Expert electrical analysis: $340 audiophile cables test “marginal”

Lee Hutchinson - Jul 31, 2015 4:00 pm UTC

Our cable adventure is coming to a close. First we took our two $340 AudioQuest Vodka Ethernet cables to Las Vegas and subjected one of them to a live listening test; listeners were unable to tell it apart from a $2.50 Ethernet cable of the same length. Then we took the cable we didn't use on stage and gutted it, exposing its innards. We found an interesting mix of high craftsmanship (a thick polyethylene sheath, genuine S/FTP construction) and corner cutting (masking tape, unterminated shields).

But listening tests and exploratory surgery would only get us so far. What we needed to cap things off was some actual, for-real electrical analysis, and for that there was really only one place we could go: Kurt Denke and Blue Jeans Cable.

To be sure, we could have rented a Fluke analyzer and done some tests ourselves, but Denke and his company have a sterling reputation in the (surprisingly deep) world of cables. Perhaps most famous for standing up to Monster Cable's lawsuit threats by telling the bigger company to go jump in a lake, Denke and his company produce high-quality tested cables of all kinds—and he’ll also test out your cables to see exactly how well they perform.

After we contacted Blue Jeans Cable and explained what we were doing, Denke agreed to take temporary custody of our remaining AudioQuest Vodka Ethernet cable and also the CableMatters Ethernet cable. Both cables were the same ones used on stage in the listening test as the "A" and "B" test legs.

Denke opened his explanation by saying that BJC isn't equipped to test to "category 7" specifications, to which the Vodka cable is labeled as conforming (indeed, there is no universally agreed-upon "category 7" standard for Ethernet cables—it has an ISO spec, but not a TIA type). However, using a $10,000 Fluke network analyzer, Denke tested out the Vodka to category 6a spec. The results were what can best be described as a "marginal pass."

"While the cable did pass 6A patch cord standards it did so within the tester's margin of error—meaning that if it were run on a variety of well-calibrated testers it might sometimes fail," explained Denke in his analysis. "The difficulty, as is typical for Cat 6A, was near-end crosstalk."

This means that for all the Vodka's sturdy construction and heft, it exhibited near-end crosstalk numbers that were borderline relative to the spec—at worst, only 0.6dB above the spec's crosstalk limit (see What Does This Report Mean? on the Blue Jeans Cable site for a more detailed explanation of the results). By way of comparison, Denke also included a test for an off-the-rack Blue Jeans Cable Cat6a patch cable of the same 1.5m length, which demonstrated a worst case headroom of 1.5dB.

With respect to the expensive Telegärtner terminators, with their individual cable taps mounted to a PCB, Denke explained that they didn't appear to do much from a signaling perspective: "My understanding is that they [the connectors] provide a very high-quality electrical termination," he said. "Interestingly, that isn't what the diagnostics on this cable on the Fluke seemed to show—the HDTDX peaks, which measure locations where crosstalk is high, were as high as on conventional terminations."

Having the signals carried on that length of PCB and not on twisted pairs doesn't appear to make much practical difference, either. "As for the PC board traces," Denke explained, "bear in mind that those can be designed to be 100 ohm impedance lines themselves, and by the nature of their rigid construction, if they're done correctly their impedance stability is superb. I haven't looked to see exactly how the board deals with the 36-45 issue"—that is, the crosstalk introduced by the standard Ethernet pinout—"but I am sure that was an important consideration to the designers."

Finally, the braided shield inside the cable drew some comments. "There is no continuity from the body of the one connector to the body of the other, indicating that the shield has not been terminated to one or both of the connector," noted Denke. "Our 6A uses an absorptive shield—that is, the cable is shielded but the shield is not terminated at either end. Alien crosstalk is the crosstalk which occurs between cables, as opposed to the internal crosstalk which occurs between the pairs in a cable. This may also be why there are unterminated shields on the Audioquest cable—I’m not really sure what the reason is there, though I had thought that the shields on Cat 7 were required to be tied to ground. It is also possible—I have no handy way to test—that they've tied the shield to one end only, though this would be highly nonstandard for network cabling."

And what about our $2.50 Cable Matters cable from Amazon, with its Cat6 rating? Turns out it was a lemon. The cable failed both a category 6 test and also a category 5e test.

"This is quite typical of Chinese Cat 6 that we see, most of which does not pass 5e," elaborated Denke. "We find that the fault is almost always in the cable, rather than the terminations—if we re-terminate the cable we generally get little or no improvement."

As with the Vodka cable, Denke provided us with test results from a "good" off-the-shelf category 6 cable for comparison, which shows a massive improvement over the Cable Matters cable in crosstalk and return loss.

This has an interesting implication for the results of our on-stage listening test. It turns out that we weren't testing the expensive AudioQuest cable against a standard Cat6 Ethernet cable—we were testing it against a really terrible, noisy, spec-failing garbage-quality Ethernet cable.

And listeners still failed to hear any difference.

For all the crosstalk on the wire, for all the potential return signal loss, for all the alien crosstalk the patch cables might have been picking up from lying on the table next to other cables, they still moved the bits; they also failed to introduce enough noise or EMI into the listening laptop's DAC to be audible to the listeners.

There are two conclusions that can be drawn from the BJC analysis and the on-stage testing. First, at least for the sample tested, the $340 AudioQuest Vodka cable only marginally passes the TIA/ISO Cat6a signaling standard. Secondly, when using them to listen to NAS-hosted audio files via SMB, the Vodka Ethernet cables have roughly the same effect on your music as do $2.50 cables that fail the Cat6 spec with unacceptable levels of near-end crosstalk and return loss.

Ultimately, I’m not an electrical engineer—I’m an IT guy-turned-writer, and I’m not an expert in all things cable-related. However, Kurt Denke is an expert. He was kind enough to include a huge amount of detail in his e-mails to me, and I’d like to close out this piece not with my own analysis, but with an extended quote from his.

Our perspective on this sort of thing is that we do not really believe in anything beyond simple spec compliance where Ethernet cable is concerned. If the cable is spec-compliant, and is installed in a network of spec-compliant devices, making it better or fancier or prettier isn't going to improve performance. And it has to be admitted, too, that a certain amount of non-compliance with spec will not always make a difference, especially in a small, low-traffic network where everything is being tested to its limits....

A lot of the things upon which cable is sold in the "boutique" market do not really bear upon spec compliance—the things that make spec-compliant cable are really quite dull (and, fortunately, are also usually much cheaper than the boutique cable!). A great Ethernet cable can be made out of ordinary copper, ordinary polyolefin, and ordinary PVC—because the critical factors in cable performance are not what kinds of materials you use, but how carefully and consistently you use them. The consistency of dielectrics, the consistency of wire drawing, the consistency of pair spacing, the arrangement and consistency of twists, the consistency of the overall cabling operation where the four pairs come together–all of these are tasks that require attention to mundane details, so that every spool of cable comes out with the right impedance and the right relationship between the pairs, spool after spool. There's a little old-fashioned craftsmanship in our job—terminating the cable, which is still a hand-tools operation—but the bulk cable quality is all the product of some rather ordinary process control, mostly fretting about how to keep those dimensions stable.

Thanks for taking the time, Kurt—we appreciate it. The only thing left to do is to figure out what the hell I’m going to do with a spare $340 Ethernet cable.

Listing image by Blue Jeans Cable