Two adc GPIO inputs with ESP32 Devkit V1

ESPHome
1

I am using a single ESP32 Devkit V1 on WiFi to monitor utility power current in Amps.

My yaml code is not working.

sensor:
  # Current measurement CT B
  - platform: adc
    pin: GPIO32 # Current measurement CT B
    name: "Power Sensor ESP32_3_B"
    id: power_sensor_esp3_3_b
    update_interval: 5s
    accuracy_decimals: 2
    #filters:
    #  - calibrate_linear:
    #    - 0.000002 -> 0.0
    #    #- 0.00372-> 7.65
    #    - 0.00372-> 0.015
    #  - lambda: return x;
    unit_of_measurement: "Amps"
    on_value:
      then:
        - script.execute: flash_led

  # Current measurement CT C
  - platform: adc
    pin: GPIO33 # Current measurement CT C
    name: "Power Sensor ESP32_3_C"
    id: power_sensor_esp32_3_c
    update_interval: 5s
    accuracy_decimals: 2
    #filters:
    #  - calibrate_linear:
    #    - 0.000002 -> 0.0
    #    #- 0.00372-> 7.65
    #    - 0.00372-> 0.015
    #  - lambda: return x;
    unit_of_measurement: "Amps"

Both CTs show the exact same current. This differs from what i actually happening as shown by my AC clamp meter, which show different incomin Amps on each phase.

Obviously I have not yet attempted to calibrate the CTs. I do not understand why the ESP32 is reporting the same current down to the milliamps on both legs.

What have I done wrong please?

btw:this works perfectly when I use two ESP8266s which only have one adc input, hence the ESP32 migration.

Thanks - Pete

2

What is the actual voltage read on your clamps (with a multimeter)? Remember the ESP8266 and ESP32 ADCs do read slightly differently, it’s possible you may need to set an attenuation value.

3

The clamp output voltage is very low. I’m reading less than 0.25V at about an 8Amp load. The CT I am using is the Emporia Gen 2 200A Current Sensor which outputs 0.0 – 0.333V (Measuring 0-200A) so it is a current based output. There is an RC network before the ADC input

Capture
Capture691×386 31 KB
That looks like this. The 22R is a ballast resistor. My circuit is only using the left hand part uo to the NodeMCU, so just the 2 x 10K and 22R + the 10uF capacitor. Obviously the CT plugs in to Tip and Sleeve of the jack.

Please remember that I am seeing differing readings on a pair of ESP8266s whereas they ar absoutely identical n the ESP32. Worse, since I’m using WiFi I cannot use the ESP32’s seconds ADC. The yamls syntax cheker won’t allow it.

I am pretty sure I have messed up my code but I cannot see where – yet.

4

There’s not much to mess up with a simple ADC - it looks fine. Your circuit looks fine too so as you can guess I am no help… :slight_smile:

My only suggestion is that your voltage at the ESP is over 1.1V and the auto attenuation isn’t working, thus why I suggested setting a manual attenuation as per the ADC instructions.

5

From the ADC page:

ESP32 Attenuation

On the ESP32 the voltage measured with the ADC caps out at ~1.1V by default as the sensing range (attenuation of the ADC) is set to 0db by default. Measuring higher voltages requires setting attenuation to one of the following values: 0db, 2.5db, 6db, 11db. There’s more information at the manufacturer’s website.

6

Thanks for your reply. I read the info on ESP32s capping out a 1.1V. I believe that this may relate to earlier versions. Why? Because I put 3V3 and GND alternately on both GPIO32 and 33. The both reprted 3V2 when held high and about 0.08V when grounded. Howver you are right. My potential divider of 2 x 10K resistors will always give > 1.1V at the GPIO pin.

Need to rethink this.

Thanks again - Pete

7

I am no expert on this subject but have gotten the ESPHome power sensor working.
The esp8266 (Wemos D1 mini) has a built in voltage divider so it can measure 0V-3.3V. I believe this hard wire on the D1 mini.
ESP32 will by default read 0V-1V but has an attenuation: parameter that can change that to 0V-3.3V
This is explained here.

Looks like adding
attenuation: 11db
to your yaml file should make the ESP32 work like the NodeMCU8266
Your schematic looks like this one for a ESP8266

8

I missed the earlier post about attenuation: