Conclusions so far
The above gives us a number of initial measurements indicating expected accuracy of the heat meters that we use. In general I would expect from these results a tolerance of ±0.1 COP or better on a SPF of 4.0 or around ~3% Accuracy. Technically the MPE (maximum permissible error) of the heat meter together with an electric meter could push this to ±0.3 or 7%, but our measurements to date do not indicate that this is common, our maximum measured error was an under-read of 3.24% on the axioma (average under-read of 1.67%). This said the number of tests we have performed is still relatively small and it seems that the BEIS testing did produce larger errors that were close to the MPE.
As time allows we will continue with further tests on our test bench here and also use it for other interesting experiments such as looking at the effect of adding radiator fans for increasing radiator outputs.
Any thoughts for other test to do and reflections on the above are very welcome!