I've just read the last page of this 8+ year old thread. I am essentially in agreement with the comments by Kijanki and Shadorne, and I think that Lightminer made an excellent first cut at a summary, which I think can be expanded upon a bit, and I've done that below.
But first let me say that I think that differences or lack of differences between digital cables are more easily explained, and are a lot more predictable, than differences between most other cables or power cords in an audio system. The key to it, though, is looking at it from a system-level perspective (as Kijanki and Shadorne, in particular, have done). Notwithstanding the observations some have offered about differences between particular cables that in their experience have been consistent across different systems, I think that (as the following summary should make clear) considering a digital cable as having a definable performance on its own, akin to a component such as a preamplifier or cdp, is wrong and is probably the underlying reason for a lot of the disagreements (based on differing experiences with different systems) which pervade this area.
That said, here is my embellished version of Lightminer's good summary:
If your DAC re-clocks really really really well (and most very modern ones focus on that now in varying degrees),
and
If your SPDIF cable is 1.5 meters or so (perhaps a bit more or less depending on the risetime and falltime of the output of the particular transport, and the exact value which the logic threshold voltage level that the receiver chip in the dac happens to have, within its rated tolerance),
and
A reasonably well-controlled 75 ohm impedance is maintained by the cable, the output stage of the transport, the input stage of the dac, and the connectors at both ends (bnc's being the best way of doing that; Canare rca's being a good second choice; ordinary rca's being marginally acceptable),
and,
the risetimes and falltimes of the transport output represent an optimal balance between being too slow (which would increase jitter due to the small amount of noise which inevitably rides on the signal) and too fast (which would increase emi and couple noise into circuit points within the system that could affect sonic performance, and also might worsen reflection effects),
and
the bandwidth of the cable does not significantly degrade risetime and falltime of the transport output,
then the difference in cables should be pretty minimal.
In that sense are we getting somewhere over the 8+ years!
But given the essential impossibility of assuring all of that, the best approach is to optimize as many of those factors as you can, then take into account the experiences of others who have worked with similar equipment (and take their experience with a few grains of salt), try a few different cables, go with what sounds best in your particular system, and don't worry about the conflicting experiences others may report with their own systems.
Re this question that Lightminer had:
Probably somewhat, but mainly just in the sense that the sonic advantages of the higher rez formats would be more significantly compromised by jitter than in the case of the redbook format. The clock rate of 24/192/2channels is a little under 10MHz, well within the cable bandwidths. However, what is important for any of the formats is that the cable bandwidth be adequate to not significantly affect the higher frequency spectral components that are associated with the signal risetimes and falltimes. If that were to happen, jitter would increase comparably regardless of format, although as I say the increased jitter would probably matter more in a higher rez format.
Based on certain assumptions that we don't have to go into here, because we are dealing with rough ballpark numbers, the 25ns rise/fall times of typical transport outputs corresponds to a "3db bandwidth" (meaning the particular frequency is attenuated by 3db) of 0.35/25ns = 14MHz. Faster rise/fall times, which are desirable in terms of jitter minimization until they reach the point where that benefit is overshadowed by emi problems or increased reflection effects, would have correspondingly greater bandwidths. So for any format it is desirable to have a cable that can pass that 14MHz (or higher) frequency without degradation. I would not assume, though, that the 14MHz number can be directly compared with the 20 and 40MHz numbers, because the cable bandwidth may not be specified based on "3db bandwidth," and (more importantly) may be based on the attenuation that results from a length much greater than what would be used in an audio system.
Many common types of coaxial cables used for video signals have 3db attenuation occurring at 100MHz at 100 feet! So I wouldn't be surprised if the 20MHz and 40MHz numbers were based on unspecified definitions that were designed to make the parameters sound relevant to an audio system, when in practice they are not.
Regards,
-- Al
But first let me say that I think that differences or lack of differences between digital cables are more easily explained, and are a lot more predictable, than differences between most other cables or power cords in an audio system. The key to it, though, is looking at it from a system-level perspective (as Kijanki and Shadorne, in particular, have done). Notwithstanding the observations some have offered about differences between particular cables that in their experience have been consistent across different systems, I think that (as the following summary should make clear) considering a digital cable as having a definable performance on its own, akin to a component such as a preamplifier or cdp, is wrong and is probably the underlying reason for a lot of the disagreements (based on differing experiences with different systems) which pervade this area.
That said, here is my embellished version of Lightminer's good summary:
If your DAC re-clocks really really really well (and most very modern ones focus on that now in varying degrees),
and
If your SPDIF cable is 1.5 meters or so (perhaps a bit more or less depending on the risetime and falltime of the output of the particular transport, and the exact value which the logic threshold voltage level that the receiver chip in the dac happens to have, within its rated tolerance),
and
A reasonably well-controlled 75 ohm impedance is maintained by the cable, the output stage of the transport, the input stage of the dac, and the connectors at both ends (bnc's being the best way of doing that; Canare rca's being a good second choice; ordinary rca's being marginally acceptable),
and,
the risetimes and falltimes of the transport output represent an optimal balance between being too slow (which would increase jitter due to the small amount of noise which inevitably rides on the signal) and too fast (which would increase emi and couple noise into circuit points within the system that could affect sonic performance, and also might worsen reflection effects),
and
the bandwidth of the cable does not significantly degrade risetime and falltime of the transport output,
then the difference in cables should be pretty minimal.
In that sense are we getting somewhere over the 8+ years!
But given the essential impossibility of assuring all of that, the best approach is to optimize as many of those factors as you can, then take into account the experiences of others who have worked with similar equipment (and take their experience with a few grains of salt), try a few different cables, go with what sounds best in your particular system, and don't worry about the conflicting experiences others may report with their own systems.
Re this question that Lightminer had:
Now that we are using 24/96 music and lets assume also 24/192, does that change the requirement on the "Freq Resp dc - 40 MHz" type of rating? What MHz is required? Interesting that Kimber D60 (quite expensive) is 40 MHz and the rest of their cables are 20.
Probably somewhat, but mainly just in the sense that the sonic advantages of the higher rez formats would be more significantly compromised by jitter than in the case of the redbook format. The clock rate of 24/192/2channels is a little under 10MHz, well within the cable bandwidths. However, what is important for any of the formats is that the cable bandwidth be adequate to not significantly affect the higher frequency spectral components that are associated with the signal risetimes and falltimes. If that were to happen, jitter would increase comparably regardless of format, although as I say the increased jitter would probably matter more in a higher rez format.
Based on certain assumptions that we don't have to go into here, because we are dealing with rough ballpark numbers, the 25ns rise/fall times of typical transport outputs corresponds to a "3db bandwidth" (meaning the particular frequency is attenuated by 3db) of 0.35/25ns = 14MHz. Faster rise/fall times, which are desirable in terms of jitter minimization until they reach the point where that benefit is overshadowed by emi problems or increased reflection effects, would have correspondingly greater bandwidths. So for any format it is desirable to have a cable that can pass that 14MHz (or higher) frequency without degradation. I would not assume, though, that the 14MHz number can be directly compared with the 20 and 40MHz numbers, because the cable bandwidth may not be specified based on "3db bandwidth," and (more importantly) may be based on the attenuation that results from a length much greater than what would be used in an audio system.
Many common types of coaxial cables used for video signals have 3db attenuation occurring at 100MHz at 100 feet! So I wouldn't be surprised if the 20MHz and 40MHz numbers were based on unspecified definitions that were designed to make the parameters sound relevant to an audio system, when in practice they are not.
Regards,
-- Al