[volt-nuts] correlate volt meters for more accurate readings

[Poul-Henning Kamp]

--------
In message <3EE48329-EDB5-44D1-AB6C-68C81777DD6E at hanler.com>, Jerry Hancock writes:
>Hello all,
>
>I’ve been running plots of my various and rebuilt volt standards (aka DAS-46 which we have working now) and during that process, someone mentioned to me about using two meters at once and then correlating the outputs.  I have a pair of both 3457a’s and 3456a’s,  and though the 3457a has the extra dig
>it, the 3456a meters seem to be more stable.
>
>So assuming I get a time-stamped dataset from both meters, for instance the 3457A meters in hi-res mode, would I just average the two samples? 

First thing you should do is plot them in X:Y mode and let that graph guide you.

If you get a straight line from lower left to upper right, the two meters track each other
and averaging will get you little more than a sqrt(2) reduction in noise.

If you can, also record a representative room temperature, plot
that on one axis and the two meter readings on the other, then think
creatively about the result...


-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.