ADAU1761 DC gain in the digital mic path

Hello,

It would theoretically be possible to use a DC Input Entry block, and add that to the signal to compensate for a constant and known DC offset, but the preferred way to eliminate a DC offset would certainly be to use the onboard HPF... Are you using the ADAU1761 evaluation board, or a custom design?

Regards,

Tyler

Hi Tyler,

Thanks for your quick response.

I am working with Mahesh. We can compute mean of the signal and compensate for DC. However, we want to understand the root cause. We are using the EI3 audio extender along with BF707 EZ-Kit. We are playing signals from PC through a speaker and measuring input signals at microphone to characterize the mic audio path.

Why do we see a DC offset on the digital input of microphone signal?

We also see that the microphone signal does not increase beyond +-2800 irrespective of the signal strength we feed for a fixed location between a speaker and on board microphone. Would you know of any reason why the signal would have such a low dynamic range?

Hi Tyler,

Thanks for the response.

Adding to the above, ADAU1761 has two omnidirectional MEMS microphones, which are analog devices part number ADMP421 ICs (U3 and U4 on the board). About this microphones, how many data bits the ADMP421 is giving and what is the format of the data?

Thanks

Mahesh Muni

Hi Tyler,

Adding to the above information,

To remove the DC offset from the captured mic signal without using the HPF, calculated the mean of the captured mic signal and subtracted from the captured mic data (2k buffer each time capturing from the digital mic), but still some DC offset observed in the signal. Here are the wave forms with DC offset(Present) and mean subtracted wave from(Removed). (used build in 'mean' function in matlab to estimate the value for 2k buffer data of int16).

Can you please help on this how to compensate the DC offset in the signal on ADAU1761 without HPF turning ON?

Also, how many data bits the ADMP421 is giving and what is the format of the data?

Thanks

Mahesh Muni

Hello,

The likely reason behind the DC offset you see is that there is an inherent DC offset in some Sigma-Delta ADCs. Since the ADMP421 has a built-in one-bit modulator, the offset you're seeing is likely inherent to the part. However, this offset is generally temperature-sensitive, so it is advisable to use a high-pass filter of some kind instead of compensating. Again, I would recommend using the onboard HPF -- it really is the most appropriate option to correct this behavior, as a 2Hz highpass. Is there a specific reason why you are not able to use it?

If you are not able to use the hardware HPF, there is an option of using the "DC Blocking" filter in SigmaStudio (located in Filters -> Miscellaneous). This is also a very low-frequency high-pass filter, and should do the job well.

As far as how many bits the ADMP421 gives and their format: the ADMP421 has a PDM output, which means that there is only one bit per channel per clock cycle.

Raj: Could you please clarify what scale it is that you're seeing +-2800 on?

Best regards,

Tyler