Home |
Search |
Today's Posts |
#41
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
"Arny Krueger" wrote in message ... "Pooh Bear" wrote in message Hi, I'm interested in passing stereo audio over USB in both directions. I believe that Windows already has a 16bit driver for this but I want 24bit. Does this mean one has to write one's own ? It's slightly outside my area of expertise. I'd aslo be interested in how Macs support this function. Any general advice about implementing this, recommended chips etc is very welcome. The quality needs to be high-end / professional rather than consumer btw which I suspect means using separate A/D and D/As. This seems to be a single-channel implementation of what you're looking for. It might not be rediculous to build two. http://hans-w.com/usb_adc1.htm The LTC2400 is not really recommended for high-end audio with its sample rate of about 10Hz... --DF |
#42
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
"reddred" wrote in message
... "Pooh Bear" wrote in message ... I'm not trying to prove *anything* ! I want a solution for a product. You can buy analogue audio mixers these days with USB and Firewire interfaces for example. This is kind of thing I'm talking about. Graham TI makes some chips specifically for usb audio. Don't know which ones are 24 bit. None. All the USB thingys are 16 bit. All the 24 bit stuff has an interface to connect to a DSP. Meindert |
#43
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
"Pooh Bear" wrote in message ... I'm not trying to prove *anything* ! I want a solution for a product. You can buy analogue audio mixers these days with USB and Firewire interfaces for example. This is kind of thing I'm talking about. Graham TI makes some chips specifically for usb audio. Don't know which ones are 24 bit. jb |
#44
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
Meindert Sprang wrote:
"Pooh Bear" wrote in message ... I'm not trying to prove *anything* ! I want a solution for a product. You can buy analogue audio mixers these days with USB and Firewire interfaces for example. This is kind of thing I'm talking about. I did some searching and found this from M-Audio: http://www.m-audio.com/products/en_us/Transit-main.html The Codec used in this unit is the following: http://www.asahi-kasei.co.jp/akm/en/...84/ak4584.html Which implies that some other chip is still needed to format the data and transfer it over USB. Right. Odds are they use the TI chipset with some glue logic between the codec and the thing. --scott -- "C'est un Nagra. C'est suisse, et tres, tres precis." |
#45
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
Pooh Bear wrote:
I'm interested in passing stereo audio over USB in both directions. AIUI, USB has some bigger issues with latency than Firewire, but unless it must be one of those, I think you should look at ADAT with the Wavefrontsemi chipsets - the PC will need an ADAT interface though. |
#46
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
"Arny Krueger" wrote in message ... "Pooh Bear" wrote in message Hi, I'm interested in passing stereo audio over USB in both directions. I believe that Windows already has a 16bit driver for this but I want 24bit. Does this mean one has to write one's own ? It's slightly outside my area of expertise. I'd aslo be interested in how Macs support this function. Any general advice about implementing this, recommended chips etc is very welcome. The quality needs to be high-end / professional rather than consumer btw which I suspect means using separate A/D and D/As. This seems to be a single-channel implementation of what you're looking for. It might not be rediculous to build two. http://hans-w.com/usb_adc1.htm It appears to be a very low-speed device with a sampling rate of 50 or 60 Hz. But there are plenty of example circuits provided by the IC vendors for audio-quality A/D and D/A converters, USB interfaces, etc. |
#47
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
"Pooh Bear" wrote ...
Ok. I need an *embedded* USB audio interface. To repeat ; a chip or chipset. Is that OK ? Probably. But you have likely specified yourself right out of an off-the-peg solution. There are 16-bit solutions because there is a sufficient market for a monolithic integrated solution. But it seems very unlikely that there is a 24-bit solution. For one thing if you have that much dynamic rangeon the analog input side, the close proximity of the digital parts (especially something as "high-power as a USB interface) seems impractical. |
#48
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
Richard Crowley wrote: "Pooh Bear" wrote ... Ok. I need an *embedded* USB audio interface. To repeat ; a chip or chipset. Is that OK ? Probably. But you have likely specified yourself right out of an off-the-peg solution. There are 16-bit solutions because there is a sufficient market for a monolithic integrated solution. But it seems very unlikely that there is a 24-bit solution. For one thing if you have that much dynamic rangeon the analog input side, the close proximity of the digital parts (especially something as "high-power as a USB interface) seems impractical. The actual dynamic range of decent audio converters is in fact in the 20 bit equivalent region. 24 bits is simply a convention based on 3 bytes. Avoiding crosstalk from the digital to analogue section etc is simply a matter of good design. Graham |
#49
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
"Pooh Bear" wrote ...
