Al Clark wrote:
> Steve Underwood <steveu@dis.org> wrote in news:ft1hp9$67d$1
> @home.itg.ti.com:
> 
>> Glenn Zelniker wrote:
>>> On Apr 1, 8:06 pm, Steve Underwood <ste...@dis.org> wrote:
>>>
>>>> Good and on-chip are not entirely compatible in the codec world.
>>>> Mediocre and on chip is certainly available. However, a 100 DSP MIPS
>>>> processor is a bit noisy to live in harmony with a state of the art
>>>> audio codec.
>>> By "good," I meant good enough for portable consumer audio
>>> applications. And there are audio SOCs extant with good CODEC specs.
>>> Look at the TAS3204 from Texas Instruments. There's a fast DSP in the
>>> same package as the converters, both of which do better than 100 dB
>>> dynamic range and better than -90 dB THD+N. I could live with this
>>> chip if it had more RAM.
>> So mediocre is OK for you, and you can probably achieve what you need on 
>> a single chip. :-)
>>
>>>> You just said codec. What are you looking for? ADC, DAC, or both?
>>> IIRC, "CODEC" is short-hand for coder/decoder. In other words, both!
>> To most people these days "CODEC" means a compression/decompression 
>> module with no conversion capabilities at all. The vagueness of the term 
>> means my question was very relevant.
>>
>> Regards,
>> Steve
>>
> 
> I certainly agree that the term has become ambiguous. A data converter that 
> contains both an ADC and a DAC should have been called something else.
> 
> Nevertheless, the ADC-DAC usage is quite common, especially with audio 
> converters. I'm not convinced that most people think 
> compression/decompression when they see this term. CODEC is a contraction 
> of coder/decoder but is more often ADC/DAC today.
> 
> In any case, I think the poster was interested in ADC & DAC specs .

Actually codec always meant compression and decompression. The original 
devices termed codecs where the digitisers for the telephone network. It 
wasn't really the ADC and DAC that got them the name codec. It was the 
inclusion of lossy compression to/from ulaw or Alaw in the conversion 
process which got them the name. I think use of the term codec for an 
ADC + DAC is just a sloppy derivation from that.

Say codec to most people these days and they think MP3, MPEG2, and so 
on. Again the term is used pretty sloppily. Most MP3 units are a co or a 
dec. Only a limited number of applications use both parts together.

Steve

Steve Underwood <steveu@dis.org> wrote in news:ft1hp9$67d$1
@home.itg.ti.com:

> Glenn Zelniker wrote:
>> On Apr 1, 8:06 pm, Steve Underwood <ste...@dis.org> wrote:
>> 
>>> Good and on-chip are not entirely compatible in the codec world.
>>> Mediocre and on chip is certainly available. However, a 100 DSP MIPS
>>> processor is a bit noisy to live in harmony with a state of the art
>>> audio codec.
>> 
>> By "good," I meant good enough for portable consumer audio
>> applications. And there are audio SOCs extant with good CODEC specs.
>> Look at the TAS3204 from Texas Instruments. There's a fast DSP in the
>> same package as the converters, both of which do better than 100 dB
>> dynamic range and better than -90 dB THD+N. I could live with this
>> chip if it had more RAM.
> 
> So mediocre is OK for you, and you can probably achieve what you need on 
> a single chip. :-)
> 
>>> You just said codec. What are you looking for? ADC, DAC, or both?
>> 
>> IIRC, "CODEC" is short-hand for coder/decoder. In other words, both!
> 
> To most people these days "CODEC" means a compression/decompression 
> module with no conversion capabilities at all. The vagueness of the term 
> means my question was very relevant.
> 
> Regards,
> Steve
> 

I certainly agree that the term has become ambiguous. A data converter that 
contains both an ADC and a DAC should have been called something else.

Nevertheless, the ADC-DAC usage is quite common, especially with audio 
converters. I'm not convinced that most people think 
compression/decompression when they see this term. CODEC is a contraction 
of coder/decoder but is more often ADC/DAC today.

In any case, I think the poster was interested in ADC & DAC specs .

Al Clark
Danville Signal Processing, Inc.

