• Extending An Audio Network Using 802.11ac Wireless

    Theoretically wireless networks should have been capable of streaming high resolution audio since consumers started adopting 802.11g 54 Mbit/s WiFi back in 2003. Streaming 8 channels (7.1) at 24 bit / 192 kHz requires roughly 37 Mbit/s of bandwidth. Thus, high resolution two channel audio should have been a breeze to sling around one's house ten years ago. Unfortunately real life stood in the way. The actual throughput of wireless network devices has never been close to the theoretical maximum. Issues such as latency, dramatic signal losses due to distance and home construction variables, and the lack of technology to harness the power of wireless signals has made wired Ethernet the only game in town. Some computer audiophiles are fortunate to have wired ethernet networks connecting their music servers to their Network Attached Storage devices (NAS) anywhere in their houses. Others are stuck placing components in less than stellar locations because they can't stretch an unsightly Ethernet cable across the living room. With the release of 802.11ac wireless routers, access points, and adapters many audiophiles previously constrained by the lack of wiring may finally have a solution for smooth streaming.





    IEEE 802.11


    Computer Audiophiles with USB, FireWire, eSATA, Thunderbolt, or internal hard drives for their music storage probably won't care much about 802.11ac. These users don't require much network bandwidth for fetching album art, track titles, and other metadata from the Internet. Plus, remote control apps are far less than demanding of one's wireless connection. However, these users may be interested in moving to a Network Attached Storage (NAS) device to move hard drives out of the listening area, enable access to stored music from any device on the network, and take advantage of native NAS applications such as DLNA /UPnP servers and network backup. Those of us already using NAS drives may be interested in cutting the Ethernet cable to allow more freedom in component placement, clean up the room aesthetics, or any number of valid reasons. Sure there's no free lunch. Many audiophiles refuse to use any wireless because the devices emit quite a bit of Radio Frequency noise into an environment where people go so far as to elevate speaker cables off the floor or hang tiny cups on the wall in an attempt to ring out that last 0.00001% of sound quality. Purposely introducing RF seems antithetical to the aforementioned approach.

    Some basic networking knowledge will assist those not as learned as some of the experts who read the pages of CA. To make the discussion easier I will use Mbit/s when discussing network speed. That stands for megabits, not megabytes. Network traffic is always measured in bits whereas hard drive storage is always measured in megabytes. A Byte is 8 times larger than a bit. Thus a network speed of 10 Mbit/s is equivalent to 1.25 Mbytes/s. Wired Ethernet connections have traditionally been 10 Mbit/s, 100 Mbit/s, and 1000 Mbit/s. Some of the cheapest laptops still feature 100 Mbit/s Ethernet cards but for the most part 1000 Mbit/s is the current standard.

    Wireless networking for the most part has come in speeds of 11 Mbit/s (802.11b), 54 Mbit/s (802.11a/g), 150-450 Mbit/s (802.11n), and 866-1300 Mbit/s (802.11ac), and 7000 Mbit/s (802.11ad). Over the last several years wireless streams have been combined to improve performance. For example combining two streams of 150 Mbit/s (802.11n) gives the user a 300 Mbit/s (802.11n) connection. One important item to remember is that data rates are to wireless networking what CPU speeds are to general computing. Alone neither CPU speed nor WiFi data rate mean very much because there are many other variables that help determine the final outcome. I look at the wireless data rates with a grain of salt. If my old wireless router supported 54 Mbit/s and my new router supports 150 Mbit/s I hope to see roughly a 3x increase in speed. That number usually has nothing to do with the theoretical data rates.

    802.11ac will be introduced to the public in two phases. The first phase of 802.11ac products will reach speeds of 1300 Mbit/s. The second phase products will reach up to 7000 Mbit/s. 802.11ac also operates solely on the 5GHz spectrum whereas 802.11n used both 2.4GHz and 5GHz. Phase 1 802.11ac products use a maximum channel width of 80MHz versus 802.11n's 40MHz max. "AC" is also more efficient because of the increase in data rates of its modulation scheme from 64 QAM to 256 QAM. The second phase of "AC" product will increase channel width from 80MHz to 160MHz, increase the number of spatial streams from three to eight, and allow multipul clients to communicate on a single channel simultaneously.

    A note about 802.11ad: This standard operates on the 60GHz spectrum. Although it's capable of 7000 Mbit/s the range is limited to line of sight only. Future computer and electronics could use 802.11ad rather than USB or any other short-wired connection.



