The FDR conversion via the 'new' method and the 'old' method use very different internal data structures for sorting the p-values -- and only the 'old' software has the printing out ('-list') option implemented. Basically, 3dFDR is 2 separate programs combined into 1 -- the 'old' program, which can be slow, and the 'new' program, which can be much faster for high resolution datasets.
It is very important to use a brain mask to avoid trying to compute q-values on trivial values from outside the brain. Otherwise, the q-values will be larger than they should be.
For some explanation of how p-values are converted to q-values, see
[
afni.nimh.nih.gov]
In particular, slide #4.
It is hard to say why your t-values didn't "survive" the conversion to q-values, since I don't have your data.