What organism?
FACS is very fast, "later" should be a fraction of the cell division (doubling) time length. Depending on the FACS you have available, you may be able to "bin" your positive cells with respect to brightness... cells that are dim have either only just started to express, or are already losing the plasmid... so I would guess you want the cells around the peak of the histogram (of brightness).
If you still have issues, I might look into the possibility of setting up a feedback loop in the FACS... where the waste/negative-cells (cells which are not bright) are fed back into the input (in case the cells have not begun to express yet, you can keep them looping in the FACS until they just start to express). The challenge here, I think, will be re-adjusting the growth media/buffer, since the sheath fluid likely will affect growth rate. You might need to implement some sort of in-line centrifugation or dialysis, so you can keep the cells happily growing.