I have a technical issue with the ADV7611, we take the I2S serial audio bus and send it to our TI processor for encoding. Recent tests have shown that some of our boards exhibit noise on the encoded
audio - a sort of loud static that overlaps the actual audio content (which can still be heard with the static). While probing the four I2S signals with my probe, I noticed that the noise completely disappears just by holding the probe on the LRCLK line.
I did some research on the Analog EngineerZone forums, and saw a thread
that discusses exactly this issue on another part (the AD1937):
In that thread, they claim to solve this issue by inverting the edge
polarity of the DBCLK output (bit clock). On the ADV7611, this pins is called
SCLK. Looking over the ADV7611 documents (hardware manual, software register
map, etc.), I did not find anything that indicates that this signal can be
inverted. Some of the video-related clocks can be inverted, but nothing about
For the moment, I have a temporary solution: I am manually placing an 18pF
capacitor between LRCLK and GND on each board, to mimic what happens when I
place a scope probe on it. This method appears to fix most of our boards (but
not all). Can you recommend a more definitive solution for this? Obviously it
would be better if we could implement a fix in firmware. But if the permanent
fix requires hardware changes, that's ok too. Just FYI, the four I2S traces
(AP, LRCLK, SCLK, MCLK) are length-matched pretty well, all ranging between
1.82 to 1.90 inches. And all use a 75-ohm series resistor in their path.
Can you help?