Wind sensor (Rev 3) thermister vs ds18b20

[bitbang3r]

I'm using the Rev 3 wind sensor with the example code calibrated for a 5v Arduino (forum won't allow me to post link to it on Github) to measure airflow inside my home air conditioner's supply duct (I'm mainly planning to use it to detect if/when the coils begin to ice up after running for too long with the blower at half-speed).

I ran some tests with some additional temperature sensor modules (BME280, SHT11, HTU21, ds18b20) physically adjacent to the r3 wind sensor (using the calibrated example code with an Arduino Mega2560 & external 5v power).

When airflow inside the duct is zero (ie, blower not running) and the air temperature inside is approximately 65-75 degrees (F), the temperature calculated from the wind sensor's onboard thermistor is generally within 2-3 degrees of the temperatures reported by the other sensors.

When the blower is running & air temperature inside the duct is approximately 45-50 degrees (F), the temperature calculated from the wind sensor's onboard thermistor reads about 10-15 degrees (F) lower than the other sensors (~39.8F thermistor vs ~48.9F ds18b20).

Since I can easily mount a ds18b20 sensor within an inch or two of the r3 wind sensor, would I get more accurate windspeed measurements if I modified the code to just ignore the windspeed sensor's onboard thermistor output & base the temperature portion of the windspeed calculations on the value obtained from the DS18B20 instead?

Example:

Code: Select all

    // instead of calculating TempCtimes100 from the windspeed sensor's thermistor,
    // fetch it directly from the DS18b20
    TempCtimes100 = thermometers.getTempC(t1) * 100;
    
    // now work backwards & use it to generate an equivalent TMP_Therm_ADunits value
    TMP_Therm_ADunits = (16.862 - sqrt((16.862 * 16.862) - (0.02 * (9075.4 - TempCtimes100)))) / .01;
    if (TMP_Therm_ADunits < 0)
        TMP_Therm_ADunits = (16.862 + sqrt((16.862 * 16.862) - (0.02 * (9075.4 - TempCtimes100)))) / .01;
    // ... continue with remainder of calculation per example code

[oz]

bitbang3r,

Since I can easily mount a ds18b20 sensor within an inch or two of the r3 wind sensor, would I get more accurate windspeed measurements if I modified the code to just ignore the windspeed sensor's onboard thermistor output & base the temperature portion of the windspeed calculations on the value obtained from the DS18B20 instead?

The short answer is yes, you are better off using another temp sensor since both the thermistor we are using for temperature and our regressions with it, aren't that accurate. Or you could just calibrate the temperature sensor from your other data - it's not going to change once you calibrate it. Or you could also just hard code in a couple of trigger points for various actions.

paul