I am aware of the issues with i2s, and agree that keeping the cable length short is desirable.
Most devices that support I2S between boxes (including the Denafrips products) use additional buffering (generally LVDS - low voltage differential signaling) which can support cable lengths of 2-3 meters.
As you point out, the longer the cable length, the more likely that slight differences in cable length and propagation delay will cause skew between the clock and data signals. Also, LVDS only has a modest amount of common-mode noise rejection, so the longer the cable, the more likely external noise will interfere with the signal.
With my DCC and DAC on shelves one above the other, I can get away with a 0.75m cable (even one that is fairly stiff), so that's what I'm looking for.
I'm a little leery about buying a hand-made cable since the consistency of the individual conductors lengths, twisting, and shielding will all have an impact on the amount of skew between signals.
I think this is the only type of cable used in audio where clock and data are transmitted separately and where there is no error detection/correction, so example-to-example differences (particularly for a hand-made cable) may be more significant for this type of cable than any other type of audio cable.
Most devices that support I2S between boxes (including the Denafrips products) use additional buffering (generally LVDS - low voltage differential signaling) which can support cable lengths of 2-3 meters.
As you point out, the longer the cable length, the more likely that slight differences in cable length and propagation delay will cause skew between the clock and data signals. Also, LVDS only has a modest amount of common-mode noise rejection, so the longer the cable, the more likely external noise will interfere with the signal.
With my DCC and DAC on shelves one above the other, I can get away with a 0.75m cable (even one that is fairly stiff), so that's what I'm looking for.
I'm a little leery about buying a hand-made cable since the consistency of the individual conductors lengths, twisting, and shielding will all have an impact on the amount of skew between signals.
I think this is the only type of cable used in audio where clock and data are transmitted separately and where there is no error detection/correction, so example-to-example differences (particularly for a hand-made cable) may be more significant for this type of cable than any other type of audio cable.