    802.11ac In The Real World


    The specs above look great on paper but can 802.11ac devices really perform near the theoretical maximum speed? Even more important, can the devices stream high resolution audio eliminating the need for wired Ethernet? The answers are no and yes* (with an asterisk). To test 802.11ac wireless in my system I used a couple different configurations.


    Configuration One:

    In this configuration I used two Apple AirPort Extreme 802.11ac wireless devices to bridge the gap between the computer connected to my audio system and the rest of my network, including my NAS, where all my music is stored. The entire network is wired with the exception of one 1300 Mbit/s 802.11ac wireless link. In simple terms, the NAS storage is wired to an AirPort Extreme, this AirPort Extreme is setup as an 802.11ac 5GHz access point, another AirPort Extreme is setup to extend the 802.11ac 5GHz network, and my computer used for audio playback is wired to the Extreme that extends the network. The following diagram should make much more sense than my simplistic description.



    Configuration Two:

    In this configuration is placed an Intel Dual Band Wireless-AC 7260 mini PCI express card in a CAPS server. The CAPS servers have slots for both mSATA and mini wireless cards in addition to the standard PCIe x1 slot. Currently this 7260 wireless card requires the 64 bit version of Windows 8. I removed the AirPort Extreme that was setup to extend the 802.11ac network because I the new internal network card is capable of 802.11ac, albeit at only 866 Mbit/s. I configured the CAPS server to join the 5GHz 802.11ac network setup for my computer audio components only. In simple terms, the NAS storage is wired to an AirPort Extreme, this AirPort Extreme is setup as an 802.11ac 5GHz access point, the CAPS computer is set to join this wireless network. There is no physical network connection near my audio system. The following diagram depicts this configuration.




    Results


    I played PCM files from 16 bit / 44.1 kHz (Bitrate 1411) through 24 bit /192 kHz (Bitrate 9216) and DSD files at 1 bit / 2.8224 MHz (Bitrate 5644). My DSD files are in the DSF file format and my PCM files are in the uncompressed FLAC format. I tried both memory playback and regular playback when using JRiver Media Center version 19.0.23 on Windows 8. I used iTunes version 11.04 on my MacBook Pro retina running OS X 10.8.4 with the Thunderbolt to Ethernet adapter for a wired connection. The Linux machines all ran a version of Debian 7.0-7.1 and MPD 17.04 to 18.x.

    A layman might think the only item of importance is file size. Thus if the bandwidth is present the file will play just fine. However, that's not the case. There's more to this than bandwidth and file size. Each playback application handles different file formats differently. I ran into an issue playing 24/192 FLAC files through JRiver v19. The files were not playing right as there were several dropouts during each track and MC was buffering about every 30 seconds. I began a day long email exchange with Matt at JRiver in an attempt to track down the issue. After several tests we concluded that reading a file over a wireless network in small packets is orders of magnitude slower than reading a file with large packets. The FLAC decoder used was reading 8 KB at a time by default. Matt sent me a couple test dll files with different packet sizes. The larger sizes such as 32 KB fixed the issue immediately. There's more nuance to this whole story including what JRiver did to remedy the issue at in this thread at the JRiver forum (Link ). There was no way to test 24/192 FLAC file playback with iTunes but I had no issues playing AIFF at high resolution through my MacBook Pro retina. My Linux machines had no issues with 24/192 FLAC playback.

    Configuration one worked very well. I believe this is because 1. The wireless connection between AirPort Extremes was theoretically 1300 Mbit/s and 2. The Extremes have no driver configuration issues as they are appliances that do one thing only. In this configuration I played 44.1 through 192 PCM FLAC files without an issue once I updated to JRiver Media Center 19.0.23 or newer. DSD playback was very close to perfect with the minor exception of buffering problems less than 1% of the time. I believe a wireless configuration like this is ready for prime time. The only hinderance for users will be the distance between wireless access points and the materials such as drywall, wood, concrete, or metal used in the structure of one's listening environment. My AirPort Extreme were spaced about twelve feet apart with no walls in between the two units.

    Configuration two was very close to configuration one but wasn't quite ready for prime time. Playing 24/192 PCM the first track would biffer for about four to five seconds before playback started. When selecting another track playback began immediately. using 192 kHz PCM I experienced some buffering issues lasting about four seconds during playback of 10% of the tracks. On about 0.05% of the tracks I heard static when Media Center moved from track to track. Playback of material 96 kHz and under was flawless. DSD playback in this configuration was a bit more spotty than PCM playback. I currently don't recommend DSD playback with the Intel 7260 AC wireless card. There are no drivers available for this card for OS X or Linux. As such I was unable to test this configuration with other computers.

