Asynchronous USB vs Network data transfers

[dpacella]

Hello...I understand that running the USB interface asynchronously has a major advantage over source controlled clocking, namely that in async mode the downstream device can precisely control clock skew as it relates to data alignment and D/A conversion. The data coming off the input buffer prior to the DAC is the same clock that handles flow control over the USB. Some hardware process monitors the buffer and asks for more data, ensuring the pipe is full as long as the audio data is streaming. Timing uncertainties are very low, as the skew is well understood and the conversions are happening in the right places. In synchronous USB the upstream master (the PC) clocks this

process. The way I understand it, this is so because the DAC sample rate and USB packet timing are independent. It is in this clock domain crossing where timing errors in conversion become a factor.

 

Network based transfer (squeezebox, transporter, etc) seems equally well positioned in the battle against timing uncertainties. in the squeezebox, for example, there is a 64 MByte RAM that is monitored by hardware flow control in the squeezebox. This ensures that the buffer before the DAC is always full (this buffer provides the elasticity between the data rate coming off the server, and the data going out for D/A conversions). It seems to me that here we have another example of optimal timing; The same clock that drives the data out of the buffer is the same clock that drives the D/A converter. If the same DAC, filters, and analog circuitry were identical between the async USB front end and the network front end it seems to me we could expect identical audio performance. What do you think?

 

[aljordan]

One advantage wired Ethernet based delivery would be natural galvanic isolation from the computer. Even with asynchronous USB DACs, I can still notice differences in performance between different PCs, and it probably has to do with the cleanliness of the USB output.

 

Alan

 

[barrows]

You are correct. Networked DACs have the potential to perform as well as anything else out there. I have heard the Linn Klimax DS, and it sounds very, very good (and John Atkinson's measurements in the Stereophile review confirm very low jitter). The problem for manufacturers is that developing a Network DAC is very difficult. Unlike USB and Firewire interfaces, where the music player software and GUI are in the computer, the network DAC has to have its own music player and GUI internally, so there is a lot of work to do to develop a networked solution.

 

[flatmap]

@Dave Pacella, while I don't know anything about what you're asking, I enjoyed your description. Very clear!

 

@aljordan, I certainly hear differences between different computer sources using the same music player (and version), same OS, and same cpu (though different clock speed).

 

For example I get more noise via my G4 (mac) laptop versus my G4 desktop. While I would say the audibly quiet desktop has other sonic virtues, I also wonder how much this has to do simply with the difference in power supply.

 

[bordin]

AES WHITE PAPER: Best Practices in Network Audio (pdf)

 

Network audio players are coming near you !

 

[Lizard_King]

I use the M2tech Hi-Face and is wonderful. no need to use what you are describing here. I am using Winxp at this time until I feel the Mac is ready and so are the drivers.

 

[dpacella]

Thanks!

Good history of network technology!!

 

Dave

 

[zeroeight]

I guess we also have to distinguish between wired (ethernet) and wireless (wi-fi) networks. Not sure how much RFI are introduced in wireless networks and how it affects sound. For those that live in apartments these days tough luck.

 

[The Computer Audiophile]

I agree wired v. wireless is a big difference and the number of variables with wireless are much greater. We should also consider that Ethernet is standardized as IEEE 802.3 whereas wireless has usually been 802.11. Thus there is technically no such thing as wireless Ethernet :~)

 

[Elprior]

Hi,

 

nice thread here.

 

IMO, network should be superior to usb for the reasons given above.

 

Now, it's clear that Linn is having a really hard-time developping its own control UI.

And I've been listening to the Klimax DS twice, without being impressed (rather the contrary in fact).

Charles Hansen of Ayre cleary states that wireless is a no-go as far as RFI are concerned.

 

OTOH, there already exists UPnP capable players (JRiver, ...).

The idea of an embedded computer running such a software with proper power supply development and digital/analog ouputs is appealing to say the least. Now that would have to be designed very carefully, so the computer does not become the issue (as in usb mode).

 

Elp

 

[bordin]

Networked audio (both wired and wireless) employs the same concept of the asynchronous USB-based audio in which perfect bits of the music file can be transported asynchronously to the player.

 

Wifi devices with proper implementation should be able to deal with some RFI interference as you can successfully download data file to your notebook computer, i.e. no bit error.

 

However, for near real-time music playback, wifi networked player can tolerate upto certain amount of RFI interference causing interruption in continuous data transfer, producing some audible glitches such as pops and clicks.

 

[Elprior]

I was refering to RFI as polluting the rest of the audio chain, I'm not questionning the usability of tcp over wifi

 

Note that Linn however recommands to use wired connection for 24/192 playback.

That's no good sign since that does not require more than 1MB/s of transfer rate in continuous streaming.

 

Elp

 

[dpacella]

Most modern homes have numerous RF transmitters and receivers. These signals travel through drywall, wood, etc. Therefore, regardless of wireless vs copper streaming network audio device in your rig, your rig is under constant exposure to the HF electromagnetic energy. I guess if one were so inclined turning off ALL sources of RF in one's home would put to bed possible image smear resulting from potential RF fratricide.

Now, as for devices with built in WIFI radio receivers, these have antennas which receive the energy (no harm here), as well as active local osciallotor and other RF LNA front end components. I suppose if the integrated radio receiver were poorly designed, one might be able to measure considerabl RF leakage from the active front end radio components, but since these have to be FCC approved, they must NOT radiate appreciable amounts of RF energy...in other words the manufacturers who design, build, and test these integrated WIFI radios have run tests for EMC compliance. As for me, personally, I have zero concern that the use of WIFI equipped network audio devices (squeezebox. etc) pre-disposes the resulting audio data both pre DAC and post DAC to any distortions.

 

Packet timing, FIFO flow control, DAC clocking etc. are another story.

 

 

 

 

[dpacella]

this is an excellent website. Thank you (Chris) for putting this together and maintaining it so well!!

 

As for PC based music servers finally being legit high end front ends, I think we are entering a fertile growth period, and the async USB seems to be one powerful enabler of this. I also think there is much promise in network based front ends as well. Perhaps the (near) future high end DACS will only have: wifi radios, ethernet, and USB, with the occasional RCA (!) connector for legacy s/pdif transports.

 

 

 

[barrows]

I have confirmed at my home, that external sources of RF interference can, and do affect the audio performance. I live at 9,300' in the mountains, above Boulder Colorado, there is no cell service, and limited TV/Radio reception (very low signal level). There is no high speed internet service, so there are no local wireless networks. My nearest neighbor is well over 200 yards distant, and the area around my home is sparsely populated. In my home there is one main source of airborne RF energy, a cordless telephone (if I have a computer powered up, wireless features on it are shut down). In my environment, my system sounds better when the cordless telephone is powered down. I will not consider using wireless music streaming at home because I have proven to myself that RFI does negatively effect the audio performance of my system.

Obviously, my environment is unusually "clean" from an RF perspective, but I suspect that many people living in stand alone homes would achieve somewhat better audio performance by attempting to reduce the level of RF energy present within the home, especially RF transmitters (wireless devices) in close proximity to the audio system.

Considering the above, I choose not use WIFI devices (remote desktop, Ipod touch, etc.) to control a computer used for audio playback: does anyone know the signal strength, and frequency differences between WIFI and Bluetooth transmission? Would using a Bluetooth device as remote, theoretically, present less of a sonic "problem" than a WIFI device?

 

[aljordan]

Bluetooth range is about 30 feet, as compared to standard wifi which is about 150 to 300 feet. Bluetooth devices are much lower powered. You could probably test the audibility of bluetooth RF by picking up am inexpensive bluetooth mouse.

 

Alan

 

[dpacella]

@barrows: wow you take our OCD prone hobby to a whole nutha level and that is very intersting that you notice audio changes (for the worse of course) with ambient RF present. I also do, to a point, care about electrically clean environment for my gear but I would not even bat an eye regarding efforts in trying to eradicate my house of RF. Whatever the effects, it's something I can happily live with. I would be sad without WIFI!. However, I stand by my statement that, from an engineering perspective, the canned WIFI radios will cause little, if any measureable (dare I even go there) disturbances on the conversion timing of the D to A process. Any RF signals (from a single transmitter or mixed from various different frequencies) riding on PCB traces, speaker cables, interconnects, voice coils , will be so far down in the noise compared to the active signal, not to mention shifted so far above nominal human ear/brain frequency response that it really is a struggle for me to conceive of how it could be an issue whatsoever. Perhaps RF is infiltrating your psychoacoustic experience? There are some cases where folks are proven to have physiologic reactions to RF fields and waves. No Joke.

 

 

 

 

[dpacella]

Cbarrows, I am totally envious. I love that area of the country. I proposed to my wife at the Stanley Hotel near estes park....it was spooky awesome

 

I used to work in Longmont when I was with Maxtor Corporation (we were based out of Shrewsbury Mass, and would go out for weeks at a time) I was working on the read/write design team optimizing transfer rates in high end SCSI drives.

