Ad7767 (CN0540) 60Hz noise pickup

I am having some serious noise issues with CN0540/ AD7767 picking up too much noise:

1. There are low frequency (<few Hz) fluctuations. Setting Vocm to 1V1 or 0V9 seems to reduce those but not entirely.  While using Vocm =2V5 or mid supply make output constantly jump up to +/-1.5v making the whole setup unusable . Disconnecting internal constant current supply seems to eliminate the issue . Wrapping the entire board in aluminum foil and connecting to board GND also solves it, even with the internal current source enabled.

2. There is a huge spike right at 60Hz mains frequency (30Db above the noise floor). It persists even with output disconnected. Removing the jumper to constant current supply cuts the  60Hz peak to roughly15Db above the noise floor. Shielding the board and connecting the shield to the board ground eliminates this spike even with the current source enabled, however connecting coaxial cable with IEPI to the input brings the 60Hz back up. 

Any suggestion/recommendations? 

  • 0
    •  Analog Employees 
    on Jan 6, 2021 6:36 PM


    Can you tell me a little bit more about your setup?

    What settings for the capture (data rate, power mode)?

    When connecting IEPE cable, is there a sensor also connected?

    If using a laptop, does running from battery only improve things?



  • Our application is for very low frequencies, so running at the lowest setting (1ksps,  Low power mode).

    CN0540 is mated with DE10 FPGA card which receives 5V from the AC adapter and also connects to the AC powered network switch. Data is transferred to the PC over the Ethernet. So can't really eliminate 60Hz power here

    Connecting and disconnecting IEPE sensor does not make a difference on the 60Hz spike. However if I short the cable at the sensor input, the spike is gone.

    We  also tried connecting RG 58 C/U coaxial cable with open end and noise is almost non existent. I don't know what type is that IEPE cable, but  it is what IEPE manufacturer supplied. Could it be impedance mismatch?  

  • 0
    •  Analog Employees 
    on Jan 12, 2021 11:25 AM in reply to Alexo57

    The CN0540 board itself shouldn't pick up the noise directly. Shorting the cable at the sensor input should indeed remove the noise anyway, since there is only one node here that should amplify any signals, and that is the input.

    If the coax cable is not picking up the noise in the open-ended case, then it suggests that the board itself is good. The input impedance is not truly high-impedance, as it expects an input from an active sensor. Because of that it should not pick up too much noise when left open-circuit. And it should not be sensitive to impedance mismatch either, as long as the sensor output impedance is not too high (<500R maybe) When the cable is open-ended the 2.5mA current from the current source flows through the input resistor in this case.

    When the sensor is connected, is it grounded only by the coax cable shield, or is there another ground return path from the sensor? Is the sensor mounted to a grounded machine?

    Do you have any alternative power supply that you could try to supply the DE10-Nano? Just in case the supplied wall-wart is misbehaving.

  • 0
    •  Analog Employees 
    on Feb 24, 2021 9:55 PM in reply to StuartADI


    Can you confirm that you are all set or not?



Reply Children
No Data