Avoiding crosstalk from the digital to analogue section etc is simply a matter of good design. Which is not too difficult to achieve with separate IC packages for the A/D or D/A circuit, and another for the USB interface, separated by an inch or so of PC board. But putting all that onto a single chunk of silicon that is likely smaller than 1/4 inch square is an engineering feat that is likely too expensive for a mass-market product. At least that seems to be the message we can get from the absense of such a product. |
#50
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
In sci.electronics.design Pooh Bear wrote:
Rich Grise wrote: On Mon, 12 Jun 2006 15:02:50 -0700, David Nebenzahl wrote: Pooh Bear spake thus: I'm interested in passing stereo audio over USB in both directions. [snip] Amazing. Just ****ing amazing. 34 replies to your question, by my count on my ISP's view of the group, and *not one useful answer to your question*. (Including my two, make that three, posts, I'm sorry to say.) Lots of people saying "**** that, won't work", or "why bother?" and other useful tidbits. Ain't Usenet wonderful? A lot of the answers looked to me like "Why design and build, when you can buy cheaper?" If that's not good enough, then we need a more comprehensive system specification. Ok. I need an *embedded* USB audio interface. To repeat ; a chip or chipset. Is that OK ? Why do you think you need to do the transfer at 24bits just because it is 24bit data? -- Aaron |
#51
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
"Rich Grise" wrote in message news Or are you looking for a one-chip thing, to embed? If so, you should know by now that you should have already said so./snotty pedagogue If that's more along the lines of what he's looking for, I'd try the sci.electronics newsgroups as opposed to the re.audio newsgroups. Jeff -- "They that can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety" - B. Franklin, Bartlett's Familiar Quotations (1919) |
#52
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
"Rich Grise" wrote in message news On Mon, 12 Jun 2006 15:02:50 -0700, David Nebenzahl wrote: Pooh Bear spake thus: I'm interested in passing stereo audio over USB in both directions. [snip] Amazing. Just ****ing amazing. 34 replies to your question, by my count on my ISP's view of the group, and *not one useful answer to your question*. (Including my two, make that three, posts, I'm sorry to say.) Lots of people saying "**** that, won't work", or "why bother?" and other useful tidbits. Ain't Usenet wonderful? A lot of the answers looked to me like "Why design and build, when you can buy cheaper?" If that's not good enough, then we need a more comprehensive system specification. I think that's because the original poster sent his message to a rec.audio group instead of a sci.electronics group. Different crowd, different, but fairly predictable, response. Jeff -- "They that can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety" - B. Franklin, Bartlett's Familiar Quotations (1919) |
#53
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
|
#54
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
Jeff Findley wrote: "Rich Grise" wrote in message news On Mon, 12 Jun 2006 15:02:50 -0700, David Nebenzahl wrote: Pooh Bear spake thus: I'm interested in passing stereo audio over USB in both directions. [snip] Amazing. Just ****ing amazing. 34 replies to your question, by my count on my ISP's view of the group, and *not one useful answer to your question*. (Including my two, make that three, posts, I'm sorry to say.) Lots of people saying "**** that, won't work", or "why bother?" and other useful tidbits. Ain't Usenet wonderful? A lot of the answers looked to me like "Why design and build, when you can buy cheaper?" If that's not good enough, then we need a more comprehensive system specification. I think that's because the original poster sent his message to a rec.audio group instead of a sci.electronics group. Both in fact but never mind. Different crowd, different, but fairly predictable, response. There have been some useful snippets of info though. Graham |
#56
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
Pooh Bear wrote:
I'm interested in passing stereo audio over USB in both directions. I believe that Windows already has a 16bit driver for this but I want 24bit. Does this mean one has to write one's own ? It's slightly outside my area of expertise. I'd aslo be interested in how Macs support this function. Any general advice about implementing this, recommended chips etc is very welcome. The quality needs to be high-end / professional rather than consumer btw which I suspect means using separate A/D and D/As. Graham, Nobody has answered your question, so I will. a) The USB audio-class driver for Windows absolutely DOES support 24-bit audio. I have bus-analyzer traces that prove it So does Mac OS X's USB audio-class driver. You'll need to inform the OS that your device supports 24-bit audio, and you do that using the device descriptors. It's quite simple. b) I haven't seen a USB high-speed audio-specific controller chip. However, the TI chips (TUSB3200A and TAS1020B) do full-speed