[App_rpt] USB hubs & USB audio
Jim Dixon and I have had disapointing results with USB hubs during the testing of USB fobs with them. We don't know why they don't work properly and we don't have the time to figure it out right now. There may be hubs which work just fine, but we just don't know as the root cause of the problem has not been uncovered. If Steve Henke W9SH knows something I don't them maybe he should speak up on the list. Steve WA6ZFT
The problem is probably complex. USB 2.0 hubs come in two flavors single- transaction translators (single-TT) and multi-transaction translators (multi-TT). Most USB audio devices are "full speed" devices (12 mbits/sec), not "high speed" devices (480 mbits/sec) so the USB hub must translate the data rate between the host and the device. The single-TT devices only have a single translator which is shared between all ports. A multi-TT hub has a translator per port. So in theory a milti-TT hub is what you should want for optimal bus usage. An additional complication is that the Linux kernel must be multi-TT hub aware to properly schedule things. I'm not current with Linux development, but I've read discussions in the past about improving the kernel's USB scheduler in this area. Here are a couple of links to info which may be useful: http://www.commsdesign.com/design_corner/showArticle.jhtml?articleID=1650536... http://www.linux-usb.org/usb2.html 73's Skip WB6YMH
I have successfully tested URIs with several high speed USB hubs. But these tests were performed without any other USB devices connected at the same time so this may not be very usefull. Skip points out the difference between full speed (12Mbps) and high speed (480Mbs) devices. This terminology is confusing. Not all USB 2.0 compliant hubs support high speed, so be sure the package says HIGH SPEED. Very few hubs support multiple transaction translators, probably because the chip sets are more expensive and almost nobody understands the need. Theoretically, one TT is all that is needed provided you don't mix USB 1.x and 2.0 devices on the same hub. Keep in mind that USB supports several data modes each with varying priorities and latency allocations. Plus USB physical layer is half-duplex so one-way communication never comes close to 480Mbps. The TT tries to handle all the traffic so don't expect good results if you plug a bunch of web cams or other bandwidth hungry devices into your hub. Hope this help. David Kramer DMK Engineering Inc. 310-544-1222 -----Original Message----- From: app_rpt-bounces@lists.illiana.net [mailto:app_rpt-bounces@lists.illiana.net] On Behalf Of Skip WB6YMH Sent: Tuesday, July 22, 2008 1:22 PM To: app_rpt@lists.illiana.net Subject: [App_rpt] USB hubs & USB audio
Jim Dixon and I have had disapointing results with USB hubs during the testing of USB fobs with them. We don't know why they don't work properly and we don't have the time to figure it out right now. There may be hubs which work just fine, but we just don't know as the root cause of the problem has not been uncovered. If Steve Henke W9SH knows something I don't them maybe he should speak up on the list. Steve WA6ZFT
The problem is probably complex. USB 2.0 hubs come in two flavors single- transaction translators (single-TT) and multi-transaction translators (multi-TT). Most USB audio devices are "full speed" devices (12 mbits/sec), not "high speed" devices (480 mbits/sec) so the USB hub must translate the data rate between the host and the device. The single-TT devices only have a single translator which is shared between all ports. A multi-TT hub has a translator per port. So in theory a milti-TT hub is what you should want for optimal bus usage. An additional complication is that the Linux kernel must be multi-TT hub aware to properly schedule things. I'm not current with Linux development, but I've read discussions in the past about improving the kernel's USB scheduler in this area. Here are a couple of links to info which may be useful: http://www.commsdesign.com/design_corner/showArticle.jhtml?articleID=1650536 2 http://www.linux-usb.org/usb2.html 73's Skip WB6YMH _______________________________________________ App_rpt mailing list App_rpt@lists.illiana.net http://lists.illiana.net/mailman/listinfo/app_rpt
So, I guess the next question is: Does anyone have a favorite multi-tt hub? ....I have used this one in the past for other linux applications and it seemed okay...I'll try one with the URI's....I guess the biggest issue here is that the hub you buy today won't necessarily be the one on the shelves tomorrow....A bit frustrating! http://www.usbgear.com/USBG-217C.html -David On Tue, 22 Jul 2008, David Kramer wrote:
I have successfully tested URIs with several high speed USB hubs. But these tests were performed without any other USB devices connected at the same time so this may not be very usefull.
Skip points out the difference between full speed (12Mbps) and high speed (480Mbs) devices. This terminology is confusing. Not all USB 2.0 compliant hubs support high speed, so be sure the package says HIGH SPEED.
