ESP32 Deepsleep + WiFi within 3s - Deal breaker

[itsmaxdd]

Hello,

I'm in need to have a system composed by an ESP32 that wakes up from  deepsleep and reports via WiFi (to influxdb particulary) the sensed data.

I'm not really sure why there is such a 3s hard constraints but i would like to understand the following:

1. Is the 3s hard constraints only during calibration?

2. My understanding is that if i skip one "nextCall" the algorithm will never calibrate (this is what is happening)

3. Between 3s and 300s there was 30s once, was it too hard to implement 😄 ?

4. My code is the exact same code of ESP32 deep sleep with the addition of static WiFi config + a single HTTP Post which during optimal operation takes 800ms-1s but if there is an interference or a collision or a delay in the "internet" part the output is the following:

08:45:12.907 > 000043687: Successfully set state from 41007
08:45:17.428 > 000048209: Deep sleep for 18446744073703422 ms. BSEC next call at 46688 ms.
08:45:17.440 > 000048209: GetTimestamp()) * 1000: 48209000, esp_timer_get_time(): 4609499
08:45:17.707 > 000048483: Deep sleep for 18446744073707405 ms. BSEC next call at 51220 ms.
08:45:17.716 > 000048483: GetTimestamp()) * 1000: 48483000, esp_timer_get_time(): 4883730
08:45:22.567 > 000053342: Deep sleep for 18446744073700687 ms. BSEC next call at 54220 ms.
08:45:22.575 > 000053343: GetTimestamp()) * 1000: 53343000, esp_timer_get_time(): 9742914
08:45:23.703 > 000054483: Deep sleep for 18446744073701405 ms. BSEC next call at 57220 ms.
08:45:23.714 > 000054483: GetTimestamp()) * 1000: 54483000, esp_timer_get_time(): 10883705
08:45:27.170 > 000057951: Deep sleep for 18446744073697468 ms. BSEC next call at 60220 ms.
08:45:27.183 > 000057952: GetTimestamp()) * 1000: 57952000, esp_timer_get_time(): 14352607
08:45:29.702 > 000060483: Deep sleep for 18446744073695405 ms. BSEC next call at 63220 ms.

Now it is clear that a "negative" or "overflowed" sleep time doesnt make sense so i was trying to avoid sleep and perform immediately another reading cycle via sensor.run(GetTimestamp()) (basically a second run of the loop function).

void loop()
{
    if (sensor.run(GetTimestamp())) {
#ifdef DEBUG
        LOG("Temperature raw %.2f compensated %.2f", sensor.rawTemperature, sensor.temperature);
        LOG("Humidity raw %.2f compensated %.2f", sensor.rawHumidity, sensor.humidity);
        LOG("Pressure %.2f kPa", sensor.pressure / 1000);
        LOG("IAQ %.0f accuracy %d", sensor.iaq, sensor.iaqAccuracy);
        LOG("Static IAQ %.0f accuracy %d", sensor.staticIaq, sensor.staticIaqAccuracy);
        LOG("Gas resistance %.2f kOhm", sensor.gasResistance / 1000);
#endif

        sensor_state_time = GetTimestamp();
        sensor.getState(sensor_state);
        // DumpState("getState", sensor_state);
        // LOG("Saved state to RTC memory at %lld", sensor_state_time);
        CheckSensor();

        update = !update;
        if (WiFi.status() == WL_CONNECTED) {
            if (/*influx_client.validateConnection() && */ update) {
                bme680.clearFields();

                // time_t tnow = time(nullptr);
                // bme680.setTime(tnow);

                bme680.addField("Traw", sensor.rawTemperature);
                bme680.addField("T", sensor.temperature);
                bme680.addField("Hraw", sensor.rawHumidity);
                bme680.addField("H", sensor.humidity);
                bme680.addField("P", sensor.pressure / 1000);
                bme680.addField("IAQ", sensor.iaq);
                bme680.addField("IAQ_acc", sensor.iaqAccuracy);
                bme680.addField("SIAQ", sensor.staticIaq);
                bme680.addField("SIAQ_acc", sensor.staticIaqAccuracy);
                bme680.addField("R_GAS", sensor.gasResistance / 1000);
                influx_client.writePoint(bme680);
            }

        uint64_t time_us = ((sensor.nextCall - GetTimestamp()) * 1000) - esp_timer_get_time();
        LOG("Deep sleep for %llu ms. BSEC next call at %llu ms.", time_us / 1000, sensor.nextCall);
        LOG("GetTimestamp()) * 1000: %llu, esp_timer_get_time(): %llu", GetTimestamp() * 1000, esp_timer_get_time());
        if (time_us < 3000000){
            WiFi.disconnect(true);
            esp_sleep_enable_timer_wakeup(time_us);
            esp_deep_sleep_start();
        }
    }
}

What proper solution can i implement to avoid the accuracy to stay at "0" but that allows me to push the data without moving to the 300s configuration?

Regards,

[BSTRobin]

Hi itsmaxdd,

Different power modes to meet different application requirements.
Low power (LP) mode that is designed for interactive applications where the air quality is tracked and
observed at a higher update rate of 3 seconds with a current consumption of <1 mA.

BSEC only needs to be called periodically, and there is no special requirement.
How to implement this requirement on the host platform requires more consideration from the host platform.