just fine. You'll be limited to 48 kHz stereo 24 bit if you want to do full-duplex; full-speed USB doesn't have the bandwidth for 96 kHz full-duplex stereo. c) If your design requires separate sampling frequencies for record and playback, then you'll need to use the TAS1020B, as its CODEC interface in I2S mode has two sets of LRCLK, BCLK and MCLK signals. The TUSB3200A's CODEC interface has only one set of BCLK, LRCLK and MCLK signals. A further complication is that if you wish to support more than one sampling frequency (even though both playback and record will always use the same frequency), then you need the TAS1020B and you can't use a CODEC (you need to use separate ADC and DAC devices). This is because the OS considers the playback and record interfaces to be completely separate, but if the hardware has only one MCLK/LRCLK/BCLK interface, clearly they're not separate. The problem occurs when the OS says, "I want record at 44.1 kHz" and the hardware goes, "OK," but the OS has no way of knowing that the playback interface is also at 44.1 kHz. This limitation is why TUSB3200A-based devices like the Edirol UA-5 have a front-panel switch that selects the sample frequency: at enumeration, the micro reads the state of the switch and alters the descriptors appropriately, and the only way to change the sample frequency is to detach the device, set the switch, and reattach. Again, since the TAS1020B has two sets of I2S clock lines, it has no problem supporting different sample frequencies. Is this helpful? -a |
#57
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
Andy Peters wrote: b) I haven't seen a USB high-speed audio-specific controller chip. However, the TI chips (TUSB3200A and TAS1020B) do full-speed just fine. You'll be limited to 48 kHz stereo 24 bit if you want to do full-duplex; full-speed USB doesn't have the bandwidth for 96 kHz full-duplex stereo. I was under the impression that there was plenty of speed with USB2.0 to handle far more than 24-bit 48 kHz full duplex. There are several devices that do 8 channels duplex at at least 48 kHz and some that do it at 96 kHz. Here's one: http://tinyurl.com/ognnb |
#58
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
Mike Rivers wrote:
I was under the impression that there was plenty of speed with USB2.0 You forget that USB2.0 high speed is faster than "full" speed. Go figure... |
#59
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
Andy Peters wrote: Pooh Bear wrote: I'm interested in passing stereo audio over USB in both directions. I believe that Windows already has a 16bit driver for this but I want 24bit. Does this mean one has to write one's own ? It's slightly outside my area of expertise. I'd aslo be interested in how Macs support this function. Any general advice about implementing this, recommended chips etc is very welcome. The quality needs to be high-end / professional rather than consumer btw which I suspect means using separate A/D and D/As. Graham, Nobody has answered your question, so I will. a) The USB audio-class driver for Windows absolutely DOES support 24-bit audio. I have bus-analyzer traces that prove it So does Mac OS X's USB audio-class driver. You'll need to inform the OS that your device supports 24-bit audio, and you do that using the device descriptors. It's quite simple. Excellent ! I presume it'll be the task of the USB device's firmware to tell the OS. b) I haven't seen a USB high-speed audio-specific controller chip. However, the TI chips (TUSB3200A and TAS1020B) do full-speed just fine. You'll be limited to 48 kHz stereo 24 bit if you want to do full-duplex; full-speed USB doesn't have the bandwidth for 96 kHz full-duplex stereo. I can probably manage with that if that's what other similar devices have to do. c) If your design requires separate sampling frequencies for record and playback, then you'll need to use the TAS1020B, as its CODEC interface in I2S mode has two sets of LRCLK, BCLK and MCLK signals. The TUSB3200A's CODEC interface has only one set of BCLK, LRCLK and MCLK signals. A further complication is that if you wish to support more than one sampling frequency (even though both playback and record will always use the same frequency), then you need the TAS1020B and you can't use a CODEC (you need to use separate ADC and DAC devices). This is because the OS considers the playback and record interfaces to be completely separate, Does it indeed ? Does this mean that the sample rates for p/b and rec don't have to be synched then ? I was wondering about generating wordclock as it happens. but if the hardware has only one MCLK/LRCLK/BCLK interface, clearly they're not separate. The problem occurs when the OS says, "I want record at 44.1 kHz" and the hardware goes, "OK," but the OS has no way of knowing that the playback interface is also at 44.1 kHz. Uhuh. Understood. This limitation is why TUSB3200A-based devices like the Edirol UA-5 have a front-panel switch that selects the sample frequency: at enumeration, the micro reads the state of the switch and alters the descriptors appropriately, and the only way to change the sample frequency is to detach the device, set the switch, and reattach. Again, since the TAS1020B has two sets of I2S clock lines, it has no problem supporting different sample frequencies. Is this helpful? It certainly is. I'd already downloaded the data sheets for both parts, so I'll go look at them in more detail. Graham |