Very few hubs support multiple transaction translators, probably because the chip sets are more expensive and almost nobody understands the need. Theoretically, one TT is all that is needed provided you don't mix USB 1.x and 2.0 devices on the same hub. Keep in mind that USB supports several data modes each with varying priorities and latency allocations. Plus USB physical layer is half-duplex so one-way communication never comes close to 480Mbps. The TT tries to handle all the traffic so don't expect good results if you plug a bunch of web cams or other bandwidth hungry devices into your hub.
Hope this help.
David Kramer DMK Engineering Inc. 310-544-1222
-----Original Message----- From: app_rpt-bounces@lists.illiana.net [mailto:app_rpt-bounces@lists.illiana.net] On Behalf Of Skip WB6YMH Sent: Tuesday, July 22, 2008 1:22 PM To: app_rpt@lists.illiana.net Subject: [App_rpt] USB hubs & USB audio
Jim Dixon and I have had disapointing results with USB hubs during the testing of USB fobs with them. We don't know why they don't work properly and we don't have the time to figure it out right now. There may be hubs which work just fine, but we just don't know as the root cause of the problem has not been uncovered. If Steve Henke W9SH knows something I don't them maybe he should speak up on the list. Steve WA6ZFT
The problem is probably complex. USB 2.0 hubs come in two flavors single- transaction translators (single-TT) and multi-transaction translators (multi-TT). Most USB audio devices are "full speed" devices (12 mbits/sec), not "high speed" devices (480 mbits/sec) so the USB hub must translate the data rate between the host and the device. The single-TT devices only have a single translator which is shared between all ports. A multi-TT hub has a translator per port. So in theory a milti-TT hub is what you should want for optimal bus usage.
An additional complication is that the Linux kernel must be multi-TT hub aware to properly schedule things. I'm not current with Linux development, but I've read discussions in the past about improving the kernel's USB scheduler in this area.
Here are a couple of links to info which may be useful:
http://www.commsdesign.com/design_corner/showArticle.jhtml?articleID=1650536 2
http://www.linux-usb.org/usb2.html
73's Skip WB6YMH
_______________________________________________ App_rpt mailing list App_rpt@lists.illiana.net http://lists.illiana.net/mailman/listinfo/app_rpt
_______________________________________________ App_rpt mailing list App_rpt@lists.illiana.net http://lists.illiana.net/mailman/listinfo/app_rpt
Hi Everyone, Well, I just tried one of the usbgear hubs listed below. It ..almost.. works. The 2 URI devices I'm experimenting with are identified and functional EXCEPT that it sounds like the audio sampling rate is wrong--way too low with aliasing distortion (or something). ugh. Probably a driver confusion issue?? Anyway, the USB hub and hardware otherwise seems to be fine. Here is an output from the usbtree utility: [root@kb4fxc-rpt ~]# ./usbtree /: Bus 02.Port 1: Dev 1, Class=root_hub, Drv=ohci_hcd/8p, 12M /: Bus 01.Port 1: Dev 1, Class=root_hub, Drv=ehci_hcd/8p, 480M |_ Port 4: Dev 2, If 0, Prod=USB2.0 Hub, Class=hub, Drv=hub/4p, 480M |_ Port 1: Dev 3, If 0, Prod=C-Media USB Headphone Set, Class=audio, Drv=snd-usb-audio, 12M |_ Port 1: Dev 3, If 1, Prod=, Class=audio, Drv=snd-usb-audio, 12M |_ Port 1: Dev 3, If 2, Prod=, Class=audio, Drv=snd-usb-audio, 12M |_ Port 1: Dev 3, If 3, Prod=, Class=HID, Drv=usbfs, 12M |_ Port 2: Dev 4, If 0, Prod=C-Media USB Headphone Set, Class=audio, Drv=snd-usb-audio, 12M |_ Port 2: Dev 4, If 1, Prod=, Class=audio, Drv=snd-usb-audio, 12M |_ Port 2: Dev 4, If 2, Prod=, Class=audio, Drv=snd-usb-audio, 12M |_ Port 2: Dev 4, If 3, Prod=, Class=HID, Drv=usbhid, 12M |_ Port 4: Dev 5, If 0, Prod=USB2.0 Hub, Class=hub, Drv=hub/4p, 480M I'll investigate further since this is ultimately a config I may want to use as well. David KB4FXC On Tue, 22 Jul 2008, David McGough wrote:
So, I guess the next question is: Does anyone have a favorite multi-tt hub? ....I have used this one in the past for other linux applications and it seemed okay...I'll try one with the URI's....I guess the biggest issue here is that the hub you buy today won't necessarily be the one on the shelves tomorrow....A bit frustrating!
http://www.usbgear.com/USBG-217C.html
-David
On Tue, 22 Jul 2008, David Kramer wrote:
I have successfully tested URIs with several high speed USB hubs. But these tests were performed without any other USB devices connected at the same time so this may not be very usefull.
