Jack--
Let me begin by addressing the title of your post. The OMAP-L137 isn't an audio codec. It's a DSP. As a professional in this field, surely you understand that the L137 (DSP), the AIC3106 (the audio codec), and the board itself all contribute to the overall audio performance. I'm not sure how you get to the conclusion that because you did not get your expected SNR, the L137 has poor audio performance. Again, it's a DSP--it's just applying processing. The only way that it will corrupt the signal is if your processing algorithm does so. If you're just doing a passthrough, then I guarantee that it's not dropping your SNR from 90 to 50 dB. For the record, this DSP has been used in literally millions of high end audio products over the last several years.
"However, its performance is very lower than that mentioned in datasheet."
As you must realize, the fact that the AIC3106 can achieve 90dB SNR per the datasheet does not imply that one can design it onto any board, with any components, and get 90 dB performance. I imagine that if you got an AIC3106 EVM, you'll see something really close to that number, because the components, layout, etc will have been optimized for audio performance. But the OMAP-L137 EVM is not designed specifically for audio. It has some basic audio capabilities for demonstration purposes, but it's not a high fidelity audio platform, and doesn't claim to be. We designed this board to demonstrate most of the peripherals on the L137, of which McASP is just one, and as such we've never taken analog audio performance measurements for the board; that was never the point.
Incidentally, as this is the OMAP forum, the 100+ AIC3106 register values that you pasted in won't mean anything to us as we don't support that device.
"We used the default setting of the codec and board. I think the default setting of a board primarily should give the best signal in terms of SNR."
This is just not a good assumption to make. There are too many variables; there's no way for us to know how the customer will want to configure the DSP and codec.
I'm not sure if you have started with some example code, or written your own from scratch. But again, I'd recommend taking a look at the code that we linked in your other thread as a starting point. Set up an example wherein the DSP is doing nothing but audio passthrough as a starting point. I don't know why you're not getting the performance that you'd like, but I guarantee that it's not because of the DSP's audio performance, and I'm pretty certain that the AIC3106, which has also been used in MANY audio products, performs as documented. In my opinion the most likely sources for error here are (in no particular order):
1. Bad DSP code
2. Bad configuration (for instance, make sure that the codec launches data on the rising edge and DSP samples it on the falling edge, or vice verse...check that clock polarities, bit depth, justification, etc settings are all in agreement between the AIC and the McASP).
3. Clock settings (have you looked at your input and output clocks? Do they look clean?)
4. Board - maybe the board design doesn't provide great audio fidelity. It's entirely possible.
--Bobby T.