Subject: Re: [nmr_sparky] Peak integration
From: Thomas Goddard
Date: Jan 19, 2006

Previous: 63

Hi Bradley,

One way to avoid the bounding contour too large problem is to turn
on Use data within rectangle in the Integration Settings dialog (it).
Then when you drag a box around a peak it only uses the data in the
box you dragged and within the lowest contour (assuming Use data above
lowest contour is also turned on). In 3D the depth of the dragged box
is set to a fixed number of planes on either side of the plane being
displayed using the View Depth dialog (vz).

I agree with Josh Ward that looking at the slices is the best way to
see if the integration worked well. A less reliable way is to show the
Fit residual in a peak list (lt) (use the Options button on the Peak
List dialog). A large number indicates a bad fit. But a small number
may not mean a good fit, because it only considers the data used to make
the fit (the data above the lowest contour) and the fit may be good in
that small region but horrible outside. Often bad linewidth results are
because not enough data is being used in the fit because the lowest
contour level is too high so only the top of the peak is seen in the
fitting calculation.

Tom