Could it be the jib airflow being different on each tack due to the roller furling? Is this with a fully unfurled jib? Or is it only when the jib is partially furled? If the jib is only partially furled the airflow can be quite affected by the sail that's wrapped around the furler.
My suggestion would be to optimise the jib sheet car position on both tacks carefully, by using the "does the top or bottom of the jib luff first" method. The car position could be slightly different on the two tacks (I know mine is)
Bristol 31.1, San Francisco Bay