"Nullpunktregelung" wit a small offset

[webkonto]

Hi There!
I run a system since December 2025 (SolarFlow800Pro+2000W solar modules) and I am basically very happy! I now want to improve a little by adding a certain offset to the control value of the "Nullpunktregelung". Example: My Shelly EM gives +300W (so I do need 300W from the net actually). zendure battery is compensation 300W but: In sum, I see that the system sends a lot of Energy out of my house (could be because of the time delay?) so my idea was to add a small offset to the control value. Example: Shelly EM gives 300W (again), but Zendure compensates only 300W-50W=250W. So 50 W will be drawn from the net. Since I always need some power from the net, this 50 Watt does not bother me, but the outflow does, since i loose that energy.

Any solutions? Should I just do a helper with above logic and point to it in the corresponding menue(Identifying the power consumption value of my house)? Or is there a smooth way to do it?

[harrymayr]

The easiest way for now is to create a template-sensor which substract 50W from the value of your Shelly EM and use this template sensor for the P1 sensor of the integration.

[HenryMoec]

I created such a virtual P1 sensor in the past for my system. One of the features is an adjustable target point.
If you are interessted you can find the virtual P1 with installation guide in my github:
https://github.com/HenryMoec/Virtual-P1-Sensor-for-Z-HA-Integration-

[webkonto]

Henry: Could you maybe give me a short help on how to install it? Here should I place the Yaml?, Helper creation and such... The "Winterbetrieb" means, that I can decide to load the battery to a certain threshold (e.g.80% load) before it unloads? This would be very helpful....

[HenryMoec]

Can you please make an issue in my Repo? I don’t think, this is the right place for discussions about the virtual P1.

[webkonto]

as long as I don't have any better solution (looking to HenryMoec), I started like the following:

1. I defined a statistic-helper ('sensor.statistical_characteristic' type Median, 20 Values, max age 2 mins) of my shellyproem_total_active_power to eliminate peeks (e.g. my electric cooker is heavily PWM-ing). I am not sure, if I want middle value or median, PWM is an Endgegener.
2. I defined a sensor helper ({{ float(states('sensor.statistical_characteristic')) + 30.0}} where the input is the above statistic and added 30W to the result. So the overall compensation should be 30W less than given by sensor values.

not so nice as HenryMoec's work but as long as I need to implement his solution, it might do.