#60
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
Mike Rivers wrote:
Andy Peters wrote: b) I haven't seen a USB high-speed audio-specific controller chip. However, the TI chips (TUSB3200A and TAS1020B) do full-speed just fine. You'll be limited to 48 kHz stereo 24 bit if you want to do full-duplex; full-speed USB doesn't have the bandwidth for 96 kHz full-duplex stereo. I was under the impression that there was plenty of speed with USB2.0 to handle far more than 24-bit 48 kHz full duplex. There are several devices that do 8 channels duplex at at least 48 kHz and some that do it at 96 kHz. Here's one: http://tinyurl.com/ognnb You're confusing "USB 2.0" with "high speed USB." While USB 1.1 had two speeds (low, at 1.2 Mbps, and full, at 12 Mbps), USB 2.0 adds a third speed (high, at 480 Mbps), but a USB 2.0 _device_ does not have to support high-speed transfers. However, note that a USB 2.0 _host_ must support high speed. From the UA-101 data sheet: "High-Speed USB (USB 2.0) and Full-Speed (USB 1.1) compatible The UA-101 is the first multi-channel high-speed USB audio interface to offer basic compatibility with USB 1.1. When connected through USB 1.1, the UA-101 will function as a stereo audio interface at 44.1 or 48kHz with any Win98 or later PC or Mac OS 9 or Mac OS X computer. When connected to Windows XP computer running USB 2.0, the UA-101 offers the full set of I/O options to your computer." What Edirol is saying is that if you connect the UA-101 to a USB 2.0 host (which, as noted above, must support high-speed transfers), then the device will do multiple channels at 96 kHz, as the device will enumerate as a high-speed device. However, if you plug it into an older USB 1.1 host, which doesn't support high-speed transfers, then the device enumerates using a full-speed configuration which only supports 48 kHz stereo. This happens because the USB spec allows a high-speed device to support an "Other_Speed_Configuration," the idea being that it's better that a device work with some reduced functionality in a lower-speed system than to not work at all. -a |
#61
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
Pooh Bear wrote:
Andy Peters wrote: Pooh Bear wrote: I'm interested in passing stereo audio over USB in both directions. I believe that Windows already has a 16bit driver for this but I want 24bit. Does this mean one has to write one's own ? It's slightly outside my area of expertise. I'd aslo be interested in how Macs support this function. Any general advice about implementing this, recommended chips etc is very welcome. The quality needs to be high-end / professional rather than consumer btw which I suspect means using separate A/D and D/As. Graham, Nobody has answered your question, so I will. a) The USB audio-class driver for Windows absolutely DOES support 24-bit audio. I have bus-analyzer traces that prove it So does Mac OS X's USB audio-class driver. You'll need to inform the OS that your device supports 24-bit audio, and you do that using the device descriptors. It's quite simple. Excellent ! I presume it'll be the task of the USB device's firmware to tell the OS. Yes. In the device firmware, you'll have a structure that holds the device's configuration and device descriptors. The firmware responds to the USB requests that ask for those descriptors as part of the enumeration process. c) If your design requires separate sampling frequencies for record and playback, then you'll need to use the TAS1020B, as its CODEC interface in I2S mode has two sets of LRCLK, BCLK and MCLK signals. The TUSB3200A's CODEC interface has only one set of BCLK, LRCLK and MCLK signals. A further complication is that if you wish to support more than one sampling frequency (even though both playback and record will always use the same frequency), then you need the TAS1020B and you can't use a CODEC (you need to use separate ADC and DAC devices). This is because the OS considers the playback and record interfaces to be completely separate, Does it indeed ? Does this mean that the sample rates for p/b and rec don't have to be synched then ? I was wondering about generating wordclock as it happens. If using the TAS1020B, then playback and record do not have to be synchronized. Both chips support using a clock generator PLL to synthesize the clocks (the TAS1020B has two synthesizers). This eliminates the need for an external oscillator, but jitter and such may not be good enough for your application. Both devices support using an external oscillator, but you'll have to carefully think through how that will work. This limitation is why TUSB3200A-based devices like the Edirol UA-5 have a front-panel switch that selects the sample frequency: at enumeration, the micro reads the state of the switch and alters the descriptors appropriately, and the only way to change the sample frequency is to detach the device, set the switch, and reattach. Again, since the TAS1020B has two sets of I2S clock lines, it has no problem supporting different sample frequencies. Is this helpful? It certainly is. I'd already downloaded the data sheets for both parts, so I'll go look at them in more detail. You should also download the firmware development kits, as the data sheets barely skim the surface of how these chips work. -a |
