[AG-TECH] Current soundcards in use with Alsa and rat?

[Eric Olson]

Thanks Bob, that seems to be the problem.  I know you had suggested that 
before, but since we hadn't changed soundcards, I didn't realize how it 
could be an issue.  It turns out that when we switched to alsa and its oss 
compatibility drivers, the name of the sound card device changed and no 
longer matched the name in hardcoded table.

I have two sound card chipset names that I'll ask be added to the 
soundcard table in the rat code ("Realtek ALC650E", and "Asahi Kasei 
AK4531").  If anyone else has other cards that need to be fixed this way, 
we should submit them, too.  If you're having the same line-in problems I 
was, you can test the fix by setting the environment variable OSS_IS_AC97 
to 1 before starting rat.  This has the same effect as adding your card's 
name to the hardcoded table.

Eric

On Thu, 23 Dec 2004, Robert Olson wrote:

> 
> This effect can be explained if you look at the architecture of AC97 
> sound chip.
> 
> Below (if you can see HTML mail, if not, the image is from the 
> datasheet at
> 
> http://www.analog.com/en/prod/0,2934,AD1819B,00.html
> 
> you will see the block diagram of a typical AC97 device. Note that all 
> inputs route through a selector, then to what this datasheet calls a 
> PGA, a programmable gain amplifier. The gain on that amplifier is what 
> controls the input capture level. This level adjustment is exposed in 
> the OSS audio drivers as the Igain or Reclevel control, I believe. I 
> don't recall what ALSA calls it. Adjusting the line-in volume level 
> controls the gain in the line-in -> output mixer -> line out signal 
> path, which is why that level must be set to zero to keep local 
> loopback from occuring.
> 
> The OSS code in rat attempts to detect if the soundcard in use is AC97 
> compatible (currently, by using a hardcoded list of device names). If 
> it determines that it is AC97, it maps the capture volume control to 
> control the igain signal and sets the line in level to zero, making rat 
> act correctly.
> 
> I haven't looked closely at the ALSA rat code, but I suspect that it 
> needs to have similar changes made.
> 
> --bob
> 
>