Nice calibration protocol, but ...
... what does it actually tell us, other than your particular unit is within spec?
A single data point doesn't mean much (this could be a rare 'golden unit'), as you are pointing out.
This data set was obtained for comparison with an Owon B35T that I obtained earlier. That isn't a bad meter, either (aside from the input protection being very dubious), but nowhere near as stable as the BM235.
This meter compares favourably to e.g. the Fluke 179 and is a lot less costly. The better Fluke meters commonly achieve better than 40% of specification when new, then will hold those specs indefinitely.
While I have no doubt that this will be true for all the others as well, isn't this a bit overkill for a low accuracy meter like this?
This isn't a standard calibration process. I chose to take 5000 count points instead of the usual ninety percent of range (5400 count) in order to make debiations from nominal more obvious, and also let the meter stabilise for far longer to see if any bias would creep in -- and it does, but only one count here or there, and that could simply be boundary conditions tipping the least significant digit.
I took down ambient temperature and RH, and also checked the auto ranging roll-over points. I stripped this from the data presented to simplify and improve readability.
While this is indeed 'overkill', I hope it may give a better picture of this meter than can be achieved by most verification methods. Having a metrology lab at my disposal, it seemed a worthwhile exercise to have a close look at the main functions on this new-to-market meter that Dave is selling.