DA7212- Ripple at Mic Bias

5 posts / 0 new
Last post
akshay_pandit
Offline
Last seen: 6 months 1 week ago
Joined: 2018-09-21 07:56
DA7212- Ripple at Mic Bias

Hi,
Overview - DA7212 has been used as audio codec with Mic 1 configured as pseudo-differential input with headphones connected to HP_L and HP_R of codec. All other other peripherals like AUX mic, Mic 2 and speakers are unused. Codec I2S and I2C lines are connected to STM32 MCU. We have BT controller for wireless communication. We are using A2Dp and HFP profiles for wireless audio communication.
We have observed continuous background noise even-tough mic is not connected to codec. Basically, we used audio recording app to record audio through mic 1 from codec.Recorded clip has noise from frequency(200Hz to 4kHz). To evaluate app and firmware are not sources of noise, a Loopback test is conducted and results in same noise. Loopback test is basically a test to loop back audio data from Mic 1 to HP_L and HP_R by selecting Mix_in_Las input to Mix_out. While root causing the issue, it is found that mic bias is not a stable DC. Mic bias has ripple, causing noise from mic input.
File attached contain figures to explain the cause. VDD_A has ripple, however I'm expecting PSRR of codec should have handle these ripple and provide a ripple free mic bias.

Attachment: 
Device: 
bmcadam
Offline
Last seen: 1 month 3 days ago
Staff
Joined: 2015-09-16 16:25
Hello,

Hello,

MIC Bias is effectively an internal LDO and takes it supply directly from VDD_MIC, which I note you are supplying 2.5V, which also supplies VDDA. Any ripple on the MICBIAS output is only going to come from the supply of VDD_MIC and we note there is no output capacitor on the second DC-DC convertor unlike the first one which has 22uF on their outputs. I believe you will need an output capacitor on that DC-DC convertor.

As a test I would disconnect 2.5V supply and then supply externally to see if the problem stills exists.

akshay_pandit
Offline
Last seen: 6 months 1 week ago
Joined: 2018-09-21 07:56
Hello,

Hello,
Thanks for reply.
I agree with the opinion that DC-DC need output capacitor.

However, VDD_MIC and VDDA can be driven by 1.8V DC-DC as well. Attached report shows waveform of Mic bias and VREF at VDDA,Mic bias at 1.8V. Significant amount of noise is still present.

The same setup has been replicated on audio shield by connecting 1.8V DC-DC supply. Attachment contains waveform of shield at Mic bias and VREF. Both waveform have consistent ripple.

It is unlikely that PSRR is not effective for these ripple. Can you please confirm this?

Attachment: 
akshay_pandit
Offline
Last seen: 6 months 1 week ago
Joined: 2018-09-21 07:56
Hi,

Hi,
As you can see in the codec circuit diagram, we have used GNDD and GNDA. It is observed that GNDA to GNDD has ripple.
By reducing this ripple, i.e, GNDA to GNDD, the issue of noise is eliminated. Would like to know the architecture of GND_A, GND_CP and GND_Sense in codec.

bmcadam
Offline
Last seen: 1 month 3 days ago
Staff
Joined: 2015-09-16 16:25
Hello,

Hello,

Gnd_A is effectively our main ground for all the Analogue and references within thr device.

Gnd_CP is the charge pump.

Gnd_Sense is the connection to the outside world for the headphone, which feeds back into the amplifier.

Let me know if this answers your question.

Best Regards,