How to calculate noise and detection limits

[Rosie]

Run replicate standard ( say from 0.01 ppm to 10 ppm )

calculate the standard error of the calibration curve

you have the slope then

LOD= 3*SD.ERROR/Slope
LOQ=10*SD.ERROR/SLOPE

Hope that help

Can someone please tell me, in simple terms, how to calculate standard error?

Thanks!
Rosie

[lmh]

sometimes your chromatography software will do it for you. Note that some software will call it a s.d., some a S.E. (As I remember, Excel and Chemstation produce exactly the same value but call it different things). This is the Excel method:
In Excel, put your concentrations/amounts in one column, and your responses in another.
In a spare cell somewhere, enter the formula =linest(Yrange, Xrange, true, true)
where Yrange and Xrange are the ranges of cells containing your data
Select the cell, and four cells below it, and one column to the right
with these cells selected, click in the window where you type the formula
press ctrl-shift-enter. This turns the formula into an array-formula, and fills all the selected cells
Look up Excel's help on "linest" to find which one is the SE (sd) of the line.

[bgranot]

Run replicate standard ( say from 0.01 ppm to 10 ppm )

calculate the standard error of the calibration curve

you have the slope then

LOD= 3*SD.ERROR/Slope
LOQ=10*SD.ERROR/SLOPE

Hope that help

To all,
Chromeleon has signal to noise ratio calculation using all of the above.
Go to column property and select signal to noise and in parameters you can set it base on the RMS noise of the baseline before and after the peak.
Do not complicate more than what is needed

[MaryCarson]

So glad somebody else revived this thread--made it much easier for me to find today. If you use 1/x2 weighting for your calibration curves, and get acceptable residuals that way, is the resulting standard error of the intercept still a reasonable estimate of the noise? Or should SE intercept noise estimation only be done off an unweighted curve? (which presumably only spans a single order or so of magnitude, and not the 4 orders of linearity I get off the mass spec)

[DSP007]

Well, at least in Pharmacy EDQM with the FDA seems to have agreed so far. EP considers average noise at 1 / 2 the height of the noise, FDA - full height between the maximum emission in the chromatogram, this requirement is more severe, if you missed it, then the European requirement is satisfied. The more you are in the U.S., your charter of the monastery - USP.
It is natural that the real peaks of impurities, substances whose structure is established should not be taken into account.

[ender]

I always go back to the definition in the Code of Federal Regulations (40 CFR Appendix B to Part 136).

Also see:

http://www.dnr.state.wi.us/org/es/scien ... cument.pdf

- ender