PS ....

On Oct 17, 4:41*pm, I wrote:
=nper((1 + rg / fg)^(fg / fw) * (1 - rw) - 1, 0, -pv, 0.0001) / fw
[....]
Setting NPER"s "fv" to 0.0001 is a kludge [...].
NPER does not work when I set "fv" to zero.

Probably because we can reduce a number by a percentage almost
infinitely (limited by computer arithmetic). It is probably good
enough to use 0.01 or 0.005 for "fv". After all, this is only an
approximation anyway.