    Before starting the music playback tests I ran some tests to determine the throughput of different configurations. I used the LAN Speed Test (lite) application running on Windows 8 (Link ).

    The first test benchmarked my network using 1000 Mbit/s wired Ethernet from a CAPS server to the network attached storage device. I needed a baseline of a configuration I'd been using for awhile. The tests each ran three times, creating a 20 MB file first, a 200 MB file second, and a 2000 MB (2 GB) file last. Obviously the end to end wired network scored far better than any wireless network, even though theoretically the 1300 Mbit/s wireless network is faster than the 1000 Mbit/s wired network. In second place the Extreme to Extreme network extension performed pretty well as evidence by this test and my music listening experience. The performance of the Intel 7260 AC wireless card scored much lower than the previous configurations. Considering on this low performance I think music playback was pretty decent. I would have expected worse if I had run the tests before listening. It's my guess that Intel will improve the wireless driver for this card and users will see a nice performance increase int he future.


    Wired Baseline

    1000 Mbit/s LAN to 1000 Mbit/s Switch to NAS

    Version 1.3.0
    OS Version: Windows 8
    Date: 08/07/2013
    Time: 11:46:44
    Program Parameters: 0
    High Performance Timer: 0.0000005486

    20 MB Test File: M:\temp\NW_SpeedTest.dat
    Write Time = 0.2738627 Seconds
    Write Speed = 584.2344160 Mbps
    Read Time = 0.3944047 Seconds
    Read Speed = 405.6747120 Mbps

    200 MB Test File: M:\temp\NW_SpeedTest.dat
    Write Time = 2.3123852 Seconds
    Write Speed = 691.9262240 Mbps
    Read Time = 3.0877013 Seconds
    Read Speed = 518.1848400 Mbps

    2000 MB / 2 GB Test File: M:\temp\NW_SpeedTest.dat
    Write Time = 23.7368352 Seconds
    Write Speed = 674.0578480 Mbps
    Read Time = 20.5680640 Seconds
    Read Speed = 777.9050080 Mbps


    Configuration One

    1000 Mbit/s LAN to First AirPort Extreme <-(1300 Mbit/s wireless link)-> Second AirPort Extreme to 1000 Mbit/s LAN Switch to NAS

    Version 1.3.0
    OS Version: Windows 8
    Date: 08/07/2013
    Time: 09:12:32
    Program Parameters: 0
    High Performance Timer: 0.0000005486

    20 MB Test File: M:\temp\NW_SpeedTest.dat
    Write Time = 0.4816608 Seconds
    Write Speed = 332.1839680 Mbps
    Read Time = 0.9061395 Seconds
    Read Speed = 176.5732560 Mbps

    200 MB Test File: M:\temp\NW_SpeedTest.dat
    Write Time = 4.3250614 Seconds
    Write Speed = 369.9369520 Mbps
    Read Time = 7.4703014 Seconds
    Read Speed = 214.1814480 Mbps

    2000 MB / 2 GB Test File: M:\temp\NW_SpeedTest.dat
    Write Time = 44.5334348 Seconds
    Write Speed = 359.2806160 Mbps
    Read Time = 45.8594413 Seconds
    Read Speed = 348.8921680 Mbps


    Configuration Two

    Intel® Dual Band Wireless-AC 7260 <-(866 Mbit/s wireless link)-> AirPort Extreme to 1000 Mbit/s LAN Switch to NAS

    Version 1.3.0
    OS Version: Windows 8
    Date: 08/07/2013
    Time: 09:29:17
    Program Parameters: 0
    High Performance Timer: 0.0000005486

    20 MB Test File: M:\temp\NW_SpeedTest.dat
    Write Time = 4.1758672 Seconds
    Write Speed = 38.3153920 Mbps
    Read Time = 1.6027083 Seconds
    Read Speed = 99.8310160 Mbps

    200 MB Test File: M:\temp\NW_SpeedTest.dat
    Write Time = 36.8361493 Seconds
    Write Speed = 43.4355920 Mbps
    Read Time = 15.8819654 Seconds
    Read Speed = 100.7432000 Mbps

    2000 MB / 2 GB Test File: M:\temp\NW_SpeedTest.dat
    Write Time = 402.8244899 Seconds
    Write Speed = 39.7195280 Mbps
    Read Time = 157.6390350 Seconds
    Read Speed = 101.4977040 Mbps



    Wrap Up


    Wireless networks are never as robust as wired Ethernet. Theoretical performance of wireless is always grossly inflated compared to actual performance in a real world application. In addition certain platforms perform differently when a wireless link is involved, as seen with my JRiver example. Advertised speed means very little. Period. That said a proper 802.11ac extended wireless network can meet the needs of many computer audiophiles. Streaming 24 bit / 192 kHz high resolution audio works well under the right conditions. If readers have the option of wired or wireless I still unequivocally recommend wired Gbit/s Ethernet because it always works and its speed is highly predictable. Those who have wireless as the only option will need to plan their networks carefully to get the most out of their music collections.







