Mine (R2T) is offset by 0.3C, and now my performance is very close to @matt-drummer (tbc still not MID comparable in accuracy, but much closer, and behaviour/patterns from the data are still valuable even without MID compliant sensors)
This excercise came about as a result of learning R1T on our BUH models (EDLA 09/11/14/16) is not of any use.
Absolute values are less important: heat output is more related to the delta between the values the sensors report, so the sensor data matching when the pump is running fast but with no compressor running (no heat input) is key for the mass flow rate formula.
This maybe an ESPAltherma & Home Assistant polling/template out-of-sync issue - ESPAltherma every 30 or so seconds from the HP, vs the sub-second power draw values from your external CT clamp when calculating the COP, could you filter out negative values via a template before sending to OpenEnergyMonitor?
This is a long thread and I admit to not reading it all through! It seems that folks are unsure of their COP calculation accuracy and I wonder if you are aware of this from John Cantor’s website (though I guess most are using numbers from ESPAltherma rather than a CT clamp):
Note, most of these products with a Current Clamp actually measure the current. They are accurate for normal electric heaters, but there will be a considerable error when used with electric motors. e.g. heat pumps. They can be checked or calibrated by running the heat pump whilst counting pulses of your electric meter over a time period. (use a stop-watch) . Electric meters give ‘pulses/kW’. This is printed on your electricity supply meter. Your actual consumption in kW can be calculated from this.
I’ve got a Pi Zero and a pair of DS18B20 temperature sensors that I will setup to measure the flow and return in my garage. I’m interested to see how they compare to R2T - R4T.
I will also try the calibration procedure noted above
John Cantor is clearly referring to the cheap and cheerful off-the-shelf devices that do indeed measure current - and often only current and guess a preset constant voltage - and therefore are only able to calculate (at best, if the display unit is mains-powered) or guess (at worst, if everything uses batteries) apparent power, which is what @johncantor infers by “there will be a considerable error when used with electric motors.” because the power factor of the motor means real power will be significantly less.
All OEM devices, when used with the voltage monitor (emonVs or a.c. adapter) will measure real power, to within the accuracy of their calibration. This can be better than 1% if done carefully.
Old fixed speed (non inverter, non variable) heat pumps could have quite a bad power factor. This gives a false high power reading. However, inverters have a very good powerfactor, so just measuring current can be good enough for simple analysis, and sometimes you can calibrate the CT, BUT be mindful that part-load powerfactor can be worse than full-load.
I personally like the very frequent readings from a CT clamp, but accuracy is never certain. Accurate pulse-counting from a kWh meter can be used along side a CT. … one for accuracy for COP assessment, the other for good 'resolution.
I really must find time to renew a lot of my website in case things have moved on. some of it is now quite old!