Async USB Dac Experiments - testers wanted...

[dynaudiorules]

Hum - async is not working then - should be being updated.. Are you able to ssh into the touch and type "dmesg" - what does it say for the last few lines while it is playing?

r6:00000002 r5:c03f6c78 r4:c7e0fa00 r3:c025ce58
[<c0245498>] (snd_pcm_do_prepare+0x0/0x34) from [<c0244fc4>] (snd_pcm_action_single+0x40/0x6c)
r4:c7e0fa00 r3:c0245498
[<c0244f84>] (snd_pcm_action_single+0x0/0x6c) from [<c02463d4>] (snd_pcm_action_nonatomic+0x4c/0x64)
r7:00000002 r6:00000002 r5:c7e0fa00 r4:c03f6c78
[<c0246388>] (snd_pcm_action_nonatomic+0x0/0x64) from [<c0248968>] (snd_pcm_common_ioctl1+0x710/0x100c)
r6:c7c7aa1c r5:00000000 r4:c7e0fa00 r3:00000000
[<c0248258>] (snd_pcm_common_ioctl1+0x0/0x100c) from [<c0249ac8>] (snd_pcm_playback_ioctl1+0x414/0x438)
[<c02496b4>] (snd_pcm_playback_ioctl1+0x0/0x438) from [<c0249b24>] (snd_pcm_playback_ioctl+0x38/0x3c)
r8:c002ec64 r7:0000000c r6:00004140 r5:00023060 r4:c7136e20
[<c0249aec>] (snd_pcm_playback_ioctl+0x0/0x3c) from [<c00ae8c0>] (vfs_ioctl+0x34/0x94)
[<c00ae88c>] (vfs_ioctl+0x0/0x94) from [<c00aebe8>] (do_vfs_ioctl+0x2c8/0x2e4)
r6:c7153e30 r5:00023060 r4:c7136e20 r3:00002000
[<c00ae920>] (do_vfs_ioctl+0x0/0x2e4) from [<c00aec44>] (sys_ioctl+0x40/0x64)
r6:00004140 r5:00023060 r4:c7136e20
[<c00aec04>] (sys_ioctl+0x0/0x64) from [<c002eae0>] (ret_fast_syscall+0x0/0x2c)
r7:00000036 r6:00019970 r5:0002e320 r4:00024828
#

[Triode]

That doesn't look right - could you capture the entire dmesg and post here or mail to me?

[dynaudiorules]

I sent you a PM with more of the result from the dmesg command

[dynaudiorules]

That doesn't look right - could you capture the entire dmesg and post here or mail to me?

email sent

[paul.raulerson]

Installed and tried on a MF VLink, A Wavelength CoSecant, and a Peachtree DAC*IT.

No synch to the VLink or CoSecant.

The Peachtree played the music cleanly, but the music "dragged" if that makes any sense.

By the way, install was flawless and the uninstall was also flawless. Good work, please don't get discouraged!

-Paul

[Triode]

Installed and tried on a MF VLink, A Wavelength CoSecant, and a Peachtree DAC*IT.

No synch to the VLink or CoSecant.

The Peachtree played the music cleanly, but the music "dragged" if that makes any sense.

By the way, install was flawless and the uninstall was also flawless. Good work, please don't get discouraged!

-Paul

Hi Paul - can you confirm whether you saw the words "USB Test Kernel" on the startup screen? I want to make sure that you actually installed the test kernel as the other case looks like it was not running. I've added to the instructions to hopefully make this clearer, but it would be good to understand what you saw in more detail..

[Triode]

Hi - I've been doing some more tests on this and have another kernel for people to try (Test Kernel #2)

Please also update to USB Audio Output 0.2 as this includes something to improve 96k playback.

