The Q vectors to use in the calculation of the current correlation function should be the real Q vectors, i.e. (2pi/L)(hkl) for a cubic box and not the normalized ones 2pi(hkl). This can be simply solved by removing the two lines that perform the conversion:

qVectors = traj.universe._boxToRealPointArray(qVectors.T)

qVectors = qVectors.T

Looking to this with Eric, we found that in the DynamicCoherentStructureFactor calculation, there is a double conversion: From real Q vectors to normalized vectors and from real atomic coordinates to fractional coordinates. This is correct, but unnecessary. One can remove both transformations and the result is the same (I checked this).

I will submit a pull request with the suggested changes for both files.