开云体育

Jeff,

that's a how long is a piece of string type question. But it's still a good one. How accurate a reference needs to be rather depends on the purpose to which the DVMs which might be checked against that reference will be put. A good rule of thumb might be to aim for a reference whose inaccuracy is 10 times less than the accuracy that is being aimed for. For example, for a meter that is accurate to 1% of measured voltage, then a reference that is accurate to 0.1% would be good. For a meter accurate to 0.3%, then a 0.1% accurate reference would be tolerable, but could do better.

But that then raises the question as to just how accurate that reference is on the day when it is used. How much does it drift with time? What is the temperature coefficient? What can one check it against? Does that checking reference have any traceability?

The AD584K is, according to the spec sheet, good to, or better than, ±0.1% on the 10?V output.

Thinking of a 3? digit DVM, that is a resolution of 0.05% at FSD, or for a 4? digit DVM 0.005% at FSD. The claimed accuracy is quite another matter, but is probably specified for the meters that you own. E.G. for a 4? digit meter of specified DC accuracy ±(0.05% + 3 LSDs) that works out, at 10?V DC to ±8?mV, or ± 0.08%. Now, in regards to the reference, the question becomes who is validating whom!

My own experience with one of the Chinese suppliers of AD854 units including a built-in Li-ion cell is that the measured voltage from the 10?V output is very consistent across my small fleet of DVMs, ranging from a few years old to 30 and more years. That's consistency is very encouraging and a tribute to the basic pocket multimeter design of PCB switchery and precision resistors. (And the utility of end-of-service pulls—best sort of recycling!)

HTH, 73, good question BTW,

Robin, G8DQX