Skip points out the difference between full speed (12Mbps) and high speed (480Mbs) devices. This terminology is confusing. Not all USB 2.0 compliant hubs support high speed, so be sure the package says HIGH SPEED.
Very few hubs support multiple transaction translators, probably because the chip sets are more expensive and almost nobody understands the need. Theoretically, one TT is all that is needed provided you don't mix USB 1.x and 2.0 devices on the same hub. Keep in mind that USB supports several data modes each with varying priorities and latency allocations. Plus USB physical layer is half-duplex so one-way communication never comes close to 480Mbps. The TT tries to handle all the traffic so don't expect good results if you plug a bunch of web cams or other bandwidth hungry devices into your hub.
Hope this help.
David Kramer DMK Engineering Inc. 310-544-1222
-----Original Message----- From: app_rpt-bounces@lists.illiana.net [mailto:app_rpt-bounces@lists.illiana.net] On Behalf Of Skip WB6YMH Sent: Tuesday, July 22, 2008 1:22 PM To: app_rpt@lists.illiana.net Subject: [App_rpt] USB hubs & USB audio
Jim Dixon and I have had disapointing results with USB hubs during the testing of USB fobs with them. We don't know why they don't work properly and we don't have the time to figure it out right now. There may be hubs which work just fine, but we just don't know as the root cause of the problem has not been uncovered. If Steve Henke W9SH knows something I don't them maybe he should speak up on the list. Steve WA6ZFT
The problem is probably complex. USB 2.0 hubs come in two flavors single- transaction translators (single-TT) and multi-transaction translators (multi-TT). Most USB audio devices are "full speed" devices (12 mbits/sec), not "high speed" devices (480 mbits/sec) so the USB hub must translate the data rate between the host and the device. The single-TT devices only have a single translator which is shared between all ports. A multi-TT hub has a translator per port. So in theory a milti-TT hub is what you should want for optimal bus usage.
An additional complication is that the Linux kernel must be multi-TT hub aware to properly schedule things. I'm not current with Linux development, but I've read discussions in the past about improving the kernel's USB scheduler in this area.
Here are a couple of links to info which may be useful:
http://www.commsdesign.com/design_corner/showArticle.jhtml?articleID=1650536 2
http://www.linux-usb.org/usb2.html
73's Skip WB6YMH
_______________________________________________ App_rpt mailing list App_rpt@lists.illiana.net http://lists.illiana.net/mailman/listinfo/app_rpt
_______________________________________________ App_rpt mailing list App_rpt@lists.illiana.net http://lists.illiana.net/mailman/listinfo/app_rpt
_______________________________________________ App_rpt mailing list App_rpt@lists.illiana.net http://lists.illiana.net/mailman/listinfo/app_rpt
David, Just so you know that it can be done, I used a 7 port USB hub with TT's on each port and filled it with 7 radio channels running simultaneously for a commercial trunking radio system and it works just fine. The server was a 2.66 GHz Pentium 4 on an ASUS motherboard. 73, Steve Henke, W9SH On Fri, Jul 25, 2008 at 9:19 PM, David McGough <kb4fxc@inttek.net> wrote:
Hi Everyone,
Well, I just tried one of the usbgear hubs listed below. It ..almost.. works. The 2 URI devices I'm experimenting with are identified and functional EXCEPT that it sounds like the audio sampling rate is wrong--way too low with aliasing distortion (or something). ugh. Probably a driver confusion issue??
