Which budget digital cable ?

Almarg, I have posted on this simple measure a few times now because its benefits are great compared to the cost involved. Yours is the most useful and enlightened response I have yet read. Thanks in turn for it.

I first heard about the phenomenon in a post here by Empirical Audio's Steve Nugent (audioengr). On his Web site's first page he has a link to a technical paper of his which explains in detail ("Paper on S/PDIF cable length in PSOnline"). I had to try it myself, of course, so I made up two otherwise identical lengths of Apogee Wyde Eye. The listening tests I did with a friend were perfectly conclusive: the longer cable was better (with our gear) by a wide margin.

Tobias & Al,

Thanks for pointing this out. Everyone seems to agree that jitter on digital interfaces are a problem sometimes a badly audible one - several AES papers refer to this. I'll have a look at the paper you mention from Steve. I have always used 2 meter or longer digital interconnect cables (Toslink) but only for convenience. Would it be the same issue for Toslink?

Certainly, most of what I have read stresses the importance of jitter rejection/immunity at the DAC end, presumably becuase changing interconnect cable lengths might work on some equipment but not others...

Would it be the same issue for Toslink?

I'm not particularly knowledgeable about electro-optics, but I don't think so. What underlies the phenomenon Tobias referred to is that for electrical signals with spectral components at rf frequencies, such as the electrical digital outputs of cd transports, a mismatch between cable impedance and load impedance results in a fraction of the incident energy being reflected back towards the source (instead of being absorbed by the load). As you may realize this is referred to as a "vswr" effect ("voltage standing wave ratio"). I wouldn't think that is applicable to light waves being detected by a sensor, and even if it were the quantitative parameters (risetimes, etc.) would be different.

Also, multi-mode fiberoptic cable (which I believe is what Toslink and most inexpensive fiberoptic links are) is subject to other sorts of effects, such as light repeatedly reflecting off of the walls of the conductor, resulting in some of the energy arriving at the destination later than that part of the energy which has taken a direct path. Not sure if that is a significant effect over the short cable lengths and the data rates that characterize audio systems (it may very well not be), but it's another indication that transmission of optical signals and transmission of electrical digital signals are just different animals.

Regards,
-- Al

Regards,
-- Al

Certainly, most of what I have read stresses the
importance of jitter rejection/immunity at the DAC end, presumably becuase
changing interconnect cable lengths might work on some equipment but not
other

In S/PDIF, timing is set by the transport's clock, as you know. That means that
timing gets screwed up (i.e. jitter gets worse) if the timing information is not
received correctly by the DAC; errors are maintained through to conversion.
One way for the DAC to deal with this is to maintain a constant 75 ohms
impedance from its input. Read Steve Nugent's paper to find out at least one
reason why this doesn't always happen, though.

Furthermore, whether or not the DAC handles impedance in its own signal
path correctly, the RCA connector, when used, invariably offers a mismatch;
there is no such thing as a 75-ohm RCA connector.

These are two causes of jitter in setups with an external DAC and digital
interconnect. The timing adjustment offered by a 1.5m cable length
diminishes jitter from these causes. Almarg's limpid post makes it easy to
understand why.

Shadorne, you may have heard a lot about jitter rejection/immunity at the
DAC because of the weakness of the S/PDIF clocking system, which uses the
transport's clock, not the DAC's, to establish system timing. ( You can read
more about this weakness at the LessLoss site. ) Double-clocking DACS like
the Apogee Mini-DAC are one approach (in my view, quite successful) to
solving this problem. Another is slaving the transport to the DAC's clock, as
done by Linn (Karik/Numerik), Cambridge (DiscMagic/DACmagic) and
LessLoss. Also, do read Steve Nugent's paper for yet more.

No matter what the DAC's internal design, though, the presence of an RCA
connector suggests that a 1.5m cable may be optimal with that unit.

Greeni, did you try Valab's digital cable? It is really good.