XNPV using calculated measures

Hi all,

Has anyone tried using the XNPV function on a calculated measure? Here's what it looks like when I try to do it:

The data in "Q3 use for XNPV" is being pulled from multiple tables, which I'm guessing might be contributing to the problem. That said, the actual XNPV calculation in the third column does in fact calculate the NPV of the first row, but then stops. And the totals are way off all across the board.

The formulae are complicated so troubleshooting them might not be worth it, but I was just wondering if anyone knew off the top of their head if what I'm trying to do is even possible (i.e. calculate XNPV of data pulled from multiple tables).

Thanks for any thoughts!