I gave it a closer look. It's more on the upper side. And it's well within the specs as expected, no doubt. But it could be better...
Given a AD584 with its test protocol (measured with an Agilent 34401A), and a measurement with an well calibrated HP 34401A (which values are slightly above), I referred to the average of both to calibrate my meters (if possible)
The average values of the 2.5/5/7.4/10V thus are 2,49961, 5,001085, 7,50091 and 10,00183 volts
My Brymen BM789 reads
2,4999 (+0,01%), 5,002 (+0,02%) in the 6V range and
2,500 (+0,02%), 5,003 (+0,04%), 7,503 (+0,03%) and 10,005 (+0,03%). If I had 5.001, 7.501 and 10.002, it would be perfect.
I'll watch this... maybe the values will get more accurate with a longer on time and near to 20 °C (it's ~24 °C in my hobby room)
And: Most of my cheapos simply do not have that resolution. So a 2.50, 5.00, 7.50 and 10.00 looks better on the first view, but does not have to.
If I compare the AD584 test protocol to the HP 34401, all deviatons are 0,02% or less.
Brymen to AD584 is again within specs (+0,02% @2.5V, +0,03% @5V (+0,04%) in the 6V range. +0,04% at 10V in 60V range.
Absolutely nothing to blame.
I try not to look at the amps socket and everything will be fine ...