I should point out what the different kernels do and what is leading me to look at this. From tests with my dac it appears that there is some form of limitation with the USB hardware in Touch related to the "Transaction Translator". This is part of the usb implementation which supports usb 1.0/1.1 devices connected to a usb 2.0 host. The specific limitation is that it does not appear to support output and input iso packets (sitd) in the same 1ms timeslot on the usb bus. The two kernels work round this in different ways and so one may be appropriate to you:

Kernel #1 - supports up to 44/48k sampling rates by restricting output packets to every other 1ms frame, thus allowing async input packets on the interviening frames. This relies on the async implementation in the dac supporting data arriving every other frame it asked for. It works with my Audiolab dac (with usb firmware by Lakewest audio) and I've verified with the author of that firmware that it should work. However it is breaking the requested parameters and so does not looks to work with several other dacs. This is probably the reason for the other test cases above failing.

Kernel #2 - supports up to 96k sampling rate by not changing the usb protocol. (Its a very minor modification of the default kernel to avoid the "not enough bandwidth message" seen with some usb devices) However it is intended to be used with an external usb high speed hub. In this configuration, the usb 1.0 to 2.0 translation is done by the external hub rather than the silicon within Touch and this seems to bypass the problem hardware. It works for me with the hub in my monitor - it is worth people testing with any "high speed" hub. Note there can be problem with rebooting the Touch, hub and dac in the right order so the Touch sees the dac, but once working seems to work ok at all sampling rates upto 96k for me.

Thanks to Dominik of Lakewest for suggesting the hub approach.

Please test and post here if these work better for you....

[Triode]

As an update to this - I've confirmed that that a cheap bus powered hub works - here's the one I've just purchased for this case (less than 10 GBP):


I think this makes the external hub option a very viable way of using a 1.1 usb dac. Interested in others trying. Needs kernel #2 above.

Note the hub must be a "high speed" USB 2.0 hub. (So the very cheapest ones around which are 1.1 are no good, but there are lots of cheap high speed usb 2.0 ones around)

[FredNowak]

I installed Kernal #2 and thought I would see what would happen without a hub. It is so good I have not bothered to hook up a hub. I have not heard any clicks or pops and will post again if they start.

Many thanks, Triode. I must also thank John Swenson for getting us to this point.

Fred

[JohnSwenson]

Hi Triode,
I finally had a little time today to try your work out. I'm using a HRT Music Streamer II as well. With Kernel #2 but no hub, I still get some clicks and pops. With the hub in place they are gone, I haven't heard one at all. I tried it at all sample rates (44.1, 48, 88.2 and 96) and get the same results, no clicks or pops, except for one track. One 88.2 tracks played with lots of clicks and pops, but it only did it once, playing it again works great. No idea what happened there.

For others that want to try the hub approach, install the software per the first post, unplug the power from the Touch, plug the DAC into the hub, plug the hub into the Touch, then plug the power back into the Touch. Use the USB audio setting to select the DAC, it will reboot and the DAC should work. Do NOT attempt any of this by "hot plugging", plugging in the DAC or hub with the Touch powered up. It won't physically hurt anything, but it won't work properly if hot plugged.

I tried kernel #2 out with an old adaptive (2706 based) DAC and it did not work at all, the Touch just continually rebooted (whether plugged into the hub or direct). I haven't had time to do any debugging on this. I haven't had time to try this DAC out with kernel #1.

I still have to try this with TT3.0, without the hub and the regular kernel TT3.0 makes a big difference in the sound of the Streamer II I wonder if with the hub in place it will make as much difference or not.

I have to try this out with my udev system. As you have figured out plugging in a USB DAC changes the assignments which can mess things up badly. I have a udev setup which makes sure all the devices get assigned a constant card no matter whether a USB DAC is plugged in or not. (but it still doesn't work with hot plugging)

Thanks so much Triode!!! This is great. Wonderful job of finding the issue and fixing it, and of coming up with a nice easy to use way of installing it.

Now I have to buy another hub, for testing I grabbed the one off the big computer, if I don't put it back my wife is going to get really annoyed in the morning when the keyboard and mouse don't work!

John S.