Posted: Thu Oct 16, 2008 8:19 pm
by danko
Slammy1, it seams to me that you’re addressing the X-axis value (time), whereas the mystery of EP S/N calculation appears to be related to the Y-axis value (i.e. signal).
Or do I misunderstand your comments?
Best Regards
Posted: Tue Oct 21, 2008 5:23 pm
by Slammy1
The EP S/N calc monitors a range of baseline in a blank injection of 20x(width at 1/2 height of the peak of interest) preferably equally spaced around the expected retention time of the peak of interest. It also measures absolute deviation, both overerestimate true instrument noise but not by a factor of 2 it would seem.
A colleague figured out how to set the fields up to calculate using noise in the blank injection (spent a LONG time on the phone with Waters tech support). His s/n results more closely match expectation based on the RSD of replicate injections. In the end, I'd have to agree with several other posters that it's not really a valid LOQ criteria, but we are forced by EP requirements to demonstrate it at the method disregard level for inpurities methods, so there you go. I guess that's why the USP is so vague.