Sunday, 29 December 2013

MEASUREMENTS: USB Cable Extension with Ethernet Cables - does it worsen jitter?

I mentioned in my "look at the sound room" post the other day that in the Emotiva XSP-1's multichannel "HT Bypass" mode, I'm dealing with some noise when the USB DAC (TEAC UD-501) is connected. In an attempt to fix the problem, I've tried a number of ways to minimize the problem - different USB ports on the motherboard, USB3 vs. USB2 ports, different USB cables, ferrite cores attached to the USB cables, even put in a USB3 PCI-E card... I've even dissected an old USB cable to disconnect the 5V line but it looks like the TEAC needs this present. All to no avail - the high pitched whine and buzz continued.

I then began looking at alternate options; here are a couple I found. The Firestone GreenKey looks interesting. I'm just not sure if it supports high-speed USB2 however which is important since the TEAC needs high speed for the 24/192+ PCM and DSD/DoP sample rates. There are other galvanic isolators like this one but they're just "full speed". Another option is with a better USB card like this Sonore SOtM PCI/PCI-E to USB which is often talked about on sites like Computer Audiophile. However, I really don't see myself spending $400 (with taxes and shipping) for a single port USB card; and I'm not even sure this will do the job I need! (Anyone with one of these cards can comment? A review like this one is useless for my purpose when there's really no discussion of whether they tried it in a noisy system and if it reduced audible background distortion.)

Hypothetically, I thought it might be interesting to try a different angle with this... Knowing that the ethernet system uses isolation transformers, what about the USB extenders using ethernet cables? Already, a cheap one was taken apart on Project Gus and it looks like these devices do indeed use standard LAN filters like the LFE8423. [My mistake, that link was for a USB-ethernet adaptor, not a USB extender like this one.]

With a quick search of eBay (here's a current auction), I decided to take a chance on one of these (also found listing on Amazon):
It includes a transmitter (computer end) and receiver (TEAC DAC end). The transmitter side will use the computer's 5V rail and a standard 5V power supply in the plastic bag is used on the DAC side - this of course isolates the power from the computer to the DAC. It also comes with a short USB A-A male cable for the USB connection from the computer to the "transmission" unit.
Notice the 5V power connector on the unit to the left for the receiving device end.
Total cost was ~$60 for the set including shipping from Asia. The unit is capable of 100m (~300ft) transmission distance and specifically rated up to the 480Mbps high speed spec. Unfortunately I don't see an easy way to crack open the little boxes to see what's inside and I figure I didn't want to risk breaking them at this point. Set up was easy and intuitive since there's barely any English in the instruction pamphlet. Plug 'n' play, no drivers to mess around with at all.

So... Did they work for my purpose? Yeah... To some extent.

They did definitely filter out the "computer noise". I can no longer hear the beeps and buzzes when the computer accesses the hard drive or when the CPU is busy processing. Unfortunately the high-pitched whine is still audible when I stand against the speakers; but much reduced - maybe 25% of previous and almost inaudible at my listening position about 9-10' away (ambient background noise is very low in my basement sound room). I tried different Cat5e and USB cables but this didn't make much different. The next time I go to the computer store, I'll see about getting Cat6a STP (Shielded Twisted Pair) cables and maybe shorter 3ft USBs to minimize any potential to pick up interference... I doubt this will make a difference however because I suspect the noise is embedded in the signal from the computer itself rather than picked up due to lack of shielding.

However, the opportunity presented in buying this item is that I can measure what happens to the DAC audio output through this USB signal conversion. If one believes that a digital cable makes a huge difference, then doing this should really create some nasty effects compared to having just a straight computer --> USB DAC cable. We could see diminution of dynamic range if doing this adds more noise to the DAC and furthermore, let's see if the dreaded jitter gets even worse! (Surely it must be awful, right?)


As usual, here's the setup used to measure the TEAC DAC output when I'm using the USB extender:

Win8 AMD A10 HTPC --> stock 3' USB (supplied in box) --> USB Extender Box --> 50 feet Cat5e cable --> USB Extender Box --> 6' Belkin Gold USB cable --> TEAC UD-501 DAC --> generic shielded RCA --> Emu 0404USB --> shielded generic USB --> measurement Win8 computer

For the direct USB situation, I used a 12' generic shielded USB cable (similar in build to the Belkin Gold USB cable) direct from the Win8 HTPC --> TEAC UD-501.

Notice how rather "unfair" this setup is for my needs... In real life I only use a 6' run of Cat5e cable, but to exacerbate any issues, I'm using a 50' run of just generic ethernet cable.



Here's the RightMark summary results. I tested 16/44, 24/96, and 24/192 for your consideration:

Notice how there's no difference whether it's a direct USB cable or the much more complicated USB-Cat5e extender setup. 16/44 as usual measures perfectly and there's not even a hint of loss in dynamic range, worsened noise, or excess distortion using the Cat5e hardware!

