|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
71.85.146.14
I'll preface my question by stating that I have very little technical knowledge aside from being an avid googler. Having said that I have always been a little curious as to why usb seems to be becoming the default audio connection for computer audio. I've seen mentioned multiple times the sample rate limitations due to bandwidth and native driver problems or lack thereof in the case of USB2 support.Thus I was always a little bewildered as to why firewire didn't become the de facto choice for the high end audio world despite the fact that it is less ubiquitous than usb. Although it never really caught on, I know there was an audio spec (A&M it seems to be called
http://www.ambisonic.net/dvd1394.html) for consumer devices which even went on to supposedly pass multichannel DSD if manufacturers such as Esoteric are to be believed. Further research into driver support eventually led me here (http://www.mlancentral.com/mlan_info/faq.php) which though designed by what appears to be Yamaha, and despite having no connection to the CE audio spec, seems to be a fully integrated audio protocol for multichannel audio and clock information over firewire with built in windows and mac support.Not to muddy the already confusing waters, but am I overlooking something in the specification, or perhaps the royalties which might be involved in the use of mlan. Anyhow, until I can afford the cool toys you guys make, I have to amuse myself somehow. And since I'm a mac user with some extra firewire ports hanging about I'd love to see some 24/96 firewire audio goodness eventually sprout from my computer. Discuss...
Follow Ups:
Just to complement all the good data that Steve and Gordon have supplied I want to mention that the A&M spec for audio over firewire is using isochronous streams as well.All the issues with a PLL and jitter on the receiving side are exactly the same as with USB. I believe the 1394 trade association has published a spec.
http://www.1394ta.com/Technology/Specifications/Descriptions/AVC_Rate_Control_10.htm
that describes rate control via secondary isochronous stream back to sender to slave the sender to the clock of the consumer. This is pretty identical to the USB audio spec.
Unfortunatryl similar to the situation in USB I don't believe there is much hardware support for this extension although I have heard some mentioning that this is supportend by some of the high end firewire equipment by Pioneer.
Cheers
The interesting thing about the mLan stuff is the multi-channel support. However, with mLan, you need licensed Yamaha solutions and custom implementations. I'm certain that it is possible to have low-jitter Fireware just as you can have low-jitter USB. There is just a larger investment in time and NRE to do this than for USB. There are a few firewire interfaces, one from M-Audio for instance, but I believe they probably use the chip that Gordon mentioned.The limitations in USB are really not in the interface or hardware currently. They are in the software I believe. If you use USB on a MAC, it is virtually problem-free, easy to set-up and you can do other work on the machine concurrently. I believe once we get the kinks ironed-out with Vista and have drivers available, that USB on PC will be just as easy and problem-free.
Newb,Ok first off the reason there is no USB 2.0 audio spec is that too many people want too many things. Midi, 7.1, mixing, you name it. It's a quagmire that some silicon companies just don't want to walk in.
Firewire is the choice for companies who make 10K or more of something a year. That is because there are many Firewire capable ARM parts that can be used with drivers to do just about anything.
Oxford is the only company making a commercially available part with I2S outputs. The problem is the jitter on these things is through the roof. You can do a bunch of stuff and get this too work and it is supported to 24/96 on both MAC/PC from what they say.
USB 1.1 spec does allow up too 24/96 it also alows for ASYNC and ISO usage. The problem is that there is only really 2 chips (Both from TI TAS1020B & TUBS3200) available and as I am finding out is a frustrating process over the last 9 months.
There are other ARM parts that are more generic from Atmel that can do the same thing as the TI parts. But the support and stuff to get these to work would take a good chunk of time.
Yesterday I had a little free time and was optimizing and trying out my new dScope analyzer with some software changes in the TAS1020B. The results were very promising. So I plugged the unit from my MAC over to the PC to see how it would work there. Well the damn PC rebooted every time I plugged it in. AHHHHHHHHHH
If you want multi channel support Firewire is the place to go. It just won't be that much quality stuff unless they invested heavily in external VCXO's for all frequencies.
