_repos.py reads repos.json, fetches the exact Git commits or registry hashes, applies the patches listed in that file, and installs the components under the paths declared there. A registry component must be fetched from an ESP-IDF shell with the ESP-IDF Python environment available.

Build and flash one test from its own directory:

cd tests/l0_shutdown
idf.py set-target esp32s3
idf.py build
idf.py -p <PORT> flash
idf.py -p <PORT> monitor

Each test directory contains a complete sdkconfig.defaults. Multi-mode tests also provide sparse mode overrides. Do not replace the base defaults file with an override. Before the first configuration, or when changing modes, remove the generated sdkconfig and set SDKCONFIG_DEFAULTS to the base file followed by the selected override, separated by a semicolon. For example:

$env:SDKCONFIG_DEFAULTS = "sdkconfig.defaults;sdkconfig.display-epd-fast.defaults"
idf.py reconfigure

Use the equivalent export SDKCONFIG_DEFAULTS="sdkconfig.defaults;sdkconfig.display-epd-fast.defaults" form in Bash. Generated build/, sdkconfig, and dependencies.lock files are local artifacts. USB Type-C is used for flashing and serial validation only; disconnect it before formal current measurement.

Measurement Method

1. Connect USB Type-C to flash the firmware and confirm TEST_START, BOARD_OK, and TEST_READY through the serial monitor.
2. Keep the same firmware and configuration, disconnect USB Type-C, and power the device only through the measurement setup.
3. Measure at nominal 3.7 V and 4.2 V after the current stabilizes. Publish the selected capture's approximately 60-second WINDOW Avg.

The firmware does not detect the power source or switch between validation and measurement modes. L0 initializes only the shared I2C bus and M5PM1; other peripherals are intentionally not initialized.

License

Project-owned source code is released under the MIT License. Third-party components retain the licenses listed in Third-Party Notices.

Dependencies

Locked component revisions, sources and licenses are listed in Third-Party Notices. Project-specific adaptations are recorded in patches/.