Frequency response at 24/192 (16/44 and 24/96 look to be exactly the same as well [not shown]):

Noise level at 16/44 and 24/96 (24/192 looks no different as well [not shown]):

No difference!

How about the dreaded jitter? As usual, let's fire up the spectrum analyzer and have a look at the Dunn J-Test output off the TEAC DAC.

Yet again - essentially no difference.


Okay, short and sweet - using a USB cable extender with a Cat5e system like this one does not degrade the audio output from the asynchronous TEAC UD-501 DAC. Even measured at a length of 50 feet (much more than my needs), there's no problem.

Subjectively I hear no degradation in the sonic output either and as noted above, it does reduce the noise with the analogue multichannel home theatre bypass on the Emotiva XSP-1 so it does seem to filter out some noise originating from the USB port... Not perfectly silent yet in my system but I'm slowly getting there :-). (This noise does not affect my stereo playback at all off the Squeezebox Transporter or TEAC UD-501, just when I listen to multichannel.)

I had a listen last night with some hi-res stereo 24/192 material - John Coltrane's Blue Train Classic Records 2001 HDAD, Neil Young Harvest off the 2002 DVD-A. Also had a listen to some DSD128 - samples from 2L's website. As usual they sounded very good off the TEAC. All the precision I had come to expect from the TEAC with standard USB cabling was there with no hint of resolution loss or evidence of digital error.

Basically, I'm stumped when I read comments on places like Audio Asylum by audiophile "audio engineer" folks who claim that even adding a ferrite core to a USB cable will do terrible things like worsen jitter (supposedly audible!). As far as I can tell, these folks also never seem to throw up a few measurements or describe what method they use to come to such a conclusion... Even if doing this worsened the sound quality, why do they blame jitter as the problem?

As far as I can tell, bits are bits; at least with a decent asynchronous DAC (I suppose there could be problems with old adaptive isochronous units). Asynchronous USB protocols like the one used with the TEAC work as they should to clean up any stray timing errors so long as the buffer isn't over run and the data is "bit perfect" (easily verifiable with a USB hard drive attached and looking for data corruption). Hard to imagine than anyone needs more than a decent (inexpensive) USB cable for quality sound if doing this kind of cable extension doesn't lead to sonic degradation. I do believe that there are many mysteries in this universe outside of the science we understand today, but digital reproduction in the audible spectrum doesn't need to be one of these! As usual, feel free to provide a link to some results if there's evidence to suggest I'm incorrect.

As we wind down the last days of 2013, I want to wish you all a Happy New Year. And of course a healthy and prosperous 2014 ahead...

Enjoy the tunes.

[2013/12/31 Update]

So, I headed off to the local computer store earlier this morning and bought some 50ft Cat6a STP (shielded) ethernet cables to try. I figure if this doesn't make a difference, the higher quality cable can be used for longer runs around the house.

Surprise! It worked... The whine is now gone. The analogue "HT Bypass" is still a bit noisier than straight balanced cable into the Emotiva XSP-1 preamp but at least there's absolutely no high-pitched noise audible with ears right up to the speakers now.

Again, subjectively, everything sounds great comparing straight USB to the USB-Ethernet extender through the TEAC DAC. Objectively, no evidence of deterioration and looks the same as previous results above:

At least for me, it looks like the use of one of these high-speed ethernet cable extenders is a good option to try for those suffering from USB-related noise in the audio system with no evidence of signal degradation (including jitter with an asynchronous DAC).

Happy New Year and stay safe with those late night parties! :-)


  1. [quote]
    I'm dealing with some noise when the USB DAC (TEAC UD-501) is connected. In an attempt to fix the problem,
    [end quote]

    I know it might sound stupid, but I have the same DAC (TEAC UD-501), and had similar "noises" coming from the PC (laptop), i.e. whines and buzzes: after cutting the central pin (ground) of the pocket size power supply of the PC, they all disappeared! (I actually cut an Euro-Shucko adapter).
    Ground loop was the origin of the problem, audible only when using the headphone output to drive a power amplifier, solution I had to think up since I needed a hard-volume controller...).
    A healthy and prosperous 2014! ;-)

    1. Thanks for the note Andrea.

      I actually did the same thing with the HTPC it's connected to with a cheater plug (3-pin to 2-pin) removing the ground connection... It unfortunately did not do the job for me. That's why I had to take these measures :-(. I'll be at the computer shop today to pick up the STP ethernet cable to give it a try,,,

  2. Have you tried a linear 5V supply (one with a real transformer in it and a linear regulator), I assume the supplied 5V adapter is switch-mode and can still induce common mode currents.
    It may not do anything though, this depends on a number of factors when connecting lots of equipment together and some of them being fed from mains.

    1. Thanks for the tip. I thought of this as well... No haven't tried as I don't have one around at the moment.

  3. Don't know if it would work better or worse, but those Monoprice USB cable extenders work just fine even in long lengths even at 24bit 192khz. They are active extenders, though powered off the USB itself.

  4. Really love to see this post, this is very interesting.

  5. Really great experiment. Thank you very much. I have a UD-501 also.