Thanks for the reply Gordon. Obviously, the limiting factor is going to be commercially available chips that make all this possible, and although I'm not surprised at the jitter problems with the oxford parts (as they are targeting low end consumer products), I did think there were more options out there. Searching some more I hit upon the Yamaha NC1 and the Dice II from wavefromt semi (http://www.wavefrontsemi.com/index.php?id=12,19,0,0,1,0) which both seem to support i2s outputs. Granted as these are made to target the pro audio set, their features are overkill for audiophile usage, however, I would think that by the same token their jitter capabilities would be much improved. I would also think you shouldn't have a problem getting ahold of the chips, although there too, I might be wrong.As a final question in trying to fully grasp the issue of jitter, I was wondering a little about VCXO's. I understand that chips need to create their own clock which is the function of the VCXO in tandem with PLL loops (or at least I think I do). Are the VCXO's on the TI chips themselves or do all chips simply provide a clock input for external VCXO's? If the latter is the case, do the TI chips simply provide a better implementation of PLL, or do chips like the Oxford parts not use this at all? In any case, thanks for your comments and feel free to be brief as I'm just curious about how all this happens.
Newb,I am not sure what the NC1 and Dice are but I don't think they are going to offer chips solutions. The NC1 goes for $300 each. That is a bit much just for the chip and drivers.
VCXO's are set to a freq like 11.2896 (most typical for 44.1K MCLK). They vary with a DC voltage applied to the control pin so they can be matched to the incoming stream with a range of around 100-200ppm (Parts per million). On the 1020 and 3200 from Ti the PLL is actually done in software and is pretty good at reducing the intrinsic jitter to a low amount.
Remember in itself Firewire and USB do not have jitter because there is no clock mutiplexed over the data like there is in SPDIF.
So intrinsic jitter is anything that exists because of the creation of clocks or the way it's made etc...
Bettering these parts usually means using ASYNC with a low jitter clock or external VCXO or custom stuff.
While you are waiting for the big dogs to show up, let me give you some market context.The 1394 Firewire standard was originally developed and brought to market by Apple. It was envisioned as a cost effective, more flexible plug and play solution the SCSI with the capability to chain up to 60 drives. If you didn't live through SCSI, each drtive had to have a unique ID number with a max limit of 9 drives. SCSI was and still is a black art, while FW is plug and play making it much better suited for physical movement between systems and hot swapping.
Given the time frame, data transfer rates were considered to be blazing fast for the price point. (Ultra SCSI remains the speed champ) Plus the FW standard was optimized for large file transfers such as one would use for digital photos, pre-press etc - in other words much bigger files then anything required for audio playback.
There was only one problem. Intel didn't have a piece of the action. So Intel launched a competing standard USB1 which was slower then FW. I don't remember for sure but its not hard to imagine that MSFT was quick to support them with the necessary drivers. In all fairness, I believe that USB was imagined as protocol that could be used with a much wider range of devices then just drives. For example Apple soon started using, and still uses USB for their keyboards, mouse and monitors.
Both standards were advanced - FW400 became FW800; and USB1 became of course USB2 which is almost as fast as FW800 though not quite as good for large file transfers. Meanwhile even uncompressed audio requires nothing faster then USB1
I don't know if there were significant differences in licensing costs or manufacturing costs. But USB took off. Safe to say that today it is very hard to find a computer without USB. There are billions of USB devices out there, while FW is a niche product used by data centers (pre SATA) and digital professionals in photography, audio, video and prepress.
Virtually all the innovation is in the USB space because that is where the market is.
BTW you should include Wikipedia in your reading. And so some searches here because both Gordon and Steve have written on this topic many times before.
Thanks for the info Xmasparty. I did in fact know most of what you were saying, which is why I included the caveat of calling USB the far more accepted interface. You might note, however, that Sony also had a hand in developing the firewire standard, although like most everything they do they went on to make it sound proprietary by calling it ilink. However, since we are audiophiles, a niche market itself, I simply thought that the technically superior firewire standard might confer some benefits, especially since it has already seen usage as a method of transferring high quality audio. Ultimately, however, it seems that you are correct as the market does indeed dictate the availability of receiver chips as noted by Gordon.
This post is made possible by the generous support of people like you and our sponsors: