Turf after maxdiff: weighted by probability

Hi,

I'm trying to replicate the turf analysis in R starting from my output of the Maxdiff.
I use the method weighted by probability, so I use the zero-centered raw scores and take the average such that I have the zero-centered item score.
Then I put these in the formula exp(Ui)/(exp(Ui)+a-1) with a the number of items shown in each task.
I tried this formula to compute the (cumulative) reach for 1 or 2 items, but the results I get are close to the output of the maxdiffanalyzer, but not quite the same.
Am I missing a step in my analysis or am I doing something wrong?

Thanks in advance!