Question about AD7176-2 THD performance 

Hi ADI,

I short  EVB AD7176-2 input pin with GND to get ENOB=19.8bit, according to MT-003, SINAD = 6.02*ENOB+1.76dB = 120dB.

And CN0310 shows AD7176-2 can measure SINAD 99.9dB.

I use APx515 to transmit pure 1kHz sinewave to test AD7176-2 RX quality, and use the raw data which get from AD7176-2 to calculate THD+N via MATLAB FFT, I can only get SINAD 75dB.

The result is really confused me.

----------------------------------------------------------------------------------------------------

1. Because I can't find the FFT tool on AD717x Eval+ Software,

could you teach me how you get the FFT as above figure 6? 

2.  The attachment is raw data, might you use the tool to help me check it?

Thanks a lot.

Homeway

ADI_output.csv.zip
  • 0
    •  Analog Employees 
    on May 17, 2018 1:37 AM

    Hi Homeway,

    ENOB is a measurement of the resolution with a sine wave input. It is a function of SINAD for a certain frequency. Thus, having the input connected to ground will not give you the exact value of ENOB in 1kHz signal. On the other hand, SINAD is the ratio of the rms value of the signal to the rms value of all spectral components including harmonics but except DC.

    The difference of your result, the 75dB, the actual SINAD, 99.9dB, can be caused by different input amplitude. The actual SINAD was measured at -6dBFs or 2.5Vpp input with 5V reference. The setup can also cause noise that can lower the measured SINAD. For this, can you show your current setup.

    Thanks,

    Daryl

  • Hi Daryl,

    Thanks for your reply.

    My current setup : 

    Hardware setting : EVAL-SDP-CB1Z connect to EVAL-AD7176-2SDZ

    Power : Power supply 9V & -2.5V via J9  

    ODR : 50kHz

    Reference : External Reference, Ref+ : 2.5V , Ref- : -2.5V

    Filter : Sinc5+sinc1

    Input signal : Single Ended 0.5Vrms 1kHz sinewave (0V offset).

    ---------------------------------------------------------------------------

    1. For ENOB, is "RMS resolution" in EVB SW equal to ENOB? 

    2. If 1. is right, then it is a little strange.

    If I measure 1kHz 0.5Vrms signal, the ENOB is not good as blow picture.

    I think when the input is pure clear with out any signal, then the measurement is only system noise, and the noise will decreases the efficient number of bits. 

    So I short the input to GND to get the ENOB on this state.

    3. If the system noise is the same no matter how the input level, then consider my input signal 0.5Vrms (-11dBFS), SINAD= 99.9dB-(11-6)=94 dB.

    It is still a gap between my test result.

    4. Could you teach me how to get SINAD for AD7176-2 via ADI's tool?

    Thanks for your kindly reply again.

    B.R.

    Homeway

  • 0
    •  Analog Employees 
    on May 23, 2018 11:38 PM

    Hi Homeway

    1. RMS resolution and Peak-to-peak resolution is not the same ENOB. ENOB is a measurement of the resolution with a sine wave input while RMS resolution and Peak-to-peak resolution are measured with DC input.

    2. The reason why you got low P-P resolution and RMS resolution is because the evaluation reads the 1khz signal as noise. If you can see the P-P Noise is 3.10646 V with is also the actual peak-to-peak voltage of your input. As what I said earlier, RMS resolution and Peak-to-peak resolution are measured with DC input and any higher frequency will be treated as noise.

    Generally, Sigma delta ADC are not designed to measure high frequency signal. If you will measure the characteristics of high frequency signal, ex. 1kHz, you can use SAR ADC. EVAL software of SAR ADC will give you the AC characteristic like SNR, THD, SINAD and ENOB. 

    For your reference, kindly check the NOISE PERFORMANCE AND RESOLUTION section and table 6 of the datasheet.

    Regards,

    Daryl

  • 0
    •  Analog Employees 
    on May 29, 2018 10:02 PM

    Hi, Homeway.

    AD717x series are designed for multiple channels dc type applications. You may want to consider using the AD7768 ? It is an 8 channel or 4 channel (AD7768-4), simultaneously sampling, Sigma Delta ADC suitable for audio type applications. At the fastest output data rate of 256kSPS this ADC can achieve a minimum of 106dB SNR. The noise performance can be optimize depending on the decimation rate selected and required input bandwidth. You can disable channels that are not needed to save power. 

    Thanks,

    Jellenie

  • Hi Daryl,

    Thanks for your reply.

    I understand the meaning for  Peak-to-peak resolution now. I have wrong concept before.

    Actually, I want to use AD7176-2 for audio band, according the paper Which ADC Architecture Is Right for Your Application?, I think I can still use Sigma-delta ADC.

    And I find my problem maybe phase noise.

    Below picture is the result that measure 1kHz sinewave with EVAL-AD7176-2, then calculate the raw data by MATLAB FFT.

    I think I need to use precision crystal.

    Can you suggest the value of capacitor C11 and C10?

    Thanks a lot.

    B.R.

    Homeway