    Comments 25 Comments
    1. ehb_dave's Avatar
      ehb_dave -
      Great article Chris. Very clear as usual. One small correction, you inverted the conversion factor on Mbits/sec to Mbytes/sec in the third paragraph: 10Mbits/sec = 1.25Mbytes/sec.
    1. DanRubin's Avatar
      DanRubin -
      Excellent piece of work. In the first illustration, you refer to the link between Airport Express units. Do you mean Extreme? Will the Express work in your scenario?
    1. MikeJazz's Avatar
      MikeJazz -
      Very Nice Chris. Thanks.
      Do you consider that with a beagle or a cubox, they fit in scenario 1? So the conclusions are basically the same, your would that be a completely different test case?

      Thanks!
    1. The Computer Audiophile's Avatar
      The Computer Audiophile -
      Quote Originally Posted by DanRubin View Post
      Excellent piece of work. In the first illustration, you refer to the link between Airport Express units. Do you mean Extreme? Will the Express work in your scenario?
      Yes I meant Extreme. Graphic updated. Thanks for noticing.



      Quote Originally Posted by ehb_dave View Post
      Great article Chris. Very clear as usual. One small correction, you inverted the conversion factor on Mbits/sec to Mbytes/sec in the third paragraph: 10Mbits/sec = 1.25Mbytes/sec.
      Thanks for checking my work!
    1. The Computer Audiophile's Avatar
      The Computer Audiophile -
      Quote Originally Posted by MikeJazz View Post
      Very Nice Chris. Thanks.
      Do you consider that with a beagle or a cubox, they fit in scenario 1? So the conclusions are basically the same, your would that be a completely different test case?

      Thanks!
      The BBB and CuBox Pro worked perfect in configuration one.
    1. Aegean's Avatar
      Aegean -
      I've been using the new AirPort Time Capsule 3TB for about a month now. I stored my 24/96 and 24/192 files on TC in iTunes library and accessing it wirelessly through ac enabled network on 2013 MacBook Air 11" and so far results are great and no complaint whatsoever.
    1. Stereoid's Avatar
      Stereoid -
      Excellent review. Thank You!
    1. Bones13's Avatar
      Bones13 -
      Thanks for the review Chris. I recently installed an ASUS router with the AC standard. I use wired for my audio network, but did wonder if the wireless speed was good enough coming up with new devices using this standard.
    1. jrobbins50's Avatar
      jrobbins50 -
      Chris, have you done any experimenting with wifi over power line? Currently, my NAS (Thecus N7700) is ethernet hard wired in my basement to a gigabit switch, which I have hard wired connected to my big rig in the basement (using a CAPS Lagoon server), but also to a 500 mpbs Powerline adaptor to move the data up to the main and second floors of our home. My CAPS1 on the main floor, and my PC laptop/Meridian Explorer system on my second floor, are connected by powerline adaptors on their respective ends. I'm running JRiver 18.XXX with Windows 7 on both of the main and second floor computers. I don't have buffering issues on the second floor, but I do run into the buffering every 30 seconds issue with 24/192 files playing off of the CAPS1 on the main floor. I attributed this to the slower processor in the CAPS1 over the modern PC upstairs. But, based on your work here, would you expect I would eliminate the buffering for the CAPS1 server by replacing the Powerline adaptor with wireless 802ac in your configuarion 1? Thanks. JCR
    1. The Computer Audiophile's Avatar
      The Computer Audiophile -
      Quote Originally Posted by jrobbins50 View Post
      Chris, have you done any experimenting with wifi over power line? Currently, my NAS (Thecus N7700) is ethernet hard wired in my basement to a gigabit switch, which I have hard wired connected to my big rig in the basement (using a CAPS Lagoon server), but also to a 500 mpbs Powerline adaptor to move the data up to the main and second floors of our home. My CAPS1 on the main floor, and my PC laptop/Meridian Explorer system on my second floor, are connected by powerline adaptors on their respective ends. I'm running JRiver 18.XXX with Windows 7 on both of the main and second floor computers. I don't have buffering issues on the second floor, but I do run into the buffering every 30 seconds issue with 24/192 files playing off of the CAPS1 on the main floor. I attributed this to the slower processor in the CAPS1 over the modern PC upstairs. But, based on your work here, would you expect I would eliminate the buffering for the CAPS1 server by replacing the Powerline adaptor with wireless 802ac in your configuarion 1? Thanks. JCR
      Hi JCR - I've heard mixed things about powerline networking. Some people love it while others say it's terrible. Wireless ac may resolve your buffering issues but it really depends on the construction of your house and distance from a wired access point. I'm 99% sure it's not a processing power issue on the CAPS 1.
    1. BVillet's Avatar
      BVillet -
      Chris, Thanks for a really great and (at least for me) very timely article. I noted in Michael Fremer's last Stereophile article that he reported Audioquest demonstrating sound quality differences between audio signals transmitted wirelessly vs those transmitted by CAT. The exact signal chain was not shown or identified. I am curious whether you noticed any degradation to sound quality
    1. The Computer Audiophile's Avatar
      The Computer Audiophile -
      Quote Originally Posted by BVillet View Post
      Chris, Thanks for a really great and (at least for me) very timely article. I noted in Michael Fremer's last Stereophile article that he reported Audioquest demonstrating sound quality differences between audio signals transmitted wirelessly vs those transmitted by CAT. The exact signal chain was not shown or identified. I am curious whether you noticed any degradation to sound quality
      No degradation noticed.
    1. Audio_ELF's Avatar
      Audio_ELF -
      Chris, a couple of things to clarify...

