Jitter and 75ohm cable length

@Beolab

Hello Fredrik,

As far as I can tell from the descriptions and specs all of those seem like reasonable choices. However the indication that the Van den Hul cable has a minimum safe bending radius of 70 mm strikes me as a bit worrisome. Also, I’m guessing that it costs a good deal more than the others, while not necessarily providing better performance in your application.

Perhaps a good way to proceed would be to order both of the cables that are sold by B&H, since they cost so little, and determining which one performs best in your system, if in fact there is any difference at all.

Good luck. Regards,
-- Al

Hello Beolab,

I would proceed based on the comment by the dCS person, which pertains to your specific equipment, rather than on the more general guideline of 1.5 meters.  As I said in an earlier post in this thread:

Summarizing all that has been said, I would put it that a length of 1.5 meters ... stands a better chance of being ideal than any other length. But as have been cited there are many system dependencies and cable type dependencies which conceivably could result in other lengths being better in some cases.  

Also, it's worth noting that the impedances of BNC connectors are more accurate than the impedances of RCA connectors.  And I would suspect that the impedances of your dCS equipment that the cable would be connecting are more accurate than the impedances of a lot of other digital equipment.  Both of those factors will minimize the reflection effects which are a key factor in the rationale for 1.5 meters.

Regards,
-- Al
 

I came across a number of articles saying you should use at least 1.5M of cable to reduce reflections in the cable so as to not harm the clock signal, yet an RF engineer said this was a bunch of "Bunk" and 1M would be better, in fact he said the shorter the better.

I am an EE with multiple decades of experience designing high speed digital circuits, and also significant experience designing RF circuits (none of it for audio). I say it is definitely not "bunk." However, there are several factors that introduce a degree of system dependency and unpredictability into the issue, so 1.5 meters should be taken as a general guideline, which will not always be optimal. And in some cases the length won't make any difference.

Also, if a very short length is practicable, say 6 to 12 inches, that should be at least as good or even better. It is the intermediate lengths that are the concern.

The basis of the length concern is not to "reduce reflections." The magnitude of the reflections is determined mainly by the closeness of the impedance match between the cable and its connectors, the input impedance of the dac, and the output impedance of the transport.

The point to optimizing length involves the TIMING of the reflections, or more properly, the arrival time at the dac input of reflections of the original signal arrival at the dac input that have re-reflected from the transport output. What needs to be avoided is re-reflection arrival time at the dac input that coincides with the mid-point area of the risetimes and falltimes of the original signal arrival, which is where clocking occurs. If that were to occur, the resulting waveform distortion would be likely to cause a significant increase in jitter. All of that timing is directly dependent on the length of the cable.

Therefore the lengths that should be avoided are dependent on the risetime and falltime of the output signal of the transport, which are normally unspecified, and can be expected to vary significantly among different transports. This paper by Steve Nugent, which you've probably seen, is based on the assumption that those risetimes and falltimes are around 25 nanoseconds. I assume that is a good rule of thumb, but I would not expect it to be precisely consistent across different makes and models.

Also, the amount of time required for the signal to propagate from one end of the cable to the other will vary among different cable designs, because propagation velocity is dependent on the dielectric constant of the particular cable.

Also, jitter can be contributed to by noise caused by ground loop effects between the transport and dac, which, if present to a significant degree, can be expected to worsen as cable length is increased. Conceivably that effect could outweigh the timing consideration in many setups.

Also, different dac designs differ widely in their jitter suppression capability, with some of them (such as the Benchmark) being nearly completely immune to jitter on the incoming signal.

Finally, if the degree of mismatch among all of the impedances that are involved is insignificant, the whole issue becomes moot.

So as I see it the recommendation to use either a very short length, or a length of 1.5 meters or a little longer, has a sound technical basis, and while not always applicable, can be expected to be applicable more often than not.

Regards,
-- Al

Clever experiment, taking advantage of the fact that the CDP has digital inputs!

