On re-reading I realized yr original query - frequency of calibration - got side-tracked. My apologies.
There may be a terminological issue here. Such has occurred previously.
The activity you refer to as "verification" I call "calibration" as far as the balance is concerned. It's true that my activity is carried out using "sub-master" weights which are derived from the certified masters but this is a perfectly acceptable procedure.
I describe the daily checking activity as "daily calibration" and no auditor has ever challenged this.
The master weights (primary standards) used to internally generate the sub-masters are sent out for annual certification by an official certification lab. using traceable standards. This method has aways been auditorially acceptable.
Can anyone help me find the regulations or guidance on this?
As per above, it maybe needs to be determined what the "this" actually is.
IMEX there are 2 requirements from a procedural POV -
(a) A procedure to defne the frequency of (my) calibration
(b) A procedure to operate/maintain the balances in routine use
For (a) -
Generic procedures exist whereby a "high" frequency of calibration (my terminology) is initially used. Standard methods/rules have been published (attached on this forum) which read something like - If no failure of specified tolerances in a X month period (eg 6-months) occurs, can reduce the frequency by Y% (typically 10-20% from memory). And so on. This method is called the "ladder" procedure. in yr case, the initial frequency might have been twice daily such that if no failures the frequency could be reduced progressively to daily. i imagine you are happy with a daily check so no reason to investigate further.
Examples of this have been attached on the forum, they are all pretty similar. The procedures usually include a proviso for action where failure of a balance to maintain (my) calibration occurs.
Hope the above makes sense.