# Long Term Drift of IC

I wanted to calculate Long Term Drift of output voltage of an IC and in datasheet the value is given as 60 ppm/√kHr. I dont know what exactly this means as there is square root of kHr. Can anyone pls help me for 3000 hours what would be the long term drift of this IC? attached is the graph from datasheet.

• My best guess would be 3000 = 3khrs so, drift = 60ppm * sqrt(3) = 104ppm......... For 2000hrs, drift=60ppm*sqrt(2)=84.85ppm which seems close to the max we are seeing in the graph.
– sai
Commented Mar 27, 2023 at 10:57
• How many kilo-hours are 3000 hours? What is the square root of that? Commented Mar 27, 2023 at 11:12

Not a clear way to notate it, but my interpretation would be:

kHr = kilo hours = thousands of hours of active use

total drift = drift rate 60 ppm/√kHr * √time = 60 ppm/√kHr * √3 = ~104 ppm

The manufacturer seems to expect the long term drift to follow 1/sqrt(x) pattern. That may be through the part stabilizing or the errors to counter each other. Looking at the data, it is would be difficult to claim such, possibly due to counter measures in place within the IC.

Why at least one value in the data set is out of the spec - the spec might mean that 1 or 3 or 6 sigma limit of the full data set is at the specification, so 68.3 / 95.5 / 99.7 % of parts comply.

• I wouldn't call it unclear; ppm/√kHr is a very common unit to cite long-term drift in. You'll see it in most datasheets for high-precision parts. Commented Mar 27, 2023 at 20:57
• Clear way would be to write the algorithm in the datasheet. I think part of high reliability design is to clarify any smallest doubts in intepretation and double check this kind of things with the chip manufacturer. Commented Mar 28, 2023 at 16:30
• Should every datasheet that cites resistances include a definition of ohm's law, then? I think drift being proportional to the square root of time is sufficiently simple and well-known to engineers who work with precision parts that it's safe to make the assumption. Commented Mar 28, 2023 at 16:54
• Also, none of the parts in the dataset are out of spec. The drift for 2 kHrs would be about 85 ppm, and the worst-drifting part appears to go from perhaps -10 ppm to +60 ppm, a drift of only 70 ppm. If you're talking about the one that ends up at about +90 ppm, note that that part starts out at +30 ppm, so the drift is actually just +60 ppm. Commented Mar 28, 2023 at 16:56
• I was about to add that I wouldn't expect a resistor datasheet to tell the algorigthm for current :) Good note on the 30 ppm inital offset. As the graph is for long term/time drift, I could also guess that it's actually part of the drift within early use. Commented Mar 28, 2023 at 20:40