TAC5212: TAC5212: Biquads of the record path do not work

Hi Sebastien,

I am curious if the all-zero filter has the effect you intend (no-pass)? You can use this tool to calculate coefficients for specific biquad filters at 8kHz FSYNC: https://www.ti.com/tool/COEFFICIENT-CALC

I would try a 1kHz notch instead.

Best,

Garret

Hello Garret,

Thanks a lot for your help. I will try your proposal asap.

But note that all-zero filters behave as no-pass (no signal) when applied to the playback path. That's why I expected the same result when applied to the record path. But maybe I am wrong... let's see!

Best regards,

Sebastien

I see, let me know the updated result.

Hello Garret,

I replaced the filter setting of the biquads of the record path with a 1kHz notch,

and modified the TAC5212 configuration accordingly: 

w a0 00 08 #page 8
w a0 08 5a 82 79 99 c0 00 00 01 5a 82 79 99 3f ff ff ff ca fb 0c cd #ADC biquad 1
w a0 1c 5a 82 79 99 c0 00 00 01 5a 82 79 99 3f ff ff ff ca fb 0c cd #ADC biquad 2
w a0 30 5a 82 79 99 c0 00 00 01 5a 82 79 99 3f ff ff ff ca fb 0c cd #ADC biquad 3
w a0 44 5a 82 79 99 c0 00 00 01 5a 82 79 99 3f ff ff ff ca fb 0c cd #ADC biquad 4
w a0 58 5a 82 79 99 c0 00 00 01 5a 82 79 99 3f ff ff ff ca fb 0c cd #ADC biquad 5
w a0 6c 5a 82 79 99 c0 00 00 01 5a 82 79 99 3f ff ff ff ca fb 0c cd #ADC biquad 6
w a0 00 09 #page 9
w a0 08 5a 82 79 99 c0 00 00 01 5a 82 79 99 3f ff ff ff ca fb 0c cd #ADC biquad 7
w a0 1c 5a 82 79 99 c0 00 00 01 5a 82 79 99 3f ff ff ff ca fb 0c cd #ADC biquad 8
w a0 30 5a 82 79 99 c0 00 00 01 5a 82 79 99 3f ff ff ff ca fb 0c cd #ADC biquad 9
w a0 44 5a 82 79 99 c0 00 00 01 5a 82 79 99 3f ff ff ff ca fb 0c cd #ADC biquad 10
w a0 58 5a 82 79 99 c0 00 00 01 5a 82 79 99 3f ff ff ff ca fb 0c cd #ADC biquad 11
w a0 6c 5a 82 79 99 c0 00 00 01 5a 82 79 99 3f ff ff ff ca fb 0c cd #ADC biquad 12

But, again, I saw no effect. Note that the same filter setting applied to biquad of the playback path works well.

Is something wrong or missing in my configuration?

Best regards,

Sebastien

Hi Sebastien,

I think I figured this out. The biquad effect is being countered by the AGC, which is gaining the notched signals back to the target level. If you disable the AGC (page 1, register 0x24 = 0x00), you should see the recording biquads take effect.

Best,

Garret

Hi Garret,

Good catch! By disabling the AGC: 
w a0 24 00 #DRC ch4 = off, DRC ch3 = off, DRC ch2 = off, DRC ch1 = off, AGC ch4 = off, AGC ch3 = off, AGC ch2 = off, AGC ch1 = off
the biquads of the record path work as expected.

Note that the all-zero filters act as no-pass (no signal) filters.

Therefore it appears that the biquads of the record path are ignored when the AGC is enabled. Can you confirm? because the datasheet does not mention this behavior.

I am a little bit frustrated because according to Figure 7-27, the AGC is located before the biquads. Thus how can the biquads effect be countered by the AGC? and what about the no-pass filter? Atm the biquads of the record path are simply ignored when the AGC is ON.

Best regards,

Sebastien

Hi Sebastien,

I have reached out to our design team for comment on the signal chain order and why the biquads are countered by the AGC.

Best,

Garret

Hi Garret,

Thanks Garret. I am waiting for your design team feedback.

Best regards,

Sebastien

Sure, I'll let you know when they get back to me.

Hello Garret,

That's really bad... but thank you very much for your effort.

Best regards,

Sebastien