MEASUREMENTS: USB Cables for Audio DACs. [2013-06-18 UPDATE]

Up to this point, as you've seen in my measurements, I have been using standard generic shielded cables for analogue and digital. As you likely also know - if you've been involved in "high end" audio for awhile - there are few issues as contentious as whether expensive cables are necessary or if they even make a difference!

I've generally steered clear of this "debate" over the years because inevitably, these discussions with friends or forum threads on-line almost never really "go" anywhere...  Nobody leaves the discussion having really learned anything, and ultimately if the discussion turns into an argument, everyone just ends up feeling "pissed off".

The conclusion one usually can get agreement on is that the best cable is no cable. Cables are passive "components", and ideally should transmit signal without loss, and does not impart any kind of signature on the quality of the signal. This certainly makes sense especially in the analogue domain where noise can easily creep in, causing various distortions or raising the noise floor. But what of digital cables? When something as ubiquitous as a generic USB cable can easily transmit >20MB/sec data to a USB hard drive, how problematic can it be to get 150KB/sec across for 16/44, ~500KB/sec for 24/96 audio, or 'just' ~1.12MB/sec for 24/192 high-resolution audio!?

After all these months of testing and posting on various topics, I figure it's at least reasonable to explore the cable issue as best I can with (admittedly) the limited selection of choices I have available around here.

For the record, up to this point in time, the most expensive USB cable I have tried has been the AudioQuest Carbon (~$160 for 5') more than a year ago connected between the PC and a Benchmark DAC1 USB. I have since sold off both items. I could not clearly hear a difference with the AQ Carbon, but then again since the DAC1 uses ASRC, I really wasn't expecting much either. Since then, I've been satisfied with the selection of generic cables I've collected over the years as computer parts or 'freebies'. It is in this light that I selected among my assortment, 3 A/B USB 2.0 cables to test out:

USB Cable A:
- 3 feet long
- not gold plated
- ferrite core on the B end

USB Cable B:
- 5 feet long
- gold plated A & B ends
- no ferrite core

USB Cable C:
- 17-feet long! (6 feet Logitech extender, 5 feet Linksys extender, 6 feet generic cable)
- not gold plated
- possible ferrite core on the extender connections?

As you can see, Cable C was made by Dr. Frankenstein. It's unreasonably long (USB specs indicate a reasonable limit at ~16'), has 2 detachable connections in between, is not even gold plated, and who knows what kind of shielding is in that thing. Furthermore, the Logitech extender portion has the thinnest wire of all the cables. Any card-carrying "audiophile" would be laughed out of town for even suggesting they use something like this!

I. Cables tested with Asynchronous USB 2.0 CM6631A interface:

Setup:

Aspire Win8 laptop --> *test USB cable* --> CM6631A USB to SPDIF --> Generic TosLink --> AUNE X1 DAC --> silver braided RCA --> E-MU 0404USB --> Aspire Win8 laptop

Let us start with what has been previously demonstrated to be a very robust asynchronous USB interface; the CM6631A. Remember that the previous tests showed that despite different laptops used, the analogue and jitter spectra were almost identical.

With the different USB cables, RightMark DAC analogue analysis (24/96 used):

Frequency Response:

THD:

Stereo Crosstalk:

No difference.

Dunn Jitter Tests (16-bit & 24-bit tests):

Cable A:

Cable B:

Cable C:

No difference.

II. Cables tested with Adaptive Isochronous USB 1 AUNE X1 interface:

As shown previously, this interface can get noisy if I play and record with the Aspire laptop (see the laptop post). Therefore I'm going to use my ASUS Taichi to play the test audio while measuring with the Aspire as usual.

Setup:

ASUS Taichi --> *test USB cable* --> AUNE X1 USB1 --> silver braided RCA --> E-MU 0404USB --> Aspire Win8 laptop

RightMark DAC analogue analysis (16/44 only):

Frequency response:

THD:

No difference.

Dunn Jitter test (16-bit only):

Cable A:

Cable B:

Cable C:

Well, the adaptive USB interface is noisier and more jittery than the asynchronous setup as we have seen before in previous posts. Depending on which cable, we can see noise showing up at certain frequencies. For example, there's a -120dB 11.7kHz spike on Cable A, 11.3kHz for Cable B, 10kHz for Cable C. But these are not the symmetrical sidebands associated with data jitter. In fact, the low-frequency jitter associated with widening of the base around the 11kHz signal looks essentially identical and perhaps most importantly, there's really nothing unusual here to suggest that Cable C is of unusual length and construction!

III. Cables tested with TEAC UD-501 Asynchronous USB 2.0 DAC: (2013-06-18 Update)

For the sake of completeness, I decided to run these same "cables" through to the direct asynchronous USB 2.0 interface of the TEAC UD-501 DAC to see if we can see any differences. Some folks asked if I could try a "name brand" cable for point of reference...  Suggestions included the AudioQuest Diamond, maybe some Crystal Cables... Well, I search high and low... And here's the Belkin Gold 6-ft (1.8m) which I managed to get on sale for <$10 locally :-)

That red stuff is just some electrical tape I put on it for identification on another project.

Here's the RightMark data for 24/96:

Frequency Response:

THD:

Stereo Crosstalk:

No difference.

Dunn Jitter Tests (16-bit & 24-bit tests):

Belkin Gold (6'):

Cable A (3'):

Cable B (6'):

Cable C (17' monstrosity):

As you can see, the jitter plots for the TEAC are quite clean. On the whole, the 24-bit noise floor is measurably lower than the AUNE X1 but the jitter plot does show a few more noise spikes (made more obvious because of that low noise floor). This is really quite academic because we're talking about noise down near -140dB!

No evidence of significant jitter differences between the Belkin Gold or any of the other cables including that terrible looking Cable C (17-footer).

Conclusion:

No evidence in these tests to suggest that the different USB cables used here with the asynchronous CM6631A USB-to-SPDIF converter, direct asynchronous TEAC UD-501 USB DAC, or adaptive isochronous USB setups should sound different (even though one would expect Cable C to be the worst). Subjectively, listening to music with Cable C through Sennheiser HD800's sounded fine. 

No evidence with the J-Test to suggest data-correlated jitter is significantly different between cables. By the way, for a good demonstration of how jitter improves with interface/cable change, look at my post on Transporter-to-Behringer connection with TosLink vs. AES/EBU. Also, remember that my Oppo BDP-105 tests were done with a single 15' USB cable - still better than Cable C in construction :-).

As usual, feel free to drop me a note if there's good data or controlled tests to suggest USB cables make a significant difference contrary to these findings.

Happy listening...