Hello Kevin,
So here's the deal.
#1: I had the -uR and -uD options interpreted backwards in the command line parser. That's now fixed and if you want the corrected version let me know what binaries of AFNI you use. For the moment, all you'll need to do is use -uR instead of -uD.
#2: For time series that have very very low SNR, no delay can be computed reliably and it is possible that the delay in seconds exceeds the stimulus period simply because of the way the program is setup. Essentially delay values don't have meaning at voxels with very low SNR (as measured by the cross-correlation coefficient). To change delays to radians or degrees, I do not perform phase unwrapping so it is possible to get angles > 360deg. for such voxels.
let me know how it goes.
cheers,
-ziad