If you don't match characteristic impedance between source, cable, and destination you might get reflections in the cable on impedance boundaries. These reflections can modify (add to) transition edge. Such modification, can create staircases on the signal transition. It makes variation in exact moment of level recognition - a time jitter. This time jitter translates to noise in frequency domain.
This characteristic impedance is pretty much SQRT(L/C). Inductance and capacitance in the cable depend on geometry and dielectric. Every cable has characteristic impedance. The chances are, that that your IC have impedance that is very close especially to destination (DAC). You often pay for high purity metals or very low absorption dielectric in analog IC, that are not important in S/Pdif cable. In addition, reflections in the cable might miss the edge originating them (hence 1.5 min recommendation) and you might be OK even when it is mismatched. You can experiment, but I would start with inexpensive 75 ohm coax. Main factor here is impedance matching and inexpensive cable might be better than fancy one.
Also, recommendations/reviews posted make no sense, since it is system dependent.
This characteristic impedance is pretty much SQRT(L/C). Inductance and capacitance in the cable depend on geometry and dielectric. Every cable has characteristic impedance. The chances are, that that your IC have impedance that is very close especially to destination (DAC). You often pay for high purity metals or very low absorption dielectric in analog IC, that are not important in S/Pdif cable. In addition, reflections in the cable might miss the edge originating them (hence 1.5 min recommendation) and you might be OK even when it is mismatched. You can experiment, but I would start with inexpensive 75 ohm coax. Main factor here is impedance matching and inexpensive cable might be better than fancy one.
Also, recommendations/reviews posted make no sense, since it is system dependent.