I verified my chuck and discovered that the jaw motion is produce by a spiral design that turn clockwise or counterclockwise. My hypothesis is that this kind of chuck for miniature processing cannot be accurate since the spiral motion gives asymetrical and not equidistant centering by really small amount but sufficient to produce oscillation.