Truncation of the signal files in FDTD

Exception of the Ar-filter of MWS, did somebody tried to extend "manually" (by filtering, extrapolation) the signal files of FDTD softwares.
I am mostly interested in the low-frequency range of the signal but I loose a lot of accuracy when I truncate the signal. Ar-filter usually doesnt work well because the signal is, or too short, or monotonic.
I have to face very long calculation time for a not-so-good accuracy until the signal decay, using small frequency range is more accurate but even longer to calculate.