I. IntroductionIndeed, "noise" can be an issue with audio systems. There's little worse that can happen to an audiophile after spending hundreds or even thousands of dollars on new equipment, plugging everything in as per "best practice", getting ready to play your first piece of music, and finding that for some reason, noise has seeped through to disturb the expectation of pristine, clean background silence from which music can blossom...
Before proceeding further, remember that noise can sometimes be subtle, and very importantly, ambient noise must be low in your sound room. It's quite possible that slight hum and low-level noise may not be significant from a normal listening distance if the ambient noise level is high (something we urban dwellers especially do have to consider!).
It is not surprising that the more complex our audio system gets, the more potential there is to experience analogue noise as an issue. Sure, we can have digital anomalies ("digital noise" would be meaningless in this context) that cause errors in the transmission through poor cables as I have previously demonstrated, and I guess one could have so severe analogue noise as to cause digital errors (I suspect very rare and likely due to poor equipment, device failure, or extreme conditions). But those anomalies have characteristic "digital" traits as I previously discussed. What we normally think of as noise is the analogue stuff... Characteristic humming at a certain (often mains resonant 50/60Hz) frequency, constant random "white" noise like a poorly tuned radio, or episodic blips and bursts associated with intermittent electrical circuit activity (EMI from computer activity, hard drive access, intentional transmission from cell phone placed too close to equipment, etc.).
If we look at the "big picture" of how noise contaminates audio systems, we can see two major mechanisms:
1. Noise from without: RF/EMI radiation inducted noise because you put an inadequately shielded device or cable close to a source of radiation like a high power computer. Just take an old AM radio and put it close to one of these radiating devices to "hear" the noise. In this case, we have to either get better shielding (no need for fancy cables or expensive ones necessarily), reduce the chance of a cable from acting like an antenna in receiving the radiation (eg. ferrite beads), and/or move sensitive devices like your DAC or pre-amp away from the source of radiation. There are standards for allowable RF radiation from consumer devices so generally this isn't a huge problem except in confined spaces. The inverse square law; placing devices away from the radiating source is you friend. (My father used to run a 1kW Ham radio transmitter in the basement in my high school years with a large antenna in the backyard - now THAT was how I learned the importance of shielding!)
2. Noise from within: Noise conducted within the electrical circuit / cable originating from the equipment itself. High quality audio demands low noise floor so a good piece of audio equipment should not be a major source of noise itself (this is of course why everyone who does measurements always include evaluation of the noise floor). These days, the device that is most likely to add electrical noise to the system is of course our computer in the sound room (hopefully your computer isn't adding much acoustic noise with loud fans). In modern systems with a DAC, conduction through the USB cable is therefore commonly suspect. Instead of just transmitting the electrical "pulses" down a digital cable, the conductor will also carry with it the unintended electrical signal (noise). If it does not get filtered out, noise becomes amplified and transduced in our speakers/headphones where we can hear the unintended distortion. It is this form of noise which can be extremely difficult to deal with since shielding and fancy metallurgical properties will likely not make any difference irregardless of expense. In other words, noisy electrical currents do not care whether it's travelling down copper, silver, gold, or unobtanium once it's in the conduction path. For these situations, the ideal is galvanic isolation which "breaks" the electrical contact but allows only the intended signals to cross - this is more difficult and costly to solve. Although I'm focusing on USB cables here, there are of course many potential sources of conducted noise such as pollution through the power system (eg. washing machine, fridge, light dimmer switches, etc.) and any electrical interface can be affected in varying degrees (coaxial S/PDIF, HDMI, Thunderbolt, DVI, DisplayPort, SATA, ethernet, etc...)
II. The Situation...Now I've got a little situation here which I think can provide a nice "case study" of what could happen, and how we can remedy the issues.
