I've looked at your notebook and it seems to make sense. As far as I can tell the pulse phases are as specified in the paper as well.
Since, as far as I can see, your calculation is correct, my suspicion falls on the evolution equation in the paper. Have you followed through in detail the reasoning in the paper? It should be possible to pick the derivation apart and verify it step by step using SpinDynamica. It's conceivable, at least, that there is a mistake in the paper, or possibly two cancelling mistakes. One place to look is the handling of the signs of the interactions and the spectral frequency axis, which are notorious sources of errors and confusion: see my attempts to clear this up in J. Magn. Reson. 126, 164-182 (1997) and J. Magn. Reson. 142, 190-194 (2000). I do not know whether the paper you cite took the steps necessary to ensure that the signs of the spin interactions are handled correctly when deriving the equations: if they did not, this could easily have caused the discrepancy. Cross-correlation effects, as exploited in TROSY and related experiments, do require extraordinary care in this regard (there is an example in my 1997 paper).
It will be interesting to see if you manage to understand what's going on. Sorry for not being of more immediate help.