Anyway, the USB hub and hardware otherwise seems to be fine. Here is an output from the usbtree utility:
[root@kb4fxc-rpt ~]# ./usbtree /: Bus 02.Port 1: Dev 1, Class=root_hub, Drv=ohci_hcd/8p, 12M /: Bus 01.Port 1: Dev 1, Class=root_hub, Drv=ehci_hcd/8p, 480M |_ Port 4: Dev 2, If 0, Prod=USB2.0 Hub, Class=hub, Drv=hub/4p, 480M |_ Port 1: Dev 3, If 0, Prod=C-Media USB Headphone Set, Class=audio, Drv=snd-usb-audio, 12M |_ Port 1: Dev 3, If 1, Prod=, Class=audio, Drv=snd-usb-audio, 12M |_ Port 1: Dev 3, If 2, Prod=, Class=audio, Drv=snd-usb-audio, 12M |_ Port 1: Dev 3, If 3, Prod=, Class=HID, Drv=usbfs, 12M |_ Port 2: Dev 4, If 0, Prod=C-Media USB Headphone Set, Class=audio, Drv=snd-usb-audio, 12M |_ Port 2: Dev 4, If 1, Prod=, Class=audio, Drv=snd-usb-audio, 12M |_ Port 2: Dev 4, If 2, Prod=, Class=audio, Drv=snd-usb-audio, 12M |_ Port 2: Dev 4, If 3, Prod=, Class=HID, Drv=usbhid, 12M |_ Port 4: Dev 5, If 0, Prod=USB2.0 Hub, Class=hub, Drv=hub/4p, 480M
I'll investigate further since this is ultimately a config I may want to use as well.
David KB4FXC
On Tue, 22 Jul 2008, David McGough wrote:
So, I guess the next question is: Does anyone have a favorite multi-tt hub? ....I have used this one in the past for other linux applications
it seemed okay...I'll try one with the URI's....I guess the biggest issue here is that the hub you buy today won't necessarily be the one on the shelves tomorrow....A bit frustrating!
http://www.usbgear.com/USBG-217C.html
-David
On Tue, 22 Jul 2008, David Kramer wrote:
I have successfully tested URIs with several high speed USB hubs. But
tests were performed without any other USB devices connected at the same time so this may not be very usefull.
Skip points out the difference between full speed (12Mbps) and high speed (480Mbs) devices. This terminology is confusing. Not all USB 2.0 compliant hubs support high speed, so be sure the package says HIGH SPEED.
Very few hubs support multiple transaction translators, probably because the chip sets are more expensive and almost nobody understands the need. Theoretically, one TT is all that is needed provided you don't mix USB 1.x and 2.0 devices on the same hub. Keep in mind that USB supports several data modes each with varying priorities and latency allocations. Plus USB physical layer is half-duplex so one-way communication never comes close to 480Mbps. The TT tries to handle all the traffic so don't expect good results if you plug a bunch of web cams or other bandwidth hungry devices into your hub.
Hope this help.
David Kramer DMK Engineering Inc. 310-544-1222
-----Original Message----- From: app_rpt-bounces@lists.illiana.net [mailto:app_rpt-bounces@lists.illiana.net] On Behalf Of Skip WB6YMH Sent: Tuesday, July 22, 2008 1:22 PM To: app_rpt@lists.illiana.net Subject: [App_rpt] USB hubs & USB audio
Jim Dixon and I have had disapointing results with USB hubs during
testing of USB fobs with them. We don't know why they don't work properly and we don't have the time to figure it out right now. There may be hubs which work just fine, but we just don't know as the root cause of the problem has not been uncovered. If Steve Henke W9SH knows something I don't them maybe he should speak up on the list. Steve WA6ZFT
The problem is probably complex. USB 2.0 hubs come in two flavors single- transaction translators (single-TT) and multi-transaction translators (multi-TT). Most USB audio devices are "full speed" devices (12 mbits/sec), not "high speed" devices (480 mbits/sec) so the USB hub must translate the data rate between the host and the device. The single-TT devices only have a single translator which is shared between all
and these the ports.
A multi-TT hub has a translator per port. So in theory a milti-TT hub is what you should want for optimal bus usage.
An additional complication is that the Linux kernel must be multi-TT hub aware to properly schedule things. I'm not current with Linux development, but I've read discussions in the past about improving the kernel's USB scheduler in this area.
Here are a couple of links to info which may be useful:
http://www.commsdesign.com/design_corner/showArticle.jhtml?articleID=1650536
2
http://www.linux-usb.org/usb2.html
73's Skip WB6YMH
_______________________________________________ App_rpt mailing list App_rpt@lists.illiana.net http://lists.illiana.net/mailman/listinfo/app_rpt
_______________________________________________ App_rpt mailing list App_rpt@lists.illiana.net http://lists.illiana.net/mailman/listinfo/app_rpt
_______________________________________________ App_rpt mailing list App_rpt@lists.illiana.net http://lists.illiana.net/mailman/listinfo/app_rpt
_______________________________________________ App_rpt mailing list App_rpt@lists.illiana.net http://lists.illiana.net/mailman/listinfo/app_rpt
participants (4)
-
David Kramer -
David McGough -
Skip WB6YMH -
Steven Henke