I’m eyeing either of these devices (besides other, non-OEM ones) to replace on old TX 3.4 and wondering about their accuracy. The Overview page estimates it as 1.2% but some components in the system suggest otherwise. For instance, we have a micro-transformer with unspecified accuracy and linearity error of 0.2%. Then in the EmonVS we have a burden resistor of 1% tolerance. This would give about 1.2% plus 0.2% of the voltage reference plus 0.5% of te CTs. This all adds up to about 1.9% at least. I’m not saying it’s bad and it should usually be better in practice. Am I looking at this right?
On the other hand, what can be said about the accuracy for low power usage? Say, 10…200W in the idle period of the nights. Did anyone measure/compare the accuracy figures in a real life installation?
The errors add in quadrature (as a quick skim, I’d need to go through the workings more carefully to be completely sure). So, it is sqrt(1^2 + .2^2 + .5^2) ~ 1.13%, not a direct summation. @glyn.hudson has run it compared to a calibrated system and confirmed the accuracy, may be able to comment further.
Yes, we’ve observed around 1% accuracy compared to a MID meter in a real installation. Component tolerances are quoted as worst case, statistically this is very unlikely for every component in the system to all be worse case at the same time.