How goes the Refined Harmonic Bin packing algorithm?

Question

I failed to find any paper that explains the algorithm in a simple manner.

I understand Harmonic(M) which goes like this: Size(1/K - 1/(K-1)] -> Type K-1 -> Pack K-1 per bin

at the end Size(0 - 1/K) -> Use Next Fit

now i just need to understand what the difference is between Refined Harmonic and normal one