Very Computer

Hi,

 I am trying to design an embedded audio player and I cant seem to
find any audio codecs anymore with 8b / 16b parallel interface ( 16b
resolution desirable). I have located some data on Cirrus CS4235 but I
dont know if its still active. Also it does a whole lot more than I
want ( I just want basic codec, with some filter is a plus). Can
anybody advise on CS4235 and/or any suitable low-cost alternatives?

Thanks,
Jai Shanker W
Shreyanet Pvt Ltd.
Bangalore INDIA
jswarrier at shreyanet dot com

You'd be better off, I think, going to a serial DAC with I2S format.
Then you get a ton of choices and you can put TI's TAS3001 inline
with the data for equalization or dynamic range compression.

If you absolutely have to have a parallel bus that makes the audio device
look like a memory device you'd be well served to add a FIFO and use
a CPLD to read from the FIFO in parallel and shift the data out to the
DAC or TAS3001 in I2S format.

--Keith Brafford

Quote:>  I am trying to design an embedded audio player and I cant seem to
> find any audio codecs anymore with 8b / 16b parallel interface ( 16b
> resolution desirable). I have located some data on Cirrus CS4235 but I
> dont know if its still active. Also it does a whole lot more than I
> want ( I just want basic codec, with some filter is a plus). Can
> anybody advise on CS4235 and/or any suitable low-cost alternatives?

So far as I know the CS4235 has been end-of-lifed: one of the projects that
I'm working on at the moment uses them and the last buy was a few months ago
as far as I remember.

There are few ISA based codecs left as manufacturers have moved to PCI
because the market is driven by desktop PCs.

    Andrew

Thanks to all. I am going with a CPLD solution and serial codec. I saw

the cost of serial codec + interface logic, plus saving the design
effort. But I guess its better to avoid EOL parts in new design.

I'm trying to interface the codec to 8051/251 family of CPUs.
Simulation shows that I'm going to have trouble even with the faster
range of those uCs available today ( actually the problem seems more
to be with Keil /Archimedes compilers producing LOTS of ASM code from
C source). I'd really appreciate it if anyone who has worked on this

or maybe 251s at 16Mhz. Has somebody done it before.

Thanks,
Jai

I finished a project very similar to what you are doing just a few months
ago.

My board used an embedded 80186 16-bit processor at 40 MHz with
an IDT 1K FIFO in the peripheral address space.  The output of the
FIFO was controlled by a Cypress 37032 CPLD that read the data
and serially shifted it into the audio chip according to the audio specs.

I only needed to maintain an average 320kbps throughput and also have
enough cycles left over to run a full TCP/IP stack and web server with
reasonably complicated device home page.  It works fine.

I'd guess that with a memory-mapped FIFO feeding a CPLD you should
have no problem getting whatever data rate you want.

--Keith Brafford

Yes it is dooable - I've made various digital delay systems for
loudspeaker alignment using the CS4231 (16 bit codec, 8 bit parallel
interface) with the configuration and DMAing being done by a regular
89C52. Yes it is *y tight for 48ksps stereo 16 bit and it is
necessary to overrev the '52 a bit (40MHz) lots of inline code (few
loops) mind you it would have been less hectic if I hadn't been using
DRAM as a storage medium (lotsa CASing and RASing).
M