>Hello, all. It's been many moons since I last posted here. I am doing
>a feasibility study for a client that entails locating a moderately
>powerful (~100 MMAC/s at better than 24x24 bits) DSP that consumes
>very little power. Furthermore, it would be even more lovely if the
>DSP lived on the same chip as an audio CODEC that did at least 16 bits
>at 44.1/48 kHz. I've scoured the web and I've also checked the usual
>suspects -- ST, Analog, TI, Atmel, etc. -- and so far I haven't found
>the perfect part (although the new ST331 from ST Micro looks like it
>might fit the bill).
>
>While doing my research, I discovered that there are many players in
>this market sector that I've not heard of before. BelSigna (owned by
>OnSemi, now AMI) comes to mind. And I've also heard word of an ARM-
>based SOC that might do what I want, although the manufacturer escapes
>me right now. My hope is that someone out there might have some
>recommendations. Who makes a system-on-chip for portable audio that
>includes a good CODEC and a reasonably powerful DSP with enough
>precision to do useful audio processing.
>
>Thanks in advance,
>
>Glenn Zelniker
>
>

Hi Glenn,

I'm not sure if you saw this one but ADI has a nice very low-power DSP
with a stereo 24-bit audio CODEC embedded - ADSP-BF522C/BF524C/BF526C. 
The Blackfin is a dual-MAC 16-bit core but does a decent job at 32-bit
precision and should get you to at least 100MMACs depending on which speed
grade you use.  There's a free download called VisualAudio (v2.5) which is
a graphical compiler and comes with a set of 32-bit audio processing
primitives provided in source code.  You'll find hand assembled 32-bit FIR
& IIR implementations amongst other things.  You'll still need the
commercial VDSP tools to build an application.

Good luck on your search!
/Dan

Glenn Zelniker wrote:
> On Apr 1, 8:06 pm, Steve Underwood <ste...@dis.org> wrote:
> 
>> Good and on-chip are not entirely compatible in the codec world.
>> Mediocre and on chip is certainly available. However, a 100 DSP MIPS
>> processor is a bit noisy to live in harmony with a state of the art
>> audio codec.
> 
> By "good," I meant good enough for portable consumer audio
> applications. And there are audio SOCs extant with good CODEC specs.
> Look at the TAS3204 from Texas Instruments. There's a fast DSP in the
> same package as the converters, both of which do better than 100 dB
> dynamic range and better than -90 dB THD+N. I could live with this
> chip if it had more RAM.

So mediocre is OK for you, and you can probably achieve what you need on 
a single chip. :-)

>> You just said codec. What are you looking for? ADC, DAC, or both?
> 
> IIRC, "CODEC" is short-hand for coder/decoder. In other words, both!

To most people these days "CODEC" means a compression/decompression 
module with no conversion capabilities at all. The vagueness of the term 
means my question was very relevant.

Regards,
Steve

Thank you.


Al Clark wrote:

> Richard Owlett <rowlett@atlascomm.net> wrote in 
> news:bs2dna01vvuVbW7anZ2dnUVZ_tajnZ2d@supernews.com:
> 
> 
>>Glenn Zelniker wrote:
>>
>>>On Apr 1, 8:06 pm, Steve Underwood <ste...@dis.org> wrote:
>>>
>>
>>[snip]
>>
>>>>You just said codec. What are you looking for? ADC, DAC, or both?
>>>
>>>
>>>IIRC, "CODEC" is short-hand for coder/decoder. In other words, both!
>>
>>I had the idea that a CODEC might take one digital representation of a 
>>signal and convert it to another. That is "coding" might be compressing 
>>a linear representation of a signal (i.e. output of a ADC) and 
>>"decoding" might be decompressing a digital stream to become a linear 
>>representation of a new signal (i.e. to be input to a DAC).
>>
>>If out in "left field", am I still NEAR ballpark?
>>
>>[In spite of a certain MIT grad insisting that I was not a newbie ;]
>>
> 
> 
> CODEC tends to have two meanings. You described one of these.
> 
> The context that the earlier post referred to is basically a device that 
> includes both an ADC and a DAC. In most cases, there is no compression, etc.
> 
> I think the usage started with mu-law and A-law telecom converters.
> 
> Al Clark
> Danville Signal Processing, Inc.

Glenn Zelniker <z-sys@z-sys.com> wrote in news:02506d65-2183-4259-ad5f-
0eff8e3d9246@d45g2000hsc.googlegroups.com:

> On Apr 1, 8:06&#4294967295;pm, Steve Underwood <ste...@dis.org> wrote:
> 
>> Good and on-chip are not entirely compatible in the codec world.
>> Mediocre and on chip is certainly available. However, a 100 DSP MIPS
>> processor is a bit noisy to live in harmony with a state of the art
>> audio codec.
> 
> By "good," I meant good enough for portable consumer audio
> applications. And there are audio SOCs extant with good CODEC specs.
> Look at the TAS3204 from Texas Instruments. There's a fast DSP in the
> same package as the converters, both of which do better than 100 dB
> dynamic range and better than -90 dB THD+N. I could live with this
> chip if it had more RAM.
> 
>> You just said codec. What are you looking for? ADC, DAC, or both?
> 
> IIRC, "CODEC" is short-hand for coder/decoder. In other words, both!

