TAS6424M-Q1: Weird noise issue and Register settings

Hello Tuan,

Unfortunately, I will need to ask a few questions before I can fully answer your full post.
The Register settings look good and the volume gain is correct.

1.  I will check this on the bench and reply.

2.  To measure noise, you should filter the signal so the out of band noise and switching frequency are not part of your measurement.  Also, wait for 30 seconds after placing into play mode to measure noise for the system to settle.  The -18dBFS oscilloscope screen capture shows aproximately 2Vp or 4Vpp signal.  A -18dBFS signal is 0.1259 x 15V/dBFS which is 1.89Vp.

All modes are described in Section 9.3.1 of the datasheet.  DSP mode is absent.  It is I2S mode that has an FSYNC pulse like TDM instead of an LRCLK.

Hello Scott,

Thank you for the response.

For the noise measurement, I understood that the frequency of noise is over 20KHz so if I apply A-weighted filter then the THD and SNR measurement may be good. But still I don't think 500mV noise like this is normal? And the noise is there always even after 30s settling time.

For the gain setting, our -18dBFS oscilloscope screen was with attenuation of 2:1 (because my differential probe max of 8V, so I used resistor voltage divider 2:1 in parallel with the load. I know this is not the good way to check audio quality, but to see the shape of waveform I think it is OK). So the peak voltage was 4V, not 2V as in screen, and peak-peak voltage was 8V, more than double the expected value.

Honestly we once got good waveform with same hardware setup and different register setting. But now due to some reasons it is not easy to request software team to change the setting, so I would like to check and understand the issue first.

So waiting for your bench test result, and also I am waiting for the EVB board arrival to check from my side.

Thank you.

Tuan,

I will wait for you THD and Noise measurements.  I will be testing your settings tomorrow with an Audio Precision. 

Please send me your latest settings, so I can test with them also.

Hi Scott,

Our latest register settings are in first post. Could you please help to replicate and check the noise and output amplitude on your board? I doubt the TDM input format setting.

Our EVB will arrive next week, and it will take some time before I can setup the software and play around with different register settings.

Thank you.

Hello Scott,

Thank you for your test result.

I did the test on our EVM, same TDM format and register settings. The noise is still present and very consistent:

 - Around 10s after a Power on reset or Software reset, there is noise pattern on and off. After 10s the noise disappears. This is acceptable.

 - Now after 10s settling time, the TDM SCLK, MCLK, and FSYNC are present, but data is zero as I am not playing. There is no noise.

 - Once I start playing (data is not zero), noise appears. And if I stop playing, noise is still there. From this moment, there is always noise pattern embedded into output signal.

 - Mute by asserting MUTE pin or set channel state control register to Mute: noise is still there.

 - Assert STANDBY pin: noise disappears. De-assert STANDBY pin: noise re-appears.

 - Trying different INPUT FORTMAT (SAP control register): same result.

The noise amplitude is around 200mV, frequency is around 100KHz.

Did you get only 7.8mVrms noise with your test setup without any filter applied? Can you check again? Please note that after POR there is no noise, noise appears once I start sending audio data.

I agree this noise is out of audio band, but seeing it embedded in the waveform is very weird. Below is 1KHz sine wave output from EVM. Is it expected behavior of this device?

Thank you. 

Hello Scott,

Then I will ignore this noise for now and apply filters for performance measurement.

We can close this issue.

Thank you for your support.