ADAU1452 and 32bit Word Samples with Filters, Delay, and Volume Control - Confirmation
Hi,
If possible, can someone confirm whether the following is possible with the ADAU1452 and Sigma Studio
16bit/44.1kHz, or 24bit/48kHz I2S signal is input on one of the ports directly to the ASRC, converted to 192kHz.
The 16bit/192kHz or 24bit/192kHz data is left justified to convert the 16bit to 32bit word size, and the 24bit is converted to 32bit word size. It is assumed that the lower bits are filled with zeros.
The 32bit word values are passed through filters for a loudspeaker crossover.
The 32bit values are applied to a delay.
The 32bit words are applied to a volume control as per the examples :
https://wiki.analog.com/resources/tools-software/sigmastudio/tutorials/gpioconditioningexample
The modified 32bit words are then output on the relevant I2S port to a 32bit DAC.
The attenuation applied to the volume control – can it be read from the I2C bus connection to the ADAU1452 ?
If it is possible, are there any issues that could cause a problem ?
Thanks and regards,
Richard.
Audio Signal Processors
adau1452
32bit
volume
SigmaDSP Audio Processors
Hello Richard,
If the I2S data is either left justified or one BCLK delay, (it is still left justified in the 32 bit slot) then it does not matter if it is 16 bits or 24 bits it will work without any special processing. You will just get 16 bit noise performance from 16 bit data and 24 bit performance if it is 24 bits. (Well, assuming the source is producing a 24 bit signal with 24 bits of performance.)
When data comes into the DSP from the serial ports, 24 bits, we add one bit on the bottom and seven bits on the top and the data is then processed as a 32 bit word inside the DSP. Again, this is by default and you cannot change it. All signals are stored and processed as 32 bit data.
The ASRC can convert from 44.1kHz up to 192kHz, no problem there. However, if the only sources are 44.1kHz or 48kHz then there is no point converting to 192kHz fs internally inside the DSP unless you just want to burn up four times the MIPS and have 1/4th the amount of available delay time. You do not gain any resolution because there is no audio above the Nyquist frequency of the incoming audio. The only reason to do this is if you have a 192kHz source and you are adding in or sometimes processing a lower sample rate then it makes sense. So I actually do not recommend you go up to the higher rate. It will make no difference.
Then your last question about going to an external 32 bit DAC. Yes, you can set the serial port to bypass the saturator/truncator and output the 32 bits from the core unchanged. In SigmaStudio you right-click on the output assignment cell and select Data Format and you will see the option there to set it to 32 bit mode.
If you still want to run the DAC at 192kHz then you can run through the ASRC on the output from the core and run the serial port at 192kHz. Again, this will not give you more frequency response but what it will do is make it more difficult to transmit the data on the PCB. It makes the PCB layout more difficult.
So my suggestion is to run at 48kHz and sample rate convert the 44.1kHz input up to 48kHz.
Dave T
Hi David,
Many thanks for the reply, much appreciated.
Thanks for confirming that the requirements can be implemented. I understand that there are no extra frequencies by upsampling to 192kHz, but should i purchase 96kHz recordings - then the system is prepared.
I previously implemented a sample rate converter with the output at 192kHz, but only 16bits, and there were no issues with PCB traces and signals at the relevant data rates. Should there be any issues, i can drop the sample rate.
The reduction in processing resource is not an issue- i am only implementing 4th order filters, a delay, and volume control. Although 32bits is excessive, and the noise floor cannot be heard, or i doubt one can ever hear the lower order bits, i wanted to ensure that i had a lot of headroom for the volume control. Many recordings do not use the full number of bits on the CD.
The ADAU products are unique - there is nothing on the market competing with the product - they seem to offer multiple solutions without having to implement a dedicated DSP that has BGA package or vast number of pins you don't need.
Thanks again for answering, much appreciated.
Regards,
Richard.
Please log in to reply.