Convert ppm/sqrt(kHr) to ppm/1000hr

[michaeliv]

Hello,

Looking at the datasheet for the LTC6655 voltage reference, it's long term drift is specified as "60 ppm/sqrt(kHr)" - parts per million per square root kilo hertz.
I've tried searching around but I can't seem to find a way to convert this to ppm per 1000 hours. Can anyone link to more information - what does the unit actually mean ?
LTC6655 voltage reference datasheet: http://cds.linear.com/docs/en/datasheet/6655fe.pdf

Thanks!

[Brumby]

Hr is not Hz

[Chalcogenide]

1 kHr = 1 kilo hour = 1000 hours.

[michaeliv]

Hr is not Hz

1 kHr = 1 kilo hour = 1000 hours.

Oulps ... I honestly read it Hz until you pointed that out . I guess it's one of those cases where your mind "auto-corrects" what it think is misspelled.
But why the square root ? Does it mean 20ppm / 31.6hr ?

[Ian.M]

It means the drift slows down with time.  You cant convert it to a linear drift with time as they have different physical units,  t^-1/2 and t^-1

[michaeliv]

It means the drift slows down with time.  You cant convert it to a linear drift with time as they have different physical units,  t^-1/2 and t^-1

That makes sense, so how would I calculate the total drift in the first 1000hr ?

[tszaboo]

Yes it does. For the first 1000 hours it is maximum 60ppm. For the first 4000 it is 120 and 180 for 9000 hours. Pretty simple. If you threat your reference well, the value might be much less, if you mistreat it, it will be worse.

[awallin]

I guess that spec means the voltage follows a random walk and could be anywhere within the +/- XX/sqrt(time) parabolic region.
Here's an image (unrelated to voltage references, just simulated random walk)
http://blog.oraylis.de/wp-content/uploads/2015/10/Rplot02_thumb.png

A really nice manufacturer or lab might buy 100 references and measure them all and put that in the datasheet - maybe there are example images out there?