Post Go back to editing

AD4111 crosstalk

To whom it may concern,

We are facing some crosstalk issues with the AD4111 in differential voltage mode. Applying 20Vpp at one channel results in 0.6Vpp at the other three channels (3% or -30dB). The crosstalk is not only seen when the input is left floating but also if we short the input. The external circuit is as suggested in the eval board.  Different tests done made us quite sure that it is not a problem on the PCB and rather inside the ADC. Are there any known problems (both with the PCB as well as with the ADC and its configuration) that could produce the crosstalk we are seeing? We did not find any numbers in the datasheet regarding the crosstalk. Because of this we do not know what to expect under “normal” conditions.

Thanks a lot in advance for your support

  • Hi, 

    May I know what output data rate did you used? Please take note that with the highest ODR of Sinc5+Sinc1 the ADC noise is around 750uVpp. Can you lower down the ODR and see if the measurement at the other channels changed? 

    Can you also clarify if you have perform internal gain calibration or change the calibration register as stated in the datasheet regarding VIN measurements? 

    Thanks,

    Jellenie

  • Hi Jellenie,

    thanks for your response. We are using the following setup

    • ADC Mode: 0x8000
    • Interface Mode: 0x0040
    • Setup Configuration: 0x1320
    • Filter Configuration: 0x0D11

    resulting in the following modes

    • 4 differential voltage inputs
    • Gain and Offset are automatically calibrated using 0x8040 and 0x8050
    • 20 SPS
    • Sinc5 + Sinc1
    • Enhanced filter for 20SPS 86dB rejection
    • No mux delay

    Are there any mistakes in our configuration?

    Thanks

  • We did some more testing:
    The crosstalk is not present if we short the input, that was a mistake made by us. Using different resistor values on the input it could be seen that the crosstalk depends on these external resistors. Furthermore, we could see that current inputs do not create crosstalk on floating voltage inputs.

    Nevertheless, is there a way to reduce the crosstalk to floating inputs? Or do we have to accept that we always must connect a source with relatively low input impedance? In some cases, this would result in problematic situations because a broken sensor wire to the ADC could be difficult to spot from the ADC readings. I know that there is the open wire detection, but using it would halve our output data rate.

  • Hi, 

    Apologies. I missed this thread. 

    In general when VIN is floating and VINCOM is selected as the other input pair. The internal resistive divider of the part will generate an ADC input voltage around mid supply. One possible solution is to operate with +/-2.5V split supply but that would add cost in the power supply circuitry, but that could result a floating input near to 0V. 

    I agree that that using the built-in open wire will halved your ODR but if you have some flexibility around the measurements for example wherein you only need to spot check an OWD then during the OWD check only will half the ODR but at normal conversions you can returned to your actual ODR. Would this be possible in your end? 

    Thanks,

    Jellenie