|
Hi C6x-ers, Does anybody have an idea if there is any input and output format for the AD535 codec on the C6711 DSK ? For e.g. whether the I/O to/from the codec is quantized to a uniform PCM format with at least 13 bits of dynamic range ? Thanks in anticipation, Raj. |
|
|
Input and Output formats of AD535 codec
Started by ●August 25, 2003
Reply by ●August 25, 20032003-08-25
|
--- Raj Sahakari <> wrote: > Does anybody have an idea if there is any input and > output format for the AD535 codec on the C6711 DSK ? I'm sure there is some kind of a "format" ;-) > For e.g. whether the I/O to/from the codec is > quantized to a uniform PCM format with at least 13 > bits of dynamic range ? and what does the datasheet say about that? as far as I remember the AD535 is set to work with 16bits at 8kHz sampling rate - I worked with that codec long time ago, but I don't remember having any problems with reading the datasheet... do you have such problems? __________________________________ |
Reply by ●August 25, 20032003-08-25
|
Thanks Wojciech, Actually yes. The datasheet does not provide the info I need. I am trying to implent an LPC vocoder which accepts data quantized to a uniform PCM format with at least 13 bits of dynamic range. The AD535 docs that I've gone thru, do not make any mention of this. So, just to rephrase my qn - does the sampled data from AD535 conform to the above (PCM) format ? ... or would some kind of scaling / quantization need to be done ? Better still, could anyone give any pointers to an audio application for C6711 DSK which uses a simple LPC codec? Thanks, Raj. ----- Original Message ----- From: "Wojciech Rewers" <> To: "Raj Sahakari" <>; <> Sent: Monday, August 25, 2003 5:24 PM Subject: Re: [c6x] Input and Output formats of AD535 codec > --- Raj Sahakari <> wrote: > > Does anybody have an idea if there is any input and > > output format for the AD535 codec on the C6711 DSK ? > > I'm sure there is some kind of a "format" ;-) > > > For e.g. whether the I/O to/from the codec is > > quantized to a uniform PCM format with at least 13 > > bits of dynamic range ? > > and what does the datasheet say about that? > > as far as I remember the AD535 is set to work with > 16bits at 8kHz sampling rate - I worked with that > codec long time ago, but I don't remember having any > problems with reading the datasheet... do you have > such problems? > > __________________________________ > |
|
|
Reply by ●August 25, 20032003-08-25
|
--- Raj Sahakari <> wrote: > Actually yes. The datasheet does not provide the > info I need. yeah... now I've gone through the datasheet... it's not specifying it explicitely, but one can assume that since they don't say anything about the output format from ADC it is uniform PCM format (2's complement - that's all I found in the datasheet) - plus the information in chapter 5.3.3 (ADC dynamic performance at fs=8kHz specifying typically 80dB) - I'm sure you have at least 13 bits of accuracy... I have this 96dB SNR for 16bit data in my mind - so I'm guessing 80dB will be either 15 or 14 bits of resolution - you're still fine then... plus - I'm sure if the ADC's accuracy was less than 13 bits - they would definitely mention something about that instead of advertising it as a 16-bits codec. > I am trying to implent an LPC vocoder which accepts > data quantized to a > uniform PCM format with at least 13 > bits of dynamic range. The AD535 docs that I've gone > thru, do not make any mention of this. as I said - I'm sure you'll be fine with AD535 at this point - remember however it's only 8kHz of sampling rate... > So, just to rephrase my qn - does the sampled data > from AD535 conform to the > above (PCM) format ? ... or would some kind of > scaling / quantization need to be done ? as I said - the ADC gives you 2's complement number with no companding (no u-Law or A-low or anything like that) > Better still, could anyone give any pointers to an > audio application for > C6711 DSK which uses a simple LPC codec? can't help you here... good luck anyway... Wojciech Rewers PS. I'm a young DSP engineer (this year's MSEE graduate) with some real life experience (15 months) - looking for an embedded/DSP job - anybody can help me? ;-) __________________________________ |
Reply by ●August 25, 20032003-08-25
|
