Boathouse Row Philadelphia, PA

Main menu:

April 2019
« Aug    


Site search


Now Reading

Beware – Nvidia nForce730a USB audio bug

Summary – You can read the details below but DO NOT BUY NVIDIA-based motherboards!! Rumors abound that Nvidia is leaving the chipset business!! I can guarantee that THEY DO NOT SUPPORT KNOWN BUGS IN THEIR CHIPSETS. The nForce 730a MCP (GeForce 8200) does not support USB audio under current Linux kernels.

My adventures with the bug in the Nvidia 730a chipset began when I decided to use a Tigertronics Signalink USB interface for my PSK31 ham radio operations. I wanted to use the Signalink with its integrated sound card to free up my on-board sound card for normal audio playback i.e web pages, videos etc. Thus, I wouldn’t be crawling under my desk every time I wanted to disconnect my transceiver to listen to a video etc.
A little about my system before the details on the bug.
Computer system:
    EVGA 730a motherboard with an AMD Phenom quad core processor  spec sheet    manual  
    Ubuntu 10.4  64 bit
    PSK31 application is fldigi
Installation of the Signalink couldn’t have been easier. Except there was no audio on the USB interface. Using ‘lsusb” showed that Linux detected the USB codec just fine.
After quite a bit of Googling I found that the clue was provided by “dmesg” in the form of a kernel error message. Whenever the USB audio interface was attached and a sound application tried to use the interface (either fldigi or simply the gnome volume control) the kernel would report “usb_set_interface failed”. This failure would be reported over and over again as the kernel tried to set the interface.

Once the interface fails, the gnome volume control “hangs” and the message “awaiting sound system response” will appear.

Some checks of the linux kernel docs shows that the usbaudio module uses the usb_set_interface call located in the usbcore module. The call is described as follows “usb_set_interface — Makes a particular
alternate setting be current”.
At this point we know that the kernel isn’t able to set the interface but we don’t know if it’s a software problem or hardware problem (motherboard or Signalink)?
First step is to eliminate the software on my system I ran the Ubuntu 10.4 live cd.   Problem was unchanged!
Second, I ran Fedora 12 live cd (could be Ubuntu implementation). Problem was unchanged.
However, it still could be the Linux kernel or modules since Ubuntu and Fedora are using the same kernel version.
Fortunately, I also have an original EeePC 701 netbook. On this netbook I run EasyPeasy 1.6, a Ubuntu derivative tailored for netbook operation. The linux kernel and fldigi application is the same as Ubuntu 10.4 and Fedora. Hooking up the Signalink and transceiver with fldigi and everything worked perfectly!! Thus, software is an unlikely culprit.
At about this time I decided to post on the Linux forum of the Nvidia site to see if anyone else had the problem. Perhaps if Nvidia monitors the site there might be a solution. Here is the thread. The only response came from Zbigniew Luszpinski in Pultusk, Poland. Zbiggy knew a great deal about the bug and had been in direct contact with Nvidia back in March 2010. Nvidia’s response can be summed up by Zbiggy’s comments

nForce 7xx and Geforce 8xxx mainboards have broken USB OHCI controller which is used for connecting USB 1.1 devices. It just hangs. Nvidia know about this bug but did not provide fix for it. I doubt if they work on this. The last contact I had with Daniel Dadap from Nvidia was at 16/03/2010 when he was collecting hardware to reproduce the bug. He said tomorrow (the day after 16/03/2010) will send me the findings. Today is 16/05/2010 and have no response. Nvidia quit chipset business recently so we are probably left without support and fixes.

I was always buying and recommending nForce chipsets for all these years to everybody since nForce2 up to and including Geforce 8200 chipset from nForce7xx family. Now I’m sad I was treated this way. This is not a problem when company releases buggy chipset when it also releases a fix. But leaving chipset business with growing number of customers who encounter this not fixed bug is ugly. I’m enough. Next mainboard will be from AMD/ATI.

The final straw came for me with the addition of an Ultra 6-port USB 2.0 PCI card. This add-in USB card uses a VIA VT82xxxx chipset rather than Nvidia. The card worked with Signalink right out of the box! Proving conclusively that there is a bug in the Nvidia chipset!

Powered by ScribeFire.

Write a comment