ADSP-SC589_EZ-Board Analog input noise

I have a ADSP-SC589_EZ-Board, and I am using it with SigmaStudio for SHARC 4.6 and CCES 2.10.0, Windows 10.

I notice that I get a 3kHz noise signal with harmonics when I connect a source with a finite output impedance. In fact, as soon as I connect a dummy source (simply a connector with an impedance of 600 Ohms between pin and GND) to the analog inout I get this noise signal.  The signal level is equivalent to 1mV p-p. This gives me a maximum S/N ratio of 50dB, which is not impressive, and it is not enough for my application as it is very specific and clearly audible. White noise would have been more acceptable.

Is this known, or is there something wrong with my EZ board?

Is there any way I can improve the situation?

BN



correction of misspell
[edited by: bnilsson at 3:23 PM (GMT -4) on 20 Sep 2021]
  • 0
    •  Analog Employees 
    on Sep 20, 2021 5:00 PM

    Hi BN,

    Have you run the POST test for the DAC/ADC to verify that your evaluation board is okay?

    This POST test can be found in the ADSP-SC5XX Board Support Package. URL here:

    http://download.analog.com/tools/EZBoards/SC5xx/Releases/Release_2.0.2/ADI_ADSP-SC5xx_EZ-KIT_Lite-Rel2.0.2.exe

    -Matt

  • 0
    •  Analog Employees 
    on Sep 21, 2021 4:24 PM

    Also, can you explain why you are connecting a 600ohm resistor between the ADC input pin and GND? 

    -Matt

  • I will try the POST, thanks for the suggestion.

    The reason I "shorted" the input with 600 ohms was a part of the troubleshooting process, to eliminate the risk that the noise signal would come from the outside.

    If I don't connect anything to the input port it is the same as connecting a source with "infinite" output impedance. In my particular case I get no 3kHz noise, only normal random white noise.

    If I connect my normal signal source, I get the 3kHz noise. How can I be sure the signal is not coming from my own signal source?

    If I want to connect an absolutely "quiet" signal source, I connect an RCA connector without cable, with 600 Ohms directly connected between signal pin and GND, mimicking a zero signal source with a realistic output impedance. Zero=0 ohms, a real SHORT from input pin to GND has a risk of introducing other problems such as parasitic ground loops, dependent on how the input circuit layout is done. The to select the impedance value here is somewhat arbitrary, 50 Ohms or 600 Ohms are common output impedances of signal sources. But preferably not zero.

    Now I have made sure that I have no signal coming from the outside, any signal on the output is generated internally in the analog input stage.

    The fact that I get "no noise" with an open input circuit,  and 1mV 3kHz with 600 Ohm connected, indicates that the input noise circuit might look something like this:

    Without the input resistor, the input noise voltage sources Vn1 and Vn2 have no effect. With the resistor connected, the input is a closed circuit with Vn1-Vn2 as the input signal, amplified by the input stage so that it appears at the output. Vn1 and Vn2 may be different, correlated or non-correlated.

    Does this explain why?

    BN

  • 0
    •  Analog Employees 
    on Sep 21, 2021 7:09 PM in reply to bnilsson

    Thanks for the additional information! I will review with some other engineers.

    BTW: when you did these measurements, was a JTAG debugger connected to the SC589-EZKIT? Several engineers have told me that having a JTAG debugger connected can cause noise on that board.

    -Matt

  • The noise is present also in a "booted runtime situation" where the EZ board is mounted in a metal enclosure with the cover on, far away form any JTAG debugger, SigmaStudio project download devices or USB cables.

    I have measured both while debugging and in the above described configuration, and there is not much difference.

    BN 

    Oscilloscope screens of 1kHz1mV in, 0mVin, and Disconnected (open input).