:: use of ALAVAR allan variance to know sample quality and noise

Allan variance software tool

https://www.eevblog.com/forum/metrology/allan-deviation-for-stability-measurements-on-voltage-references/

download here
http://www.alamath.com/alavar/

the tool can be used to determine the quality of measurement. ie: it will show if a long term measurement has become unstable due to environmental noise? or the measurement tool could be unstable or noisy.

this is the plot of 10v range K2015 shorted self noise 5x20NPLC

of interest is the red/green group of lines sloping downwards. the slope suggests the data capture is good to about 200-300 samples before diminishing returns sets in. slope not going downwards suggests more instability or noise.

this plot is 100v 10NPLC, the sampling is noisy and unstable right from the start. longer sampling does not improve the quality of capture.

this plot is 0.1v 10NPLC, the samples are slightly better than the 10v, up to 800-1000 stable samples range.

this plot is 1v 10NPLC, good samples seem to go up to 1000.

looks like all is in order?
not quite yet
look at 10v range again @ 25NPLC

does not look like a good capture of samples. sampling more does not seem to improve quality greatly.

now 10v @1NPLC, noisy and unstable data good for up to about 20 samples before diminishing returns sets in.
 
the furthest range i have tried 10x100NPLC 10v, the samples start to behave well from here up to about 400 samples.

a very interesting look at 10v 10x10NPLC. some improvement to wards 100-200 samples range.

NPLC10 10v
it seems like the K2015 is not built to measure 10v. it requires a massive amount of filtering to get good sample quality.

NPLC100 10v, the 10v range seems to add noise to itself electronically. by external math, the noise seem to be reduced.

all the above are done with the input shorted. the DMM basically measures its own operating noise at the various voltage range and NPLC settings.

what does "pure" noise look like in ALAVAR?

9999parts whiteFM noise
1 part flickerFM noise
generated by alaNOISE generator
(simply a strong white + weak flicker noise)

https://www.alamath.com/alanoise/

this one is 99parts whiteFM + 1part flickFM

it is starting to look like something we saw up there.

and finally

nearly equal parts of flicker and white noise added together.

so where does white noise becomes so strong it makes a mess of the samples?

hot silicon junctions?
pdf about noise sources esp RTS type noise

https://www.google.com/url?sa=i&rct=j&q=&esrc=s&source=images&cd=&cad=rja&uact=8&ved=2ahUKEwjh1JPQv6fkAhW6I7cAHRJ1DyQQjhx6BAgBEAI&url=https%3A%2F%2Fpdfs.semanticscholar.org%2F7873%2Fae6957bc02d04c2ec96fad836985ec3cc90c.pdf&psig=AOvVaw3n13PcP771gPDT3R_ac4LW&ust=1567148061629268