WiFi affects ADC1 raw data (IDFGH-6623)

[dizcza]

Environment

• Development Kit: TTGO-T8-ESP32
• Kit version: v1.7
• Module or chip used: ESP32
• IDF version: v5.0-dev tag (master branch has a websocket client example broken - I haven't bothered reporting this; that's why I've stuck to v5.0-dev; master branch has the same problem)
• Build System: idf.py
• Compiler version: xtensa-esp32-elf-gcc (crosstool-NG esp-2021r2) 8.4.0
• Operating System: Ubuntu 20.04
• Using an IDE?: Only to edit the code. The program was flashed from a terminal with the idf.py flash command.
• Power Supply: USB

Problem Description

When WiFi is active, in particular, an ESP32 websocket client is sending data to a server, adc1_get_raw() returns wrong unrealistic values. ADC values are read from an analog SDP806 differential pressure sensor. Without the Wifi running, the ADC1 values are correct - near zero. When Wifi is active, I see outliers with huge (>1k) ADC1 raw values.

The analog sensor is connected to pin 35, ADC1 channel 7.

I'm aware of ADC limitations but they concert ADC2 only - I've read this part of the documentation carefully and intentionally chosen ADC1 to avoid WiFi-related conflicts. If there are limitations to ADC1, please add them to the docs.

Related issues

I first thought it's related to #7781, but Bluetooth, to my surprise, does NOT affect ADC1 readings. Only Wifi does.

This issue supplements #6682 though in my case the wifi does not get disconnected when I call adc1_get_raw() periodically.

Expected Behavior

adc1_get_raw() returns small values around zero when the WiFi is running.

Actual Behavior

adc1_get_raw() returns unrealistically large values when the WiFi is running.

Code to reproduce this issue

websocket_app_start() is taken from the websocket example https://github.com/espressif/esp-idf/tree/master/examples/protocols/websocket.

void task_send() {
     char data[512];
     
     while (1) {
            esp_websocket_client_send_bin(client, data, sizeof(data), portMAX_DELAY);
            vTaskDelay(pdMS_TO_TICKS(100));
     }
}


void app_main() {
    ...
    websocket_app_start()

    ESP_ERROR_CHECK(adc1_config_channel_atten(board->sdp.adc.channel, ADC_ATTEN_DB_11));
    adc1_config_width(ADC_WIDTH_BIT_12);
    esp_adc_cal_characterize(ADC_UNIT_1, ADC_ATTEN_DB_11, ADC_WIDTH_BIT_12, DEFAULT_VREF, &m_adc_chars);

    xTaskCreatePinnedToCore(task_send, "task_send", 4096, NULL, 0, NULL, PRO_CPU_NUM);

    ...

    // A periodic timer is set up to poll ADC1 at 1000 Hz freq
    int adc_raw = adc1_get_raw(board->sdp.adc.channel);
    int aout_v = (int) esp_adc_cal_raw_to_voltage((uint32_t) adc_raw, &m_adc_chars);
    // aout_v values are printed

Full code:

1. ADC1 readings: https://github.com/dizcza/esp32-sdpsensor/blob/dc6edd0b88562a05a875bffb4dbd0ddb8ff043cf/main/sdpsensor_adc.c#L66
2. A websocket client sending data in another thread: https://github.com/dizcza/esp32-sdpsensor/blob/dc6edd0b88562a05a875bffb4dbd0ddb8ff043cf/main/record_sdp.c#L243

Other items if possible

• sdkconfig
• elf file

[jack0c]

@dizcza Can you try commit after f3adbf9

[dizcza]

I don't have the module with me at the moment.
You are free to close the issue and if I find it's unresolved I could reopen it.
Thanks.

[Alvin1Zhang]

Thanks for reporting and updates, feel free to reopen.

[clemrdh]

I am having the exact same issue ATM

[piemanau]

@dizcza Have you tested it after this change? As discussed on this issue esp-rs/esp-idf-hal#468, I am still struggling.

[dizcza]

@piemanau I haven't tested it in years. I no longer use WiFi + ADC1 in conjunction.