Carlos kindly gave me a copy of the DOSY spectrum he showed on his blog last year.
I am experimenting alternative processing routes. Here is a detail of the caffeine peaks, after applying the rudest (and probably simplest) treatment. The decays have been linearized (by taking the logarithm). The slope of the line is proportional to the diffusion coefficient. The final results are reported as a normal DOSY spectrum.
Click on the thumbnail to see the image at natural scale.
There is less signal/noise in the tails of the peaks, obviously, therefore the error increases: graphically we see the wings of a butterfly.
Being that it's impossible to correct the phase perfectly, some butterflies are asymmetric.