I am not sure at all cause this is now a POW R2 - at least the second release we had gotten back then to measure solarroof production where it was working right while the first was not really working.
I am not sure if a shelly would really help a lot or a tuya might be better cause cheaper and everywhere available but the SonOff figures are looking like a lottery or gamble.
224 Volt x 0,51 A must end at least in a 112 Watt figure, for sure not in 70,6 W unless there would be something going on what we know from DC as pwm.
You simply need to measure Voltage and I guess that this figure will be quite right .
The second point will be the current. Both can be achieved by a multimeter easily but it does not really help cause you might only need the right calibration.
That power meter will deliver a Wh measurement that you can crosscheck first by measuring a device you can be sure that it has such power consumption like an old OSRAM bulb with 80W or so. After running such a bulb for a quarter hour you should see a 20,x Wh figure just to make sure what your power meter is measuring. All claim to be accurate but it needs a proof like this or a water kettle running 6 minutes with its 2000W which will mean 6 / 60 hours x 2000 W = 1 / 10 x 2000 Wh = 200 Wh
And from there on you can then start to measure what the aquarium socket will use in parallel while the SonOff plug is plugged into that device and reports its own figures.
You can then compare live figures in the sensor pane with the figures of such power meter and I hope you will get a kind of constant factor as deviation.
The black box for you is the algorithm the sonoff sensor uses for the Wh calculation locally and how accurate this is.
I would try to get my own Watt calculation and let the rieman integrall do the math or measurement over time better known as consumption. And this figure can be compared with the power energy meter.
Cause such aquarium would here be called a power or energy monster considering the 110 W power in a 247 use case which would simply mean roughly factor 9 to get the annual kWh consumption: 990 kWh p.a.
that would cause a 300€ bill here … or a third of a ususal house hold consumption of 2,x persons
And therefore it is more important to get this figure right than what most of our 10 plugs measure which are about 100 - 300 kWh a year except that one that powers the EV with usually 2,3 kW
At least you seem to have the pow2 type as the sensor says which was far more accurate than the previous version cause we had back then both types in use.
The longer you check your values the more questions might occure cause the 4,47 kWh figure for 10 days does not really fit to the sensor screenshot where you could see a most likely daily consumption of 0,94 kWh - 0,87 kWh per day or roughly 4 kWh in 4 days.
What do you power through that POW R2 ?
a pump and a water heating / cooling … if so then that would show a smaller constant consumption for the 247 running pump while the heating or cooling would kick in only from time to time which would show a different pattern like a fridge.
But there is a huge gap between the Watts in your screenshot and the consumption except you would always have created a screenshot during heating / cooling phase cause based on the watts calculated by Volt x Ampere = 112 W over 24 hours the consumption would be a lot higher than just ~ 1 kWh as the sensor had shown.
Here you can get ELV energy master basic that will measure the figures quite accurate but I guess that his energy meter is product you might only get in the EU.
My tuya base figures like V and A do not fit together too but at least the W are right and Watts over Riemann integral deliver a figure which is 100,5% of the power meter figure so the HA reported consumption is about 0,5% higher.
Maybe you find a different device where you can start easier than such a combined use case where at least 2 or more consumers are powered through the POW R2 I guess cause an aquarium usually needs light, warm and a water pump which are running independently and at least warm is controlled by sensor I guess, while light might be running in a time schedule and water pump too or even constantly.
I would try to measure 1 device at a time like that bulb just to get behind what values might be right and what are wrong. It took me back then about 2 months to get behind that whole mess how far off these plugs can be till I found the tuya being accurate and also across the batch in a usual use case where max. 1000 Watt consumers turn on / off cause in spikey use cases the accuracies goes down too, I mean a lot of 1000 W on / off switches per day you usually find when micro waves are running a programm of heating rice.
