AD7684 issue (strange behavior)

Hi,

We are trying to modify the input stage of an already manufactured board by replacing AD7793 by AD7684 in order to measure up to 1KHz signals. I have attached a simplified schema of our input stage and two waveforms we have obtained from