Data in cube is different from what I loaded

A strange problem just happened... I loaded a cube from a flat file and everything looked OK. PSA has correct data. Update and transfer rules are direct assignments. nevertheless one of the characteristics change its value. I do not think a conversion routine has any influence on this although I could be wrong. For example I loaded a flat file with A = '050501' and the cube table gets loaded with '500005'. I've debugged up to update rules and the value is always correct.

If I load the cube through SEM the same thing happens. I write a value and a different value is written in the infocube.

Have any of you experienced the same issue?