Wojciech Rewers- > --- Raj Sahakari <> wrote: > > Actually yes. The datasheet does not provide the > > info I need. > > yeah... now I've gone through the datasheet... it's > not specifying it explicitely, but one can assume that > since they don't say anything about the output format > from ADC it is uniform PCM format (2's complement - > that's all I found in the datasheet) - plus the > information in chapter 5.3.3 (ADC dynamic performance > at fs=8kHz specifying typically 80dB) - I'm sure you > have at least 13 bits of accuracy... > > I have this 96dB SNR for 16bit data in my mind - so > I'm guessing 80dB will be either 15 or 14 bits of > resolution - you're still fine then... 96 dB is theoretical limit for 16-bit codecs, no board ever built hits that mark. Every bit is another 6.02 dB, so 13 bits would be about 78 dB theoretical. If TI's data sheet says 80 dB then 13.x bits are valid. But, to WHICH parameters dB measurement are applied is also very important: SNR, typical dynamic range, spurious free dynamic range, etc. But that's another story... Jeff Brower system engineer Signalogic > plus - I'm sure if the ADC's accuracy was less than 13 > bits - they would definitely mention something about > that instead of advertising it as a 16-bits codec. > > > I am trying to implent an LPC vocoder which accepts > > data quantized to a > > uniform PCM format with at least 13 > > bits of dynamic range. The AD535 docs that I've gone > > thru, do not make any mention of this. > > as I said - I'm sure you'll be fine with AD535 at this > point - remember however it's only 8kHz of sampling > rate... > > > So, just to rephrase my qn - does the sampled data > > from AD535 conform to the > > above (PCM) format ? ... or would some kind of > > scaling / quantization need to be done ? > > as I said - the ADC gives you 2's complement number > with no companding (no u-Law or A-low or anything like > that) > > > Better still, could anyone give any pointers to an > > audio application for > > C6711 DSK which uses a simple LPC codec? > > can't help you here... > > good luck anyway... > > Wojciech Rewers > > PS. I'm a young DSP engineer (this year's MSEE > graduate) with some real life experience (15 months) - > looking for an embedded/DSP job - anybody can help me? ;-) |
Reply by ●August 27, 20032003-08-27
|
Hi C6x-ers, Thank you all for your invaluable help. Your suggestions have gone a long way in helping us find the right solutions. Regards, Raj. ----- Original Message ----- From: "Jeff Brower" <> To: "Wojciech Rewers" <> Cc: "Raj Sahakari" <>; <> Sent: Tuesday, August 26, 2003 1:54 AM Subject: Re: [c6x] Input and Output formats of AD535 codec > Wojciech Rewers- > > > --- Raj Sahakari <> wrote: > > > Actually yes. The datasheet does not provide the > > > info I need. > > > > yeah... now I've gone through the datasheet... it's > > not specifying it explicitely, but one can assume that > > since they don't say anything about the output format > > from ADC it is uniform PCM format (2's complement - > > that's all I found in the datasheet) - plus the > > information in chapter 5.3.3 (ADC dynamic performance > > at fs=8kHz specifying typically 80dB) - I'm sure you > > have at least 13 bits of accuracy... > > > > I have this 96dB SNR for 16bit data in my mind - so > > I'm guessing 80dB will be either 15 or 14 bits of > > resolution - you're still fine then... > > 96 dB is theoretical limit for 16-bit codecs, no board ever built hits that mark. > Every bit is another 6.02 dB, so 13 bits would be about 78 dB theoretical. If TI's > data sheet says 80 dB then 13.x bits are valid. > > But, to WHICH parameters dB measurement are applied is also very important: SNR, > typical dynamic range, spurious free dynamic range, etc. But that's another story... > > Jeff Brower > system engineer > Signalogic > > plus - I'm sure if the ADC's accuracy was less than 13 > > bits - they would definitely mention something about > > that instead of advertising it as a 16-bits codec. > > > > > I am trying to implent an LPC vocoder which accepts > > > data quantized to a > > > uniform PCM format with at least 13 > > > bits of dynamic range. The AD535 docs that I've gone > > > thru, do not make any mention of this. > > > > as I said - I'm sure you'll be fine with AD535 at this > > point - remember however it's only 8kHz of sampling > > rate... > > > > > So, just to rephrase my qn - does the sampled data > > > from AD535 conform to the > > > above (PCM) format ? ... or would some kind of > > > scaling / quantization need to be done ? > > > > as I said - the ADC gives you 2's complement number > > with no companding (no u-Law or A-low or anything like > > that) > > > > > Better still, could anyone give any pointers to an > > > audio application for > > > C6711 DSK which uses a simple LPC codec? > > > > can't help you here... > > > > good luck anyway... > > > > Wojciech Rewers > > > > PS. I'm a young DSP engineer (this year's MSEE > > graduate) with some real life experience (15 months) - > > looking for an embedded/DSP job - anybody can help me? ;-) > |
