Why do digital cables sound different?

Whether different digital cables sound different is very dependent on the DAC design. Digital cables will typically NOT drop bits. Any degradation has to do with jitter that is introduced due to the cable losses. DAC's that completely re-clock the data stream will be much less sensitive to the quality of the digital cable and any jitter that the cable causes. However, most DAC's are not like this and therefore have some sensitivity to jitter.

Digital interconnects that exhibit low-capacitance, low-dielectric-absorption and generally low-losses will introduce the least jitter into the digital bit-stream.

This is why they sound different.

Zilla - it is impossible for a digital cable to act as a "tone control". What happens is that the jitter caused by the cable losses results in frequency modulation of the analog signal. This can cause lack of clarity and focus. In general, the better clarity and focus, the better the digital cable.

Jt25741 - You are dead-on when you say that jitter is the main issue with digital cables. However, with some cables, you get what you pay for. Every cable manufacturer has a digital cable and many of them are not even close to 75 ohms characteristic impedance. They need to be 75 ohms. As for the connector, the best you can do with an RCA is to get the impedance right up to the entry point of the jack. After that there will be a discontinuity. Jacks are never 75 ohms. Some manufacturers do this. I do.

Sean wrote:

If one could resonate a cable that introduced a perfect non-reactive 75 ohm load at that frequency, you would end up with no standing waves.

There may be no standing waves to a pure sine wave at that frequency, but you would certainly experience reflections with digital edges. There are only a couple of ways to eliminate reflections with digital signals, including:
1) series terminate (75 ohm driver) into a 75 ohm Zo cable - reflection from the end is absorbed at the driver.
2) parallel terminate a low-impedance driver into a 75 ohm Zo cable using a 75 ohm resistor at the end - no reflection.
3) Series and parallel terminate - (75 ohm driver) into a 75 ohm Zo cable with a 75 ohm resistor at the end - this cuts the signal in half, but eliminates all reflections.

Sean - Transmission-line effects are the main concern with digital cables. Characteristic impedance matching is a big part of this. However, dispersion of the digital signal is also caused by dielectric absorption, which can cause jitter, so just matching to 75 ohms is not sufficient to minimize jitter.

"Shouldn't a cable that has a higher velocity factor be less prone to signal deterioration / absorption due to the signal spending less time in the cable ?"

There will generally be less absorption in a high-velocity cable because in order to get high-velocity, you need a low dielectric constant. Low dielectric constant results in lower capacitance and lower dielectric absorption. The time that the signal transits the cable (propagation time) is really of little consequence itself. This will obviously change depending on the length of the cable. The rise-time of a SP/DIF signal is on the order of 20 nsec, so you would have to have 100 feet to equal the risetime. Technically, this makes impedance a non-issue for a 6-foot SP/DIF cable. However, in practice, impedance discontinuities do impact the sound, particularly the image focus and detail, by adding to the jitter.

As for dielectrics, PVC is at the bottom, getting progressively better with foamed poly, solid Teflon, foamed Teflon, expanded Teflon and finally air. I use Expanded Teflon in my Digital cable. It is hard to put a percentage on the improvement without measuring it. I have plans to purchase a Tek CSA803 communications analyzer, which will measure jitter accurately to a few picoseconds, so I will eventually be able to measure this.

Actually the SPDIF signal is digital, not dual-phase sinewaves. It appears more like a "trapezoidal" wave, which is a square-wave with controlled slew-rate rising and falling edges. This is to minimize radiated emissions for FCC regulations. The signal is around 3 MHz with 12-15 nsec rise and fall times.

Red - I think you're mad actually, just kidding....

I have no explanation other than the usual parametrics, R,L,C and dielectric absorption. Also, dont forget metallurgy.

One thing that I didn't mention before was the length/speed effect that I explain in this paper I wrote for Positive-feedback:

http://www.positive-feedback.com/Issue14/spdif.htm

ading - What engineering school did you go to????? Ive never heard such drivel....

Spluta wrote:
"From what ive seen there is a major flaw in the spec itself The #1 contributer of EMI is the rise time. The quicker it is, the more EMI emitted. A common misconception is the Freq is the culprit not true. Its seems to me that the rise times are faster than needed for the application.

Not true at all. The risetimes are purposely slowed in 99% of transports in order to pass FCC and CISPR. This actually contributes to jitter. If the terminations and impedances are matched, there will be very little EMI, even with very high edge-rates. Please read my PFonline paper.