Hi guys,
I am doing some studies about ASRC IP, mostly on architecture. So I am trying to look for some reference IPs.
Although I found some from TI, Analog Device... but they are datasheets; there isn't much explanation on the architecture. I did find one from Xilinx with product guide; but it is discontinued IP so I don't know if it is a reliable resource.
I hope you could recommend me some IPs with detail review on architecture.
Thank you.
I am doing some studies about ASRC IP, mostly on architecture. So I am trying to look for some reference IPs.
Although I found some from TI, Analog Device... but they are datasheets; there isn't much explanation on the architecture. I did find one from Xilinx with product guide; but it is discontinued IP so I don't know if it is a reliable resource.
I hope you could recommend me some IPs with detail review on architecture.
Thank you.
Thanks abraxarito.
I read this one, but I'm not quite sure I understand how the servo tracking loop and the RAM FIFO work.
I read this one, but I'm not quite sure I understand how the servo tracking loop and the RAM FIFO work.
The TI patent US 7262716 B2 provides some more information on what TI has put on the SRC4392 and similar chips. As far as I remember, it's quite readable for a patent.
US 7330138 B2 describes an asynchronous sample rate converter used by ESS. They use some weird structure that only works when you want to interpolate and that only changes one sample per input sample period. It must have an inherent sinc-shaped roll-off, because they repeat the input sample. US7953782 and US7436333 also have something to do with it.
Old ASRC like AD1890 is preferable to understand the architecture. It has some structural information in the datasheet. If you want to use ASRC, I recommend the latest one like SRC4392
http://www.analog.com/media/en/technical-documentation/data-sheets/AD1890_1891.pdf
ASRC must have very high OSR more than GHz to achieve enough SNR(more than 120dB) for audio application because of its inherent nature. But GHz sampling rate and large coefficients are not applicable. TI or AD probably use some trick to implement such function into a real chip. It usually includes some patents. Understanding of the architecture doesn't always require to search related patents. But their trick is also one kind of the architecture.
http://www.analog.com/media/en/technical-documentation/data-sheets/AD1890_1891.pdf
ASRC must have very high OSR more than GHz to achieve enough SNR(more than 120dB) for audio application because of its inherent nature. But GHz sampling rate and large coefficients are not applicable. TI or AD probably use some trick to implement such function into a real chip. It usually includes some patents. Understanding of the architecture doesn't always require to search related patents. But their trick is also one kind of the architecture.
- Status
- Not open for further replies.