[linux-audio-dev] Re: GPL Audio Hardware

James Courtier-Dutton James at superbug.co.uk
Thu Apr 6 12:32:41 UTC 2006


Lee Revell wrote:
> Lately however Creative/EMU's DSP cards have been getting great reviews
> (X-Fi, EMU 1212M, etc) - they seem to have fixed a lot of the common
> complaints about the SBLive! based devices (crappy codecs, horrible
> resampling).  And they are still damn cheap ($200-300 tops IIRC) but
> have no Linux drivers.  This might be an area to investigate, but I
> suspect you would need years of development to produce a device of that
> quality for anywhere near Creative's price point.
>   
Creative have actually donated an X-Fi and an EMU1212M to me.
No datasheets, but at least having the hardware is a start. (Creative 
pass me some datasheets for other cards though)

I have currently got as far as horrible noises coming from the EMU1212M. 
So, after some more reverse engineering, we should have that working.
The X-Fi is a totally new beast, so that is a long way off currently.

My point is that with reverse engineering, we can get these commercial 
cards working with Linux, so do we really need to build our own open 
source ones?
Sound cards are not nearly as complicated as graphics cards, so the 
reverse engineering task is less and therefore do-able.
Saying that, the X-Fi might be an exception.

If I was designing a GPL sound hardware, I would want to separate the 
DAC and ADC tasks from any other processing done by the card.
I.e. The card would do all the DSP, FPGA processing and only have 
digital external links. Then have the DAC and ADC as a separate external 
device.

The main task would be designing a good interface to the DAC and ADCs 
that would not require any firmware on the DAC and ADC device.

The EMU1212M has this separation. It has two parts:
a) The EMU1010 is the DSP/FPGA card.
b) The EMU0202 is the DAC/ADC device.

There is another area of GPL development work that might be of interest 
to the list and that is the Software Defined Radio open source work 
going on at the moment. A lot of that is based around using a sound card 
for the ADC, but sampling at 96khz/24bit.
One has an external device that does the down mixing of RF to an IF of 
about 32khz. The sound card then samples that IF at 96khz and then 
processes the IF.
The main requirement for SDR is dynamic range, so the 24bit is 
important. So the aim of getting a very good ADC for 96khz/24bit in the 
SDR market, is the same as that in the Audio market. So, you might be 
able to combine efforts.

James






More information about the Linux-audio-dev mailing list