      The CAPS server you put the 802.11AC card in, was that an Atom based server? Could you perhaps try the card in the i7 based server?

      Second, do you have a new Apple laptop with AC to try?

      Your summary would make me think that 802.11AC is capable of the streaming, but there was an issue with the Mini-PCIe card.

      Eloise
    1. The Computer Audiophile's Avatar
      The Computer Audiophile -
      Quote Originally Posted by Audio_ELF View Post
      Chris, a couple of things to clarify...

      The CAPS server you put the 802.11AC card in, was that an Atom based server? Could you perhaps try the card in the i7 based server?

      Second, do you have a new Apple laptop with AC to try?

      Your summary would make me think that 802.11AC is capable of the streaming, but there was an issue with the Mini-PCIe card.

      Eloise
      Hi Eloise - You're mostly correct :~) 802.11ac is capable of streaming but the mini PCIe card has issues. I believe the issue are driver related and possibly related to its dual antenna approach rather than triple antenna approach. I could put the card in an i7 based machine but I don't think that will matter. The CPU wasn't even close to maxed on my Atom. It never got over 50% utilization. Point taken though.

      I don't have a new Apple MacBook Air with the ac card. Although I understand the new Airs have WiFi issues.
    1. elcorso's Avatar
      elcorso -
      Hi Chris,

      I believe all Macs have WiFi issues on this Mountain Lion's time. Maybe we have to wait to Mavericks solution?

      Roch
    1. jerico's Avatar
      jerico -
      Quote Originally Posted by The Computer Audiophile View Post
      Hi JCR - I've heard mixed things about powerline networking. Some people love it while others say it's terrible. Wireless ac may resolve your buffering issues but it really depends on the construction of your house and distance from a wired access point. I'm 99% sure it's not a processing power issue on the CAPS 1.
      FWIW I tried power line networking for a while. Speed wise it was fine, but they caused an audible buzz through my speakers that I couldn't get rid of (isolated circuits weren't an option for me). Eventually I just went all wired.
    1. kevinm's Avatar
      kevinm -
      I have the intel 7260 ac card in a Windows 8 laptop, and it performs flawlessly.

      Great article Chris. I highly recommend an AC network. I have used one for about a year now, which has included streaming video and/or audio to up to 8 devices (not including computers that are on the network) from intranet servers and the internet. Actually, the only devices hardwired to the router are servers and a network printer/scanner. Everything else is wireless.

      We have never had an issue with dropouts or performance.
    1. matt123's Avatar
      matt123 -
      Does anyone have recommendations for non-Apple ac routers?
    1. kevinm's Avatar
      kevinm -
      I use the Linksys EA6500 and WUMC710.
    1. The Computer Audiophile's Avatar
      The Computer Audiophile -
      Quote Originally Posted by matt123 View Post
      Does anyone have recommendations for non-Apple ac routers?
      I've heard good things about the ASUS ac routers. Make sure you get the fastest version.