--- Begin Message ---
|
Subject: |
Re: [Openvortex-dev] au8830 - AC3 pass through SPDIF |
|
Date: |
Sat, 05 Mar 2005 13:24:36 -0300 |
Hi,
> >>
> >> The EN_CODEC1 flag, is there to enable the AC97 channels 4 and 5. AFAIK,
> >> the bits in that register (VORTEX_CODEC_EN) map as follows:
> >>
> >> 8,9: AC97 channels 0 and 1
> >> 10,11: AC97 channels 2 and 3 (used as AC98 modem channels)
> >> 12,13: AC97 channels 4 and 5 (rear/headphones channels)
> >>
> >> For more details:
> >> http://galadriel.mat.utfsm.cl/~mjander/aureal/techdoc/techpaper.html
>
> This mean that EN_CODEC1 should only be enabled if au88x0 have a quad codec
Leaving it enabled should not harm i guess.
> >>
> >> > http://lists.gnu.org/archive/html/openvortex-dev/2003-11/msg00008.html
> >> >
> >> > 3) Do we need to change the ADB routing when using AC3 passthrough ?
> >>
> >> Hmm, good question. Maybe yes, because the current SPDIF PCM device
> >> route does stereo deinterlacing (separating two stereo channels). The
> >> SPDIF interface has 2 input sinks. For AC3 data, 2 interlaced channels
> >> does not make ant sense, because it could be 4, 6 or more, encoded in
> >> who knows what. Its just some raw data that must be passed through
> >> intact. It may be that in AC3 passthrough mode the SPDIF transmitter
> >> just re-joins the 2 data streams resulting in the original raw bit
> >> stream, hopefuly not messing it up. Maybe someone with more SPDIF
> >> knowledge can comment on this ?
> >>
>
>
> http://galadriel.mat.utfsm.cl/~mjander/aureal/techdoc/routes.png
>
>
> 3) Do we need to remove the MIXOUT to SPDIFOUT routes or just mute
> the corresponding MIXIN to MIXOUT (0x14) when using AC3 pass through SPDIF ?
One route target can't be targetted by two routes at the same time. That
would cause an error, most probably blocking other routes. The SPDIF
output targets must be connected to one single source each. AFAIK the
current code disconnects the SPDIF from MIXOUTS when using the "spdif"
PCM device, to prevent any conflict.
> >> > 6) What else will be needed for the au88x0 driver to support
> >> >wavetable synthesis ?
> >>
> >> The main problem is, that the audio routes from the WT engine are broken
> >> and thus the DMA engine does not seem transfer any data.
>
>
> # more /proc/asound/card0/pcm3p/info
> card: 0
> device: 3
> subdevice: 0
> stream: PLAYBACK
> id: AU88x0 WT
> name: wt
> subname: subdevice #0
> class: 0
> subclass: 0
> subdevices_count: 64
> subdevices_avail: 64
>
>
> # aplay -D hw:0,3 test.wav
> Playing WAVE 'test.wav' : Unsigned 8 bit, Rate 11025 Hz, Mono
> aplay: set_params:857: Sample format non available
>
>
> 4) How can we use "aplay" to play test.wav on the card 0 device 3 ?
The problem with the WT device is either a bug in the DMA code or a bad
route. The WT DMA works almost indentical to the ADB DMA engine, but
with its separate set of registers. I have looked over and over
uncountable times, without finding any clue. The most uncertainty is in
regard of the WT routes. The only thing i know is that each WT bank has
32 channels, each channel with some processing capabilities (pitch,
delay, filter, gain, etc) and there are 32 + 6 sources toward the
routing bus for each bank. I guess that the first 32 sources are mapped
to each individual WT dma channel and the other extra 6 sources are some
kind of downmix of the entire bank. But i am still clueless about why
having 32 individual sources for each channel, if we have only 32 mixer
inputs, and many of them are already being used ? There is no way to
make use of all of them, thats why i guess that there must be some
downmix of each WT block available somehow.
> >> > >
> >> > > > 5) What additional kcontrol and routine are needed for SPDIF IN ?
> >> > >
> >>
> >> AFAIK, there is no direct SPDIF input interface on the au8830 chip. It
> >> looks like that the SPDIF input is implemented using a I2S port or one
> >> of the coporecessor interfaces. Since they are all sincronous serial
> >> interfaces, they probably are almost the same and almost compatible
> >> between each other, requiring little tweaking/additional circuitry on
> >> the interface daughter-board.
>
> http://sourceforge.net/projects/aureal
>
> Most likely SPORT interface.
I guess you are right.
Best Regards
--
Manuel Jander
Electronic Engineer
--- End Message ---