Allow 3 digits after decimal point for coef_ext and coef_int

[akopachov]

For now, when configuring VTherm TPI parameters it seems to be impossible to fine-grain coef_int and coef_ext values as they seem to be limited to only 2 decimals after the point

Would it make sense to allow at least 3 decimals after the point?

[jmcollin78]

Hello, of course I can do that but I'm pretty sure that 0.011 wont change anything versus 0.01. coef-ext have a small impact on the power_percent.

Are you sure you need that ?

By the way, i would post a change in the next release for this.

[akopachov]

Well, I just found this set of tuning rules

1. If the target temperature is not reached after stabilization, increase coef_ext (the on_percent is too low),
2. If the target temperature is exceeded after stabilization, decrease coef_ext (the on_percent is too high),
3. If reaching the target temperature is too slow, increase coef_int to provide more power to the heater,
4. If reaching the target temperature is too fast and oscillations occur around the target, decrease coef_int to provide less power to the radiator.

And in my case, it looks like sometimes VTherm continues heating at 15-20% level when the target temperature is 0.5-0.7 deg above the set point, so I though that it might be a good idea to try setting coef_ext to something smaller than default 0.01 (according to the p.2 above). But I can't do this because 0.01 is the smallest possible value higher than zero here.

Does it make sense or am I just missing something?

[jmcollin78]

I can help on tuning if you give me regulation curves like explains here: https://github.com/jmcollin78/versatile_thermostat/blob/main/documentation/en/additions.md#regulation-curves-with-plotly-to-fine-tune-your-thermostat

Be careful to put the power_percent in the curves it is very important.

[akopachov]

Oh, that would be fantastic! If you don't mind, could you please take a look at the graph below

What I don't understand here, is why TRV is heating when the current temp is definitely higher than the target temp.

TPI values for this TRV are set as this:

[jmcollin78]

Ok what I see clearly is that the coef_ext is too high. I will post a fix tomorrow in the next release.

I don't known if it is possible but I think you set it to 0.

To be sure it could be interesting that you post the curves when heating is required (this is not the case here)

[akopachov]

It is possible to set coef_ext to 0, so I did it and now it looks like there is no overheating, but there is some small underheating, which means coef_ext should be increased, if I understand things right.

So, looks like in my case coef_ext should be somewhere in between 0 and 0.01 :)

[akopachov]

Another VTherm where I have coef_int = 0.55 and coef_ext = 0.01 acts in a very similar way as well

I guess it will require coef_ext to be between 0 and 0.01 as well

[akopachov]

Third VTherm seems to be doing just fine with coef_int = 0.6 and coef_ext = 0.01

Well, probably not ideal and still might need coef_ext to be just a bit smaller than 0.01, but at least there is no drastic over- or under-heating

[akopachov]

And this is the case when the VTherm actually needs to heat, so it does, but seems to be overheat. This is the same VTherm as I posted here with the same parameters (coef_int = 0.6, coef_ext = 0.01), just found the case where it has to heat.

[akopachov]

Overall, would like to express my appreciation and say thank you for your hard work on VTherm and your quick reactions and turnarounds on issues and discussions. Also, I really like how much configurable you made it and how nicely and precisely it is possible to tune it for different cases

[jmcollin78]

I have release this morning the fix and you should be able to set the coef_ext to something between 0 and 0.01 now. Let me know.

Thank you for the appreciation, it is always a pleasure to read this ❤️.

[jmcollin78]

Looking at the curve, it seems your issue is not a matter of coef_ext because, the temperature continue to grow up while the power_percent is 0. Coef_ext will act on the power_percent but it is already null. So this will not change your issue.

You have a source of heat that is not controller by VTherm it seems.

This is particularly obvious here:

and here:

For this one, the coef_ext need to be increased indeed:

[akopachov]

Indeed, in the kitchen, for example, oven could be a source of heat from time to time. In another room, temp can go up if there are more people there. So yeah, I understand not everything depend on VTherm. But anyway, being able to set values smaller than 0.01 would let me tune things precisely, I believe

[akopachov]

Looks like in the latest release coef_ext can be set to integer numbers only

[jmcollin78]

It turns red but you can validate the window. Very strange ....

[jmcollin78]

On my env:

[akopachov]

Yeah, indeed, still can save the settings even with this field being red. So, no more than just some visual quirks 😅

Thank you.

[jmcollin78]

Will try to understand what happens