I have no knowledge as to how practical it might be to have the Kimber cable re-terminated with Canare's, but why bother? If the 3' HAVE/Canare sounds just about identical to playback via the CDP's built-in transport (with no cable and no S/PDIF interface involved), that would seem likely to be as good as it gets!

Regards,
-- Al

If one suspects they have a Jitter problem with their system, how might it manifest it's impact on the analog output.

I second Kijanki's excellent comments, and I would add the following thoughts, which I composed before seeing his response:

IMO it's very unpredictable, and the symptoms will vary widely depending on the spectral characteristics of the jitter, which can be expected to be a very complex mix of discrete frequency components and broadband noise-like components. Some of those components will be correlated with the values of the 1's and 0's defining the music data (which is different than being correlated with the music itself), and some will not be.

Perhaps essentially all that can be said is that there will be a loss of clarity, and an increase in distortion.

The following papers may be helpful. The second one, although highly technical, conveys a sense of how complex it all is, and by implication (as I see it) that the effects of digital cables and digital interfaces should not be thought of in the same kinds of ways that we use in describing the effects of audio frequency analog cables, e.g., "overly warm" (notwithstanding the fact that a given cable may create that perception in a given specific system):

http://www.positive-feedback.com/Issue43/jitter.htm

http://www.scalatech.co.uk/papers/aes93.pdf

Regards,
-- Al

I'm not sure what happens with balanced cables. Impedance is 110 ohm and voltage levels are much higher but at the same time slew rate is likely higher and reflection induced jitter taking over noise induced jitter. Maybe Al can help here?

I'm not sure either, mainly because I don't have any specific knowledge of what risetimes/falltimes/slew rates tend to be for typical AES/EBU outputs. My suspicion is the same as yours, though, that those parameters are likely to be faster than for typical S/PDIF outputs.

Also, while on the one hand the higher AES/EBU voltage levels (assuming the particular equipment in fact conforms to the AES/EBU voltage standards) and the balanced operation would seem likely to help with respect to noise-induced jitter, on the other hand I would expect that in many or most cases balanced digital cables will provide less accurate control of characteristic impedance than a good 75 ohm coax will typically provide.

Excellent elaboration in your posts, btw, on the distinction between noise-induced jitter and reflection-induced jitter, and the competing tradeoffs that result.

Best regards,
-- Al

Rpg, 1 meter is about 39.37 inches. And of course 12 inches = 1 foot.

Therefore 3' is about (12 x 3)/39.37 = 0.91 meters, which DOES NOT conform to the length recommendation.

6' is about (12 x 6)/39.37 = 1.83 meters, which DOES conform to the length recommendation.

I've talked about these cables earlier choosing the 3' cable as the champ, but now I'm finding the 6' cable easier to listen to. How can this be explained and may the 6' rout look better on paper and on the ears.

Chances are that less jitter is present with the 6' length, but as explained earlier that is not a certainty. Also, as explained in the section entitled "Jitter Correlation to Audibility" in the Steve Nugent paper I linked to a few posts above, depending on its spectral (frequency) characteristics jitter can sometimes be euphonic in character, and/or mask or compensate for inaccuracies elsewhere in the system.

So the bottom line obviously is to try to decide based on listening, but I would say that if the cables sound different but neither can be determined with confidence to be "better," chances are it would be best to go with the longer one.

Regards,
-- Al

With this cable manufacturer, I can order any length, is there an "ideal" length, or are there too many system dependencies to determine this?

Summarizing all that has been said, I would put it that a length of 1.5 meters (which is virtually the same as 5 feet, to within less than an inch), stands a better chance of being ideal than any other length. But as have been cited there are many system dependencies and cable type dependencies which conceivably could result in other lengths being better in some cases.

In this case, if you decide that the 6' cable you have is preferable to the 3' cable, my suspicion is that changing the 6' cable to a 5' cable of the same type would be unlikely to make a significant difference.

Regards,
-- Al