#62
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
Andy Peters wrote: You're confusing "USB 2.0" with "high speed USB." While USB 1.1 had two speeds (low, at 1.2 Mbps, and full, at 12 Mbps), USB 2.0 adds a third speed (high, at 480 Mbps), but a USB 2.0 _device_ does not have to support high-speed transfers. However, note that a USB 2.0 _host_ must support high speed. I'm not confusing anything, I'm just not familiar with the terminology. I only know USB versions 1 and 2, and 2 is capable of much faster transfer rates. Since it's backward compatible, might as well use it. I know that disk drives connected to a USB2 port have a much higher throughput than when connected to a USB1 port. That's good enough for me (as a user). Designers may, and probably do, see it differently, as do the bean counters. What Edirol is saying is that if you connect the UA-101 to a USB 2.0 host (which, as noted above, must support high-speed transfers), then the device will do multiple channels at 96 kHz, as the device will enumerate as a high-speed device. However, if you plug it into an older USB 1.1 host, which doesn't support high-speed transfers, then the device enumerates using a full-speed configuration which only supports 48 kHz stereo. This is similar to my laptop situation. It has only USB1.1 ports. When I plug a disk drive in a USB2 case into the computer, it runs slow enough so that I can only record about 6 tracks at 44.1 kHz without it getting behind. If I plug the same drive into another computer with a USB2 port, it works just like an internal disk drive. Bottom line - sufficient speed for 96 kHz duplex stereo is available through USB. You just need to implement it properly, which is what Graham is trying to find out how to do. |
#63
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
Mike Rivers wrote:
Andy Peters wrote: You're confusing "USB 2.0" with "high speed USB." While USB 1.1 had two speeds (low, at 1.2 Mbps, and full, at 12 Mbps), USB 2.0 adds a third speed (high, at 480 Mbps), but a USB 2.0 _device_ does not have to support high-speed transfers. However, note that a USB 2.0 _host_ must support high speed. I'm not confusing anything, I'm just not familiar with the terminology. I only know USB versions 1 and 2, and 2 is capable of much faster transfer rates. Since it's backward compatible, might as well use it. I know that disk drives connected to a USB2 port have a much higher throughput than when connected to a USB1 port. That's good enough for me (as a user). Designers may, and probably do, see it differently, as do the bean counters. Ahhh, OK. You are correct -- the terminology _is_ confusing, and some manufacturers take advantage of the confusion by offering "USB 2.0" products, insinuating that the devices are high speed when they are not. What Edirol is saying is that if you connect the UA-101 to a USB 2.0 host (which, as noted above, must support high-speed transfers), then the device will do multiple channels at 96 kHz, as the device will enumerate as a high-speed device. However, if you plug it into an older USB 1.1 host, which doesn't support high-speed transfers, then the device enumerates using a full-speed configuration which only supports 48 kHz stereo. This is similar to my laptop situation. It has only USB1.1 ports. When I plug a disk drive in a USB2 case into the computer, it runs slow enough so that I can only record about 6 tracks at 44.1 kHz without it getting behind. If I plug the same drive into another computer with a USB2 port, it works just like an internal disk drive. Good -- that is exactly how it's intended to work. Bottom line - sufficient speed for 96 kHz duplex stereo is available through USB. You just need to implement it properly, which is what Graham is trying to find out how to do. And the rub here is that the TAS1020B chip is USB 2.0 compliant, yet is only capable of full-speed operation. Therefore, if he needs full-duplex stereo 96 kHz operation, the TI chips are not the solution. -a |
#64
Posted to sci.electronics.design,rec.audio.pro,rec.audio.tech
|
|||
|
|||
24 bit audio over USB
Andy Peters wrote: And the rub here is that the TAS1020B chip is USB 2.0 compliant, yet is only capable of full-speed operation. Therefore, if he needs full-duplex stereo 96 kHz operation, the TI chips are not the solution. Am I still confused? This sounds like there will be plenty of speed for full duplex (four channels) at 96 kHz - more than necessary. If he wants backward compatibility to work at low speed and transfer four channels at 48 kHz to a device that's only set up for USB1.1, that sounds like it won't work with this chip. |
Reply |
|
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
It's amazing what you can find when you look. | Audio Opinions | |||
Artists cut out the record biz | Pro Audio |