Here's a picture of my setup in the soundroom:
As you can see, I have a number of pieces to this system obviously interconnected with all the wires in the back. The Paradigm Signature S8 speakers are powered by the two Emotiva XPA-1L monoblock amps sitting on the floor with my Technics SL-1200M3D turntable in the center. Signals going to the monoblocks are from the center Emotiva XSP-1 preamp (turned off with the amber LED visible mid-center rack).
I have a center channel on the rack (Paradigm Signature C3) so this is a surround system (5.1) with the rear channels behind out of view and the subwoofer (Paradigm Signature SUB 1) on the left beside the left S8. Surround processing duties are done by the Onkyo TX-NR1009 receiver situated on the bottom rack (between the turntable and Emotiva XSP-1) connected to my computer through HDMI.
Stereo audio is quite straightforward, I have the TEAC UD-501 DAC (top-left rack) and Squeezebox Transporter (mid-right rack) sending signals to the XSP-1 preamp. The Transporter is just a streaming device so it gets data from the wired ethernet network (which as I have shown is generally immune from issues). The TEAC UD-501 however is connected by USB to the computer to the far left of the image (with the white Logitech keyboard on top beside a GIK Acoustics sound panel at the first reflection point).
I use generic (various Mogami and Monoprice) balanced XLR interconnects throughout the system for both the TEAC and Transporter, as well as the preamp to the monoblocks. The noise floor in stereo mode is measurably very low and I really could not wish for any better! The only times I bother with single-ended interconnects in this system is with the Technics turntable to the XSP-1 pre-amp phono input (which sounds fantastic) and when I connect the Onkyo to the Emotiva XSP-1 preamp when I listen to surround sound through the "Home Theater Bypass" input (which as you will see was problematic) .
As much as I love the Emotiva XSP-1 pre-amplifier, there was one issue I ran into from a sound quality perspective early on. It was exquisitely sensitive to noise in the "Home Theater Bypass" analogue inputs. This is significant only when I enter surround sound mode where the output from the Onkyo is sent to the pre-amp for the subwoofer and two front channels (unbalanced signal). The signal would be polluted by both a hum as well as computer-associated EMI noise. So... How does one remove this nasty noise?
III. Ground Loop Noise (Hum)First, let's talk about audible low frequency hum. Most of the time, this is due to ground loops. Here's a nice summary from Associated Power Technologies of what happens. Basically, when a number of electrical components are interconnected, the shielding ties the ground potentials of the equipment together. There would not be an issue if the world were perfect and every single outlet potential was exactly the same. Unfortunately, it isn't perfect and variations in grounding point potentials will lead to small currents injecting noise into the system. Typically, this ground loop "hum" will either be 50/60Hz correlating with the mains AC frequency of your local standard or a resonant multiple (eg. 100/120, 150/180, 200/240Hz, etc.).
In my system, with the highly sensitive "Home Theater Bypass" analogue input, here's what it sounds like:
After making the video, I realized the little camera microphone is having a hard time picking up that low hum (maybe even filtering it out?) - trust me, although faint, it was audible in my room... In fact, I can record it and analyze the spectrum to see this:
There it is, the 60Hz ground loop with resonance up at 120Hz and 180Hz. Like I said, it's faint but I'm a perfectionist and I could hear it in late night listening sessions when the world is asleep :-).
Because the ground loop is caused by differing ground potentials, the easiest ways to deal with this problem is to connect all the devices to the same ground tap. This could be the same power outlet or in this case, I'm connecting them all to the same Belkin PureAV PF60 power conditioner I have been using in the system for many years. Thankfully, this fixed the issue for me as per this second video:
As you can see, I also show another "trick of the trade" if needed in the video... The "cheater plug". Basically, an inexpensive <$5.00 adapter plug that disconnects the ground pin hence removing potential for ground loops through that pin (and also I may add removing electrical noise from the computer/power supply through the ground connector). As I said in the video, be aware of potential safety issues. Also, it may or may not work depending on your computer power supply. Consider using a Ground Fault Circuit Interrupter (GFCI) like this one if you intend to do this longer term. As you can see, they're not too expensive (~$20). An even more permanent and esthetically pleasing method would be just to replace the wall socket with something like this. Isolation transformers would be another option but these tend to be more expensive. Another potential device would be the Hum-X - I've never used one but have heard good things.
With the ground loop inaudible, the spectrum now looks like this:
Beautiful humless noise floor :-)!
IV. USB Cable Conducting Electronic Circuit NoiseIn a way, what I showed above is a bit artificial... The truth is, in order to just demonstrate the hum problem to show you guys that issue, you may have suspected that in the video above, I had already isolated out what is actually even more problematic - electrical noise from the computer circuitry itself, transmitted in the wiring causing distortion where there is susceptibility such as the analogue "Home Theater Bypass" input of the XSP-1. In the video below, I plugged in a typical shielded USB cable to my TEAC DAC directly to show you what the interference sounds like:
Nasty eh? As you can hear, the pre-amp analogue input is picking up all kinds of computer noise and this is getting amplified and sent to the speakers. You can hear JRiver starting up for example; likely a composite of noises from SSD data transfer and program execution. There's also a distracting high pitched whine which actually the camera picked up very well (it sounds louder in the video than hearing it live in the room). This is the infamous 8kHz "packet noise" originating from the high-speed USB "microframe" rate (1kHz for "full"/slow speed, see here under "Frames and Microframes"). And yes, it can be measured:
As you can see from the data to the left, where I put the cursor (8kHz), there is a -92dB peak measured by my E-MU 0404USB. We'll look at this again a little later.
What to do about this? USB noise isolation of course!
To be honest, I have no idea if specific USB audiophile products work. For example, here's a "review" of products from Schiit, iFi, and UltraFi, without any objective context (ie. measurements of a situation in which the device worked). From reading the article I have no idea what has improved sonically other than rather generic and subjective opinions, so how is the consumer supposed to know what to buy? Since I live in Canada and haven't seen any dealers with these units to borrow and test, I really do not want the hassle of ordering and returning if it doesn't actually work for me.
So, I figured, let's just do this logically myself. We basically need to isolate noise from two places: 1.) the serial data pins of the USB cable (D+ and D-), as well as 2.) minimizing noise from the power pins (+ & -).
For the data pins, the best way I can think of is to use an optical technique. This is similar to using a TosLink cable instead of coaxial for S/PDIF back in the "old days" for galvanic isolation. And recently, Corning has come out with this interesting USB 3 Optical Cable (~$100):
As you can see, it's quite long at 10m (33') and meant to be used as a USB extender. However, it is capable of USB 3 (5Gbps!) speeds and so can be repurposed in the future for other duties if I don't need it for my sound system. Also, since it's capable of very high speed, there should be no issue with transmitting DSD128, DoP, and 24/384 PCM to the hi-res USB DAC. Remember that this can be an issue with some USB isolation devices like the Firestone GreenKey which goes up to 24/96 only.
It's important to realize that the optical USB cable does not provide 5V USB power (and thus noise on the voltage pins will not be transmitted) and so it will not work directly with devices that need it to operate (for example USB stick DACs like the AudioQuest Dragonfly or AudioEngine D3). Because my TEAC UD-501 required the USB 5V to be present for the interface, the straightforward way to do this is therefore using a powered USB hub to provide the 5V. I decided to try a spare USB 3 hub I had along with some cables to connect to the DAC:
Nothing special, just one of these generic USB 3 hubs from a local store. I suspect that USB 3 hubs may be a bit better as electrical tolerances probably are superior to standard USB 2 devices in order to accommodate the higher transfer speed.
So... Did the combination of removing ground loop hum and using the optical USB cable (plus powered hub) work? Yes:
Here's how the FFT spectrum looks now (specifically focused at the 8kHz spike):
Compared to the spectrum above, although the technique did not completely remove the 8kHz "packet noise", it is now down to about -120dB, plus the overall noise floor is down another 5-10dB. Basically, there has been a >20dB suppression of the packet noise which has now rendered it inaudible at normal listening levels. Also, no more audible "processing noise" when I open or close programs like JRiver, or do other tasks on the computer.
Remember, the issues I'm showing in this post were always just when the XSP-1 preamp was in the analogue "Home Theater Bypass" surround mode which was susceptible to noise. Normally, I never have a hum or hear electrical noise from the TEAC DAC output directly or through playback in stereo mode from the XSP-1.
For completeness, here are some measurements of the TEAC UD-501 using RightMark Audio Analyzer 6.4.1 PRO comparing the 33' Corning USB 3 optical cable with direct 10' USB 2 wire cable. As usual I'm measuring off my E-MU 0404USB, at "high resolution" 24/96:
No difference; and a few of the graphs if you care for the details (just click to enlarge):
|THD + N|
I remain unconcerned about jitter with modern asynchronous USB interfaces (not that I was ever really worried even in the S/PDIF days about audible distortion caused by jitter even though measurable).
V. ConclusionI hope this little discussion and demonstration using my set-up helps as a practical way to understand noise in a high-fidelity audio system. As expected, suggestions here may or may not work in your situation but I suspect it will help in many cases if you're running into trouble with noise. I have already received E-mail from folks with the Emotiva XSP-1 who have experienced the same noise issue with the analogue "Home Theater Bypass" input.
I repeat. I have never run into these noise issues with the direct DAC output! It is only with the analogue bypass mode on this pre-amp that I hear the noise. Here then is a good example of how the "system" as a whole can be susceptible to analogue noise... The digital transmission itself is fine and in fact the DAC output does not show the 8kHz contaminant nor ground loop hum. And if I did not set up a multichannel system with the XSP-1 preamp, I would never have realized there could be a problem.
A few conclusions then:
1. If you hear a hum; usually low frequency 50/60Hz (+ resonant multiples), think ground loops resulting from the AC power configuration. First thing to do is to reconfigure the power cables and see if you can connect the devices as close to each other as possible. Using a common power bar/power conditioning device worked for me. Even if you don't use a "cheater plug", it's worth having just to temporarily diagnose a problem by "lifting" the ground pin off various devices to have an idea of the "culprit" device. They're cheap. (Another link for those who want to read more about ground loops.)
2. Yes folks, computers are noisy! Indeed, the analogue noise can be transmitted down cables like the USB cable and cause audible interference. I demonstrated how the Corning Optical USB cable can be used to minimize the noise originating from the computer. It's logical. It can be measured. No magic.
3. Despite the use of the optical cable, the 8kHz high-speed USB "microframe packet noise" is still present but much reduced in my "case study". Not audible anymore, so that's good enough. It is possible that this residual signal is originating from the generic USB hub (PHYsical transceiver chipset) I'm using to provide the 5V power. It is possible that a device like the Uptone Audio Regen could result in a better, lower noise outcome. Another possible solution could be a high quality USB card like the oft-mentioned but expensive SOtM product. These solutions need to be objectively tested however and I suggest evaluation using a set-up like what I have here where there really is a noise problem rather than folks just plugging them into a system where there might not have been an audible issue to begin with!
4. Objective measurements of analogue output from my DAC using the optical USB cable shows no difference compared to a shielded 10' USB 2 wire cable. Likewise, no difference in the J-Test FFT to suggest audible jitter anomaly at 16-bit/44kHz or 24-bit/48kHz. Rest easy if you have one of these 10m (33') cables to connect your audiophile DAC. I have been using the Corning optical cable for almost 2 months now and it has worked flawlessly with DSD64/128, and PCM all the way to 24/384 with the TEAC DAC. Subjectively it sounds no different from any other functioning USB cable I have tried.
