ESP32C3 light sleep by UART

[eos1d3]

Board

ESP32-C3-DevKitM-1

Device Description

ESP32-C3-DevKitM-1 connected with STM32-G030 via UART

Hardware Configuration

Serial1 RX pin: GPIO3
Serial1 TX pin: GPIO2

Version

v2.0.4

IDE Name

PlatformIO Espressif 32 (5.1.0) > Espressif ESP32-C3-DevKitM-1

Operating System

macOS 12

Flash frequency

40MHz

PSRAM enabled

no

Upload speed

115200

Description

There is a solution for this issue. This solution helps UART wake up. But there are still issues:

When you keep uploading the same sketch (without power reset), or pressing reset button, sometimes wake up may cause system reset with the following message:

SP-ROM:esp32c3-api1-20210207
Build:Feb  7 2021
rst:0x8 (TG1WDT_SYS_RST),boot:0xe (SPI_FAST_FLASH_BOOT)

When restarting happens, it may keep on restarting for a long time, or may resume normal wake up after some restarting. And it may restart again.

But It seems power reset may help sometimes.

And for using esp_sleep_enable_timer_wakeup only, there is no restart issue.

Sketch

void setup()
{
    Serial.begin(115200);
    Serial1.begin(115200, SERIAL_8N1, RXD2, TXD2);
    while (!Serial || !Serial1) {
        delay(500);
    }

    // GPIO 20 is the default UART_0 RX pin
    gpio_sleep_set_direction(GPIO_NUM_3, GPIO_MODE_INPUT);
    gpio_sleep_set_pull_mode(GPIO_NUM_3, GPIO_PULLUP_ONLY);

    int wake_thresh = 3;
    uart_set_wakeup_threshold(UART_NUM_1, wake_thresh);
    esp_sleep_enable_uart_wakeup(UART_NUM_1);
    Serial.flush();
    Serial1.flush();   
}

Debug Message

ESP-ROM:esp32c3-api1-20210207
Build:Feb  7 2021
rst:0x8 (TG1WDT_SYS_RST),boot:0xe (SPI_FAST_FLASH_BOOT)
Saved PC:0x40380000
SPIWP:0xee
mode:DIO, clock div:1
load:0x3fcd6100,len:0x438
load:0x403ce000,len:0x918
load:0x403d0000,len:0x24e4
Attempting to boot anyway...
entry 0x403ce000
ESP32 Chip model = ESP32-C3 Rev 3
SDK version = v4.4.1-472-gc9140caf8c

Other Steps to Reproduce

No response

I have checked existing issues, online documentation and the Troubleshooting Guide

• I confirm I have checked existing issues, online documentation and Troubleshooting guide.

[SuGlider]

@eos1d3 - I tried this and I can't reproduce the issue.

Please, make sure you are using Arduino Core 2.0.4, because this output SDK version = v4.4.1-472-gc9140caf8c is not for 2.0.4

[SuGlider]

TG1WDT_SYS_RST is usually related to using SPI Flash pins or some task that takes too long to "return".

[eos1d3]

Today I have more tests. And to simplify the test, I focus on GPIO light sleep first.
The sketch is listed below:

#include <Arduino.h>
#include <HardwareSerial.h>
#include <WiFi.h>
#include "time.h"
#include <esp_sleep.h>
#include <driver/uart.h>
#include <esp_task_wdt.h>

#define LED1 0
#define LED_UART 1
#define LED_RUN 10
#define RXD2 3
#define TXD2 2
#define WAKE_PIN GPIO_NUM_9
#define RX_DATA_MAX_SIZE 10

unsigned long ledUARTLastChange;
uint8_t rxData[RX_DATA_MAX_SIZE];

void setup()
{
    Serial.begin(115200);
    Serial1.begin(115200, SERIAL_8N1, RXD2, TXD2);
    while (!Serial || !Serial1)
    {
        delay(500);
    }

    pinMode(LED1, OUTPUT);
    pinMode(LED_UART, OUTPUT);
    pinMode(LED_RUN, OUTPUT);
    pinMode(WAKE_PIN, INPUT_PULLUP);

    Serial.printf("GPIO Wake up ok? %d\n", esp_sleep_is_valid_wakeup_gpio(WAKE_PIN));
    gpio_wakeup_enable(WAKE_PIN, GPIO_INTR_HIGH_LEVEL);
    esp_sleep_enable_gpio_wakeup();
}

void loop()
{
    if (!digitalRead(WAKE_PIN))
    {
        Serial.println("== Sleep ==");
        Serial.flush();
        Serial1.flush();
        esp_light_sleep_start();

        Serial.println("== Wake up ==");
    }
}

And platformio.ini is

[env:esp32-c3-devkitm-1]
platform = espressif32@5.1.0
board = esp32-c3-devkitm-1
framework = arduino
monitor_speed = 115200
monitor_port = /dev/cu.usbserial-14140
upload_speed = 1500000
upload_port = /dev/cu.usbserial-14140

I tried espressif32@5.1.1, espressif32@5.1.0 and also espressif32@4.4.0
5.1.0 is said Added support for Arduino v2.0.4

And all these versions will lock up the CPU and no console message after just a few press of the test button. Mostly it will not last for 10 times of button press.

Then I tried ESP32 IDF 4.4.2 and the offical lightsleep example. I do not modify the code and it also has the same lock up issue.

Finally I tried IDF 5.0 beta1 with the same lightsleep example, only this version works normally without any lock up.

[SuGlider]

#define LED1 0

GPIO 0 is used by the board to reset it when the sketch is uploaded.

[SuGlider]

#define WAKE_PIN GPIO_NUM_9
#define RX_DATA_MAX_SIZE 10

GPIO 9 and 10 are used to access Flash.
It is used in QIO mode, but not in DIO.

[eos1d3]

@SuGlider Hi, thanks for the information.

GPIO0 can be used and is not reset purpose. While GPIO9 is the boot button and it is used by ESP32 IDF lightsleep example. So GPIO0 and GPIO9 are not the root of cause.

But from your suggestion, I found something:

1. when using GPIO9 as wakeup pin, if I use on board button, there is no lockup issue.
2. when using GPIO9 as wakeup pin, and connect it via a jumper wire with 0.1uF cap to a breadboard. The filter cap must create the lockup issue in around 5 press for any of the buttons. When I remove the cap, it is much better but it will lockup in 100 times or more.
3. Now use GPIO4 as wake up pin, with the 0.1 cap. it is not easy to get the lock up but it will lock up finally. Removing the cap seems no lock up.

So can this be noise issue of the jumper wire (15cm) with the breadboard? But it is strange that using ESP IDF 5.0 beta1, no lock up occurs.

[SuGlider]

@eos1d3 - nevermind.... I forgot that you are using the C3. GPIO 9 and 10 are problem for the ESP32.
I'll try your pin setup and see if I can reproduce the issue.

[SuGlider]

upload_speed = 1500000
upload_port = /dev/cu.usbserial-14140

I can see that the upload speed is not standard:
Please change it to 921600
upload_speed = 921600

[SuGlider]

Then I tried ESP32 IDF 4.4.2 and the offical lightsleep example. I do not modify the code and it also has the same lock up issue.
Finally I tried IDF 5.0 beta1 with the same lightsleep example, only this version works normally without any lock up.

This is interesting.... I'll try it as well.

Can you also send a picture of your board/circuit?
The STM32-G030 is also configured at 115200 - 8N1?
How is it connected? Common GND is used?

[SuGlider]

And all these versions will lock up the CPU and no console message after just a few press of the test button. Mostly it will not last for 10 times of button press.

There is the button bouncing effect... Not sure how it affects your exemple.