16 responses Add your response
Mofi, thanks for the mention.
The way I would put it is that the length that has the greatest chance of being optimal is a very short length, such as 6 to 8 inches, if that is practicable. If a longer length is necessary, the length having the greatest chance of being optimal is 1.5 meters, or a little more, such as 6 feet.
However it is still possible that there will be some systems in which 1 meter will be the best length. At least a few members here who have tried different lengths have reported that to be the case in their systems.
The reason for the uncertainty is that what length will be optimal depends on a complex set of generally unspecified, unknown, and unpredictable system-dependent variables. Including the exact input and output impedances of the components that are being connected, the risetimes and falltimes of the signal being sent into the cable (i.e., the amount of time the signal takes to transition between its higher voltage and lower voltage states and vice versa), the propagation velocity of the particular cable, the jitter rejection capability of the DAC, the susceptibility of the particular components to ground-loop issues, the AC power distribution scheme in the particular setup (which can affect susceptibility to ground-loop issues), and even the possibility that supposedly objectionable amounts of jitter in the timing of D/A conversion, that might result from a non-optimal length, might be subjectively preferable to the particular listener in the particular system, depending on its frequency spectrum and other variables.
As to why cable manufacturers sell 1 meter lengths, as I say it might be preferable, at least subjectively, in some systems. Or it might make no difference in many cases. And some manufacturers and listeners figure to be unaware or unconcerned about length-related effects. And a 1 meter length can be sold at a lower price than a 1.5 meter length, everything else being equal. Among other reasons that are conceivable.
Some years back, I owned a California Audio Labs Delta CD transport and their Alpha DAC. I bought a .5 Meter Kimber Orchid, as the Orchid was supposed to be an excellent AES/EBU cable. It was terrible(the crappy Monster coax I tried first, sounded better). I replaced it with a 1.5 Meter Orchid and found bliss. Just my lesson learned about digital reflections and cable lengths. If I ever happen to again need an AES/EBU cable, I'll be shopping for an Orchid.
Rodman, It is system thing, as Al stated, but in your case it is possible that your transport had about 10ns transition time. Beginning of the transition resulted in reflection from the end of the cable (impedance boundary) coming 5ns later (5ns/m x 0.5m x 2) deforming shape of original transition in the middle, affecting threshold crossing time (jitter). With 1.5m cable first reflection came back 15ns later completely missing transition. Of course it is only a speculation. In addition propagation time is dielectric dependent. (assumed 5ns/m comes from 60% of light speed). Best thing is to try what works best.
A notable statement by the person with whom you were debating was:
Are they aware that the S/PDIF interface has a defined jitter margin? If I am remembering correctly, it is +/- 20 nanoseconds for 1 Fs (44.1 kHz/48 kHz) transmission, which is what the vast majority send over S/PDIF interfaces.I would refer him to the following paper, co-authored by Professor Malcolm Hawksford, a distinguished British professor/academician/researcher/PhD/DSc/Fellow of the AES (Audio Engineering Society)/AES Silver Medal honoree "for major contributions to engineering research in the advancement of audio reproduction":
Is The AES EBU / SPDIF Digital Audio Interface Flawed?
An excerpt from its conclusion:
A simple model of jitter error audibility has shown that white jitter noise of up to 180 ps can be tolerated in a DAC, but that even lower levels of sinusoidal jitter may be audible. These limits place tough constraints upon digital interface design....180 ps is less than 1% of the 20 ns figure cited by the person you were debating. Also, Professor Hawksford’s paper was presented to the AES in 1992; presumably the increased resolution provided by many of today’s systems makes low level jitter even more of a concern than it may have been in 1992.