Hi, team,
My customer is using PCM5122 for an audio decoder, PCM5122 is after their DSP. They need to achieve 90dB THD+N for 2Vrms output.
However, as below test, they can only achieve 88dB on their own board. To locate the issue, we tested on our PCM5122 EVM with I2S input from AP, still the THD+N was only 88dB.
Below is picture for AP test.
Hi Nicole,
The PCM5122 is only specified for 92dB THD+N for an input of -1dBFS.
This is because the full-scale output of the PCM5122 can sometimes see distortion caused by lower (though still valid) CPVDD voltages which results in a lower magnitude VNEG voltage. This can be seen more on devices that have higher gain error (larger full-scale voltage) and when there is a resistive load on the output.
I recommend the following:
1. Test the output at -1dBFS
or
2. Do some or all of these changes:
A. Increase the CPVDD voltage to a larger value: 3.4V.
B. Reduce resistive loads on the output (high resistor values are better).
C. Change the device from VREF mode to VCOM mode. In VCOM mode the output magnitude is determined by the supply voltage so the output will scale to be slightly lower if the VDD supply is lower. This can have increased noise as the noise on the VDD pin is only filtered by the VCOM capacitor.
Thanks,
Paul
Hi Nicole,
I received your email.
To clarify, THD+N of 88dB at 0dBFS can be considered expected behavior.
In the PDS we specify a typical value of -93dB at -1dBFS, and a max of -83dBFS. The best performance of the device occurs at a lower amplitude, as you have seen.
The actions I listed above can help improve the THD+N performance, but ultimately the performance at full-scale is set by process variation. It is mostly impacted by the gain error of the internal reference and the I/V conversion stage. For this reason, sometimes the performance is improved by switching to the VCOM mode, as the internal reference is not used.
Thanks,
Paul
