Without pointing me to herehttps://www.home-assistant.io/integrations/derivative and Wikipedia which obviously I could go to but choose to come here! …
… can someone explain what it is actually very simple and layman terms , and , can I use it to better process a temperature sensor that has a resolution of 0.5 c , that is to say every time I get a reading from the device sensor it can only ever be plus or minus 0.5 c from the last reading …which isn’t very good for thermostat control
Derivative probably isnt what you would want in this case. Derivative processes the source into a rate of change. The easiest example would be a distance sensor, with a derivative applied, you would get the rate of change, or speed.
Derivative processing can be used in PID controllers (Proportional, Integral, Derivative) but aren’t typically used in residential temperature control.
A resolution of 0.5 degrees is fairly common in temperature sensors. That said, the accuracy of the sensor is more important than resolution in a temperature sensor. You will want to make sure your sensor reports an accurate temperature. That is, when it reports 20.0 C, is it actually that, or does an independant, calibrated sensor report 21.5? If its off, it may need calibration (a whole separate topic in itself!).
Temperature controllers usually employ averaging multiple readings together will reduce the measurement error with single readings. This, along with calibration (if required), will help with good temperature control.
I bet it can take bigger steps.
Place it the freezer or oven.
I’m wondering if the OP meant that the because of the 0.5 C resolution, they cannot see smaller differences ie if its 20 C, they can only see 19.5, 20, 20.5… but they would like to see 19.9, 20, 20.1 etc…
But I totally see how the message can be interpreted as you indicated.
There is no calculation that can increase the accuracy of the current value of a sensor. You might guess other values from the past inbetween two readings after you got the second. But then it would predict temperatures in the past, not the present. Assuming the next value will change in the same direction as the previous is incorrect. But on top of that, as said values are rounded. So when you get 20.5 it may very well actually be 20.3.
That’s perfectly fine for thermostatic control of room temperature. Mine even works fine with +/-1°C resolution.
Maybe not so good for critical industrial process control. But if you want to do that then get a better sensor. In fact get a better sensor and a PID controller.
You absolutely don’t want to use the derivative for this. There are several reasons.
What are you trying to accomplish? I think people here can likely point you in a better direction.
YES!
that’s itt my friend
The room temp sensor is quality, perfectly integrated with HA. Not changing this
Sensor resolution of 0.1⁰c
however
The integration drives an Octopus heat pump controlled via Onecta app.
I automatte this with HA Onecta integration.
Onecta resolution is 0.5 steps and this is fixed so I can’t change this.
So the issue is how to create/translate to/derive a sensor that replaces the onecta sensor from the temp sensor., I may not be making sense because I don’t know how to explain without going into detail of onecta implementation which I do not know.
maybe describing it as a trend works? The trend in room sensor fluctuation needs to be translated to resolution of 0.5⁰c , because adjusting by 0.5⁰c is not as responsive as 0.1⁰c
That’s actually the complete opposite of what couch67 was talking about.
You want to round your “quality” 0.1°C resolution sensor to have a 0.5°C resolution so it affects the thermostat more often.
That’s easy using a template sensor.
configuration.yaml
template:
- sensor:
- name: Low resolution temperature
unit_of_measurement: "°C"
state_class: measurement
device_class: temperature
state: "{{ (states('sensor.indoor_temperature') | float * 2) | round(0) /2 }}"
availability: "{{ has_value('sensor.indoor_temperature') }}"
You can also do this with a template helper using just this bit in the GUI state box:
{{ (states('sensor.indoor_temperature') | float * 2) | round(0) /2 }}
@tom_l
Maybe not that either.
I am not explaining very well.
I am struggling to identify root cause for heating fluctuation and sprinkling ideas over the entire design, targeting perhaps the wrong approach.
Let me go away and think about what is happening, rather than asking for fixes to a problem that might not be the problem. Once I understand the domain I can return and state the problem and effect which should help remove confusion - for that’s all I am created at the moment.
By way of defence, I am learning Pi, Linux, Docker, HA, Onecta, Daikin heat pumps, and more, within short number of weeks so you will understand and hopefully have some sympathy that I will not be as concrete as you folk in any of the aforementioned so bear with me please. I need to stop, reflect, and pose a better question.
Ok, but make sure to tell us what the problem is, not the problem with your chosen solution. See: https://xyproblem.info/