• Resolved

PCM6240-Q1: PSRR,CMRR, THD+N vs input amplitude

Prodigy 110 points

Replies: 3

Views: 69

Part Number: PCM6240-Q1

Hi, 

I have 3 question in PCM6240 datasheet,

1) In the table, the PSRR is 92dB at 1kHz.

But in the picture, the picture displays can't reach the limit.

I want to test this parameter, could you tell me how to test, especially in the DC-coupled differential microphone input(INxP=6V and INxM=2V)?

2) In the table, the CMRR is 70dB at 1kHz.

Is this parameter tested in DC-coupled differential microphone input? 

And could you tell me how to test?

3) In P22, the x-axis in the curve of THD+N vs Input Amplitude can reach to -12dB,

why this input amplitude can reach 0dB, because the nom value is 10Vrms for microphone input in the table of P14. 

  • Hi,

    The plot of PSRR is from a typical unit and is not meant as a strict limit. There will be some expected variation around the 92dB value given in the spec table. To test PSRR you need to inject a tone into AVDD. Bypass capacitance is always recommended but you would either need to remove any bypass capacitance or overdrive it with the signal injector and measure the resulting voltage at the IC pin. You can then convert this to dBFS and measure the resulting output level from the ADC. Note that the tone should be large enough such that you can distinguish it outside of the noise floor of the IC, but not so large that you push the supply outside of the recommended operating range. 100mVpp is commonly used. Using a high number of FFT points when you analyze the output spectrum can help reduce the FFT noise floor to more easily distinguish the tone. The PSRR is then the difference between the ADC output level and measured tone on AVDD. You would repeat this for any other frequencies you want to measure the PSRR at. 

    The CMRR parameter testing follows a very similar procedure but instead it is the common-mode input that you apply your signal to. In the test conditions it is mentioned that it is tested in AC coupled mode with a 1Vrms signal on each pin. 

    The curves vs input amplitude are only shown to -12dBFS because this is the practical limit when DC coupling with the bias scheme given and an 8V micbias. With an 8V micbias and INxP at 6V and INxM at 2V, each pin can swing 4Vpp for a total 8Vpp swing since it is differential. Converting to RMS this is 2.828Vrms, which is around -11dBFS with a 10Vrms input. Allowing for some margin away from ground and the 8V micbias rail results in around -12dBFS as the max recommended input for these conditions. For signals larger than this you will start to see the distortion increase significantly.

    Best,

    Zak

    Regards,

    Zak Kaye
    Precision Data Converter Applications

  • In reply to Zak Kaye:

    Hi, 

    Thanks for your great support.

    1) About PSRR your said the difference between the ADC output level and measured tone on AVDD.

    I want to know should we consider the attenuation of ADC(if PGA gain=0, DVC gain=0)?

    2) I have a general question about how to test the parameters about microphone input.

    As you know the test condition INxP is 6V, INxM is 2V in the datasheet, if we use AP's balance input, the offset of INxP and INxM is only setup the same voltage. And the AP's drive ability is very strong, so the ADC's INxP and INxM couldn't work at suitable bias voltage.

    Do you have any good suggestions, thanks.

  • In reply to user6199457:

    Hi,

    The analog PGA gain should not have an impact on the result, but it does depend on the exact path through which the supply noise couples into the signal. These are generally non-linear parasitic pathways. The digital volume control will have an impact since this is just a bit-shift of digital data and the supply noise would definitely be coupling prior to this stage. You should keep the digital volume at 0dB during PSRR measurement.

    Depending on the AP you are using there should be an option to set the output impedance of the balanced analog out to 600 Ohms. If you do this, set MICBIAS to 8V, and change the bias resistors on your board to 300Ohms , this should bias the inputs to the level you want!

    Best,

    Zak

    Regards,

    Zak Kaye
    Precision Data Converter Applications