For those who may have followed this blog for awhile, you may recognize that this is not the first time I wrote about a device that reduced noise with the "Home Theater Bypass" mode of the XSP-1 preamp. More than a year ago, I bought an inexpensive USB to Ethernet extension device that actually did the trick. I did not measure the 8kHz level with that but I don't believe it attenuated the noise as well as the optical cable shown in this post. It's still a fine solution for my problem until this optical USB 3 cable caught my eye!
If you look at the measurements in that previous post, you will see that I could not find a difference in the DAC analogue output there either between a direct USB feed compared with going through many feet of ethernet cable. Again, just more evidence that a good DAC is likely not going to sound any different between USB cables even when the data is processed through transceivers to convert to ethernet or optical fibres as in today's post. Subjectively I can't say with any certainty that I have heard differences.
I have been planning to write this piece for about a month now so I think it's interesting that this post just came out the other day. There are some interesting links in there, but I'm ultimately unclear of the practical intent other than to remind us that noise can cause problems and can be transmitted through a digital cable. Sure, it's possible. Let us consider this concluding paragraph for a moment:
"One approach is to just shrug this information off based on the belief that EMI and RF noise levels in our home and in our network-connected hi-fi are just not worth worrying about. There's no experience like no experience. Another approach is to try different solutions including different digital cables and see if they make a difference. There's no experience like experience."Seriously, are those the only approaches? Why so black or white!? May I suggest a third approach, and IMO best approach... Become "more objective" in the analysis. "Experience" is nice but fully made valuable when paired with understanding of WHY something works (or not) especially as regards to scientifically engineered devices. To not explore why something is better than another (eg. Hmmm, it seems my system could be noisy. I tried and I think this fancy cable sounds awesome compared to that fancy one... Don't really know why... Just try it and experience for yourself!) is akin to superficially "shrugging off" an opportunity to learn; to understanding oneself and the technology behind higher fidelity. Nobody needs to enter this endeavour essentially blind using just trial-and-error! Why not try a bit of objectivism and confirm what is "heard"? Yes, develop experience but through understanding the system and showing others in what way a device (like the optical USB cable or if one believes in a $166/m USB cable) provides value. Demonstrate the actual problem and potential magnitude of the problem so the reader can evaluate with some confidence that they should give something a try because of a demonstrable deficiency that was improved rather than vague testimony. That IMO is what a review should be about, especially for passive devices like cables which as we know can be quite a financial "investment"! (I hear the term "accountability journalism" being tossed about; not sure what to make of this for the majority of audiophile "reporting" these days.)
One final and I hope obvious comment on this noise "issue". Don't worry unless you hear a problem! I've read comments over the years from some and you'd think unless someone invested in thousands of dollars of power conditioning, dedicated electrical lines, expensive cables, esoteric devices (cable lifters?) and removed every switching power supply within sight of the system, one would never achieve a system free from noisy distractions! Nonsense. I bet that the vast majority of setups employing reasonable equipment, decent cables, connected together in a logical fashion are just great as is. It's the systems with numerous components hooked up (like mine with both a pre-amp and surround processor switching between modes of operation with some vulnerability to noise), or maybe highly sensitive systems like very low voltage phono cables, step-up transformers, and phono-preamps that are most susceptible.
Okay. I hope this post has been useful. Let me know of corrections you would suggest and other interesting links. I'm sure there are many things missed and other ways to control noise I have not mentioned; remember, it is a "case study" so what I discussed here works for me and I bet the techniques (like using the optical USB 3 cable) will work for many out there also.
I recently watched an interesting documentary from BBC - "When Albums Ruled The World" (2013). Lovely archive footage and historical review:
The sun's out and the Springtime temperatures are gorgeous on the West Coast... Time to head off...
Enjoy the music and please don't forget the ongoing Digital Filters Blind Test! Get me your results and tell your high-resolution audiophile friends to try! Come, experience...