Hi all.
I am not good at using filter integration, so I try to ask you some help to you
I have a thermostat sensor (a temperature probe connected to a shelly) that gives me data like these:
I would like to smooth a bit the curve in order to better use the temperature to command my heater. What is the best way to use filters in order to remove all those up and downs?
Well, it’s not a matter of visualization, but more of data… Let’s say I have a thermostat using that temperature sensor, the heater is switched of as far as the temperature reach the target, but maybe that was just one of the peak, and the real temperature is 0.3/0
5 degree lower…
You can adjust the minimum cycle duration of the thermostat so it doesn’t switch on/off. As well you can adjust the scan interval of the sensor in many cases. What is the setup of the temp sensor?
The temp sensor is just a DS18B20 probe connected to a shellu plus addon, exposed in HA.
Maybe I try to verify the statistic sensor…
As far as the minimun cycle duration, If I’ve correctly inderstood how it work it just say that if tha state is on/off for less than x time, do not switch on the other state. But here the heater could be on from 2 hours, or maybe 4, before it reach the set temperature, I do not know how much time it needs to reach it
The minimum cycle duration is more about if temp is close to set temp it doesn’t keep turning on and off every min. e.g. if set temp is 21.5 and your sensor temp is 21.4 its on but 30s later it 21.6 it would turn off then 30s later it back to 21.3 and it turns on. So if you set min cyc dur to 20 mins if it went 21.4–>21.6—>21.3 and back and forth it won’t switch off until it’s >21.5 for 20mins consistently.
My temp sensor is esp32 with BMP280 and it only reads every 90s or so. Not as much jitter as yours. Is your sensor exposed to higher air flow(a draft)? What about a bit of a cover for the probe.