ADC steps on big increments on ESP32

ESPHome
1

I am reading a sensor whose current output is 2.4V. I connected it to a Wroom ESP32 dev board. I set the configuration to show me raw values (0 - 4095). and this is my configuration file:

sensor:

# TANK
  - platform: adc
    pin: GPIO35
    attenuation: 11db #ESP32 defaults to 0db measuring 0-1V, 11db = 0-3.9V
    update_interval: 5s
    id: water
    # accuracy_decimals: 1
    unit_of_measurement: "%"
    icon: "mdi:water-percent"
    name: "Water Tank"
    state_class: "measurement"
    raw: true

And these are the result readings:

Screen Shot 2022-04-15 at 10.45.26 AM
Screen Shot 2022-04-15 at 10.45.26 AM2282×872 33.3 KB

My question is, why are the readings 512 points or 256 points apart? Why am I not getting the full 4095 points of detail?

2

Because your signal is noisy?

Because you are not putting in 3.9V from your sensor.

3

a mean why are the increments exaclty 256? If its noise why don’t I get lets say 113 points of difference or any other random number?

4

On the ESP32 the voltage measured with the ADC caps out at ~1.1V

it’s 4095 at 1.1v, the attenuation extends it up to 3.9v. 256 is probably the min step. Now since the adc component has been updated a few months ago, auto is recommended for attenuation. Although the max reading is 3.12v it may be worth trying. Read this for more detail.

5

Why would 256 be the minimum? These are 12 bit ADCs. 2^12=4096.

6

Maybe I am getting this wrong but @ 0 - 1.1v you get 12 bit resolution. If it’s stretched now to measure 0 - 3.9v wouldn’t the resolution be less and then the steps larger?

7

There should still be 4096 steps between 0 and 3.9v

1 Like
8

That would be my understanding that you should get the full resolution between 0 and 3.9V. With 256 points per step its equivalent to a 4bit resolution.