openvortex-dev
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Openvortex-dev] Recent Bugs.


From: Manuel Jander
Subject: [Openvortex-dev] Recent Bugs.
Date: Sat, 20 Mar 2004 04:10:43 -0400

Hi everyone,

- Regarding the Distorted sound "bug":

I increased the output gain of the PCM output Hardware mixer of the
Aureal Vortex driver in the latest patches, because if not, we loose
audio resolution. I noticed this while recording the mix output, because
the signal would be max one quarter of the full 16 bit scale, using only
14 effective bits.

The digital data coming from the DMA is 16 bit signed data. Through the
mixer it is increased two times by 6dB (8 byte, MIX), 12dB total.
That is a signal power gain of:
                                2^(12[dB]/3) = 16

Since sqrt(power) =~ voltaje, we have a total voltaje gain of sqrt(16)=4
(3dB equals gain factor 2). That means the 16 bit signal becomes 18
bits.
AC97 codecs ar 18 bit codecs, so we are fine with that.
If we feed just 16 bit data into the AC97 codec, we get a much worse
signal to noise ratio (S/N).

Test: I wrote a little program that outputs a 500Hz sine wave with 32767
amplitude, and i get no clipping, just a clean sine.

The problem for the AU8830 is that the EQ has some little gain, and the
audio samples get too big for the DAC's causing clipping. I modified my
tree to lower the gain of the mixers that go into the EQ (front).

Conclusion: Gain needs to be lowered to give a guard margin for the
extra gain from the EQ.


- Regarding the fast play bug:

        I still have no proof why it happens, but the original fix (disabling
the first DMA channel) does not seem to work in every case. Indeed i'm
not using that workaround anymore, and i get that problem very few
times.
I have observed that if i use 1 stereo channel and a mono channel at the
same times on my Diamond MX300, the fast playback problem disappears
until i reboot. I have reproduced this behavior 3 times so far, without
any exception.
My OOAOO MX300 Ultra (chinese card manufactured in 2003) has a newer
chip and works OK all the time.

I think Jeff Muizelaar is right about that revisions go backwards on
AU8830's.

Conclusion: I'm very sorry, but i still don't know why this damn bug
happens :(

Well, thats all for now.

And of course, another patch :D. I cannot tell if it will work better
for you or not. It applies against linux kernel 2.6.5-rc1
http://galadriel.mat.utfsm.cl/~mjander/aureal/alsa/aureal-2.6.5-rc1-b.diff.gz


Best Regards

Manuel






reply via email to

[Prev in Thread] Current Thread [Next in Thread]