 

regards,

Dave

 

[barrows]

I believe the way high frequency interference (well beyond audible frequencies) creates problems with the music signal (in the analog domain, through pickup via PCB traces, interconnects, speaker cabling, etc.) is through IM distortion. IOW, the high frequency signal modulates the music signal, creating distortions/artifacts in the audible band.

I do not have the expertise to comment on whether airborne external sources of RFI could affect things in the digital domain (and these circuits are often, but not always, well shielded from airborne interference, as you note)-a digital engineer would have to weigh in on that aspect.

Consider AC power conditioners: the noise they remove from the AC is well above the audible frequency range (as an example, the Shunyata units do not start filtering until well over 100 kHZ) and the level of this noise is quite low in relative level, but the affects of AC line filters are easily audible on music playback.

The differences I hear are subtle in nature, for sure, but they are nevertheless there, and quite repeatable. I would describe the RFI "sound" I hear as things like: too much of a "clang" or "hard" sound to pianos, lack of harmonic detail in delicate cymbal work (or what could be termed the white noise effect-where cymbals lack actual tones), and edgy sound to horns, especially well recorded trumpets and saxophones.

I have been on a bit of a mission to reduce the apparent effect of RFI, and the results have been rewarding for me-most recently I added some DIY RC networks to the ends of my speaker cables (at the speaker) resulting in a slight reduction in glare on female vocals, and very low level high frequency details. Of course, someone living in an urban area, especially in apartments, may have so much RFI around that having a home WIFI network will not make a bit of difference (perhaps copper foil wall covering is needed, a Farady Listening Room!); I would bet most folks in that kind of environment would benefit more than I do from some speaker placed RF filters/dampers.

Yup, I love living in Colorado!

 

[guymrob]

Hi everyone,

 

I was looking could anyone tell me why there is much difference in sound quality btw USB and network audio playback? I recently got myself a Marantz NA7004 network audio and start buying Hi Res Flac 24/96 music downloads. Initially I played on USB hardisk connected to the USB port and the sound quality sounds as good as SACD; I also owned many SACDs and playback on Marantz SA-KI Pearl player.

 

When I finished setup my router and Music server and port some of the Hi Res Flac 24/96 and playback on the NA7004, I immediately notice the highs are roll off and overall it sound 'thin' and 'air' in the recordings just vanished. After much unsucessfully tweaking I finally give up network audio and use the USB instead

 

Setup:

Marantz SA-KI Pearl SACD player

Marantz NA7004 network audio player

Adcom GFP-555mkII preamp

Adcom GFA 5500 power amp

Dynaudio Contour 1.8mkII speakers

Cables by taralabs and Goertz

 

[bordin]

Hi guymrob,

 

It would result from differences of the jitter type and level between those two data input channels.

 

Have you also compared USB harddisk v.s. USB flash drive playback ? Try different brands and models as well. You may be more surprised.

 

--bordin

 

[guymrob]

Thanks Bordin,

 

Yes I tried two different type of USB thumbdrives, one Sandisk and Kingston. I have a hard time telling them apart. So I decided to use a USB hard disk due to higher storage.

 

However, the difference btw USB and network becomes much different even when playback 16/44.1 flac. The USB just sounded very much better. I still have no clue.

 

I agree the jitter may play a part here, since USB is a direct connection btw the storage device and the player itself, it may have significant lower jitter. While network require NAS server--> router-->Cat 6 LAN cable--> network player. Not forgetting NAS server is a computer by itself.

 

Do you know where I can get some info regarding on jitter measurements on USB and network audio?

 

--Guy

 

 

[firedog]

The network on it's own can't cause jitter. Jitter is an audio phenomenon, and the network protocols are, by definition, bit perfect.

 

Jitter occurs during/after a network or data signal has been turned into audio. Jitter has to do with mis-timing of the audio signal; the network data transfer is purely data and doesn't have the audio/timing info.

 

Your network based playback may sound worse, but the cause isn't the network per se. You need to look for the cause somewhere else.

 

[Audio_ELF]

Please note that connecting a HDD or memory stick to the Marantz (or other player) via USB is completely different to an "asynchronous USB" connection which is where this thread began.

 

Eloise

 

[ted_b]

network DAC and an asynch USB DAC, you are, at best, describing the differences between USB data storage and a NAS....but most likely simply the differences in how your DAC reads files from the USB port vs how it reads on the network.

 

This is a good subject for another thread, but this discussion is about audio-delivery, not data delivery to the server.