ADI has announced the C versions of the Blackfin BF52x. The C versions are 
very small and have an internal codec. Blackfins have good low power 
performance.

I'm not sure how well this fits, but you might also look at ADI's sigmaDSP 
family.

Al Clark
Danville Signal Processing, Inc.

Richard Owlett <rowlett@atlascomm.net> wrote in 
news:bs2dna01vvuVbW7anZ2dnUVZ_tajnZ2d@supernews.com:

> Glenn Zelniker wrote:
>> On Apr 1, 8:06 pm, Steve Underwood <ste...@dis.org> wrote:
>> 
> [snip]
>> 
>>>You just said codec. What are you looking for? ADC, DAC, or both?
>> 
>> 
>> IIRC, "CODEC" is short-hand for coder/decoder. In other words, both!
> 
> I had the idea that a CODEC might take one digital representation of a 
> signal and convert it to another. That is "coding" might be compressing 
> a linear representation of a signal (i.e. output of a ADC) and 
> "decoding" might be decompressing a digital stream to become a linear 
> representation of a new signal (i.e. to be input to a DAC).
> 
> If out in "left field", am I still NEAR ballpark?
> 
> [In spite of a certain MIT grad insisting that I was not a newbie ;]
> 

CODEC tends to have two meanings. You described one of these.

The context that the earlier post referred to is basically a device that 
includes both an ADC and a DAC. In most cases, there is no compression, etc.

I think the usage started with mu-law and A-law telecom converters.

Al Clark
Danville Signal Processing, Inc.

Glenn Zelniker wrote:
> On Apr 1, 8:06 pm, Steve Underwood <ste...@dis.org> wrote:
> 
[snip]
> 
>>You just said codec. What are you looking for? ADC, DAC, or both?
> 
> 
> IIRC, "CODEC" is short-hand for coder/decoder. In other words, both!

I had the idea that a CODEC might take one digital representation of a 
signal and convert it to another. That is "coding" might be compressing 
a linear representation of a signal (i.e. output of a ADC) and 
"decoding" might be decompressing a digital stream to become a linear 
representation of a new signal (i.e. to be input to a DAC).

If out in "left field", am I still NEAR ballpark?

[In spite of a certain MIT grad insisting that I was not a newbie ;]

On Apr 1, 8:06&#4294967295;pm, Steve Underwood <ste...@dis.org> wrote:

> Good and on-chip are not entirely compatible in the codec world.
> Mediocre and on chip is certainly available. However, a 100 DSP MIPS
> processor is a bit noisy to live in harmony with a state of the art
> audio codec.

By "good," I meant good enough for portable consumer audio
applications. And there are audio SOCs extant with good CODEC specs.
Look at the TAS3204 from Texas Instruments. There's a fast DSP in the
same package as the converters, both of which do better than 100 dB
dynamic range and better than -90 dB THD+N. I could live with this
chip if it had more RAM.

> You just said codec. What are you looking for? ADC, DAC, or both?

IIRC, "CODEC" is short-hand for coder/decoder. In other words, both!

Glenn Zelniker wrote:

> Hello, all. It's been many moons since I last posted here. I am doing
> a feasibility study for a client that entails locating a moderately
> powerful (~100 MMAC/s at better than 24x24 bits) DSP that consumes
> very little power. Furthermore, it would be even more lovely if the
> DSP lived on the same chip as an audio CODEC that did at least 16 bits
> at 44.1/48 kHz.

What exactly are you planning to do with it?

I doubt if there is a general purpose programmable DSP like that however 
there are many dedicated DAPs with the integrated codecs and 
mp3/dolby/wma hardware made by Cirrus, AKM, AD, NXP, ST, TI.

It is all very commercial so that they will not sell you the parts and 
will not even give you the datasheets unless you have all of the 
necessary licenses and you are committed for the industrial quantities.

> I've scoured the web and I've also checked the usual
> suspects -- ST, Analog, TI, Atmel, etc. -- and so far I haven't found
> the perfect part (although the new ST331 from ST Micro looks like it
> might fit the bill).

> While doing my research, I discovered that there are many players in
> this market sector that I've not heard of before. BelSigna (owned by
> OnSemi, now AMI) comes to mind. And I've also heard word of an ARM-
> based SOC that might do what I want, although the manufacturer escapes
> me right now. My hope is that someone out there might have some
> recommendations. Who makes a system-on-chip for portable audio that
> includes a good CODEC and a reasonably powerful DSP with enough
> precision to do useful audio processing.

How many words: "good", "portable", "reasonable", "powerful", "enough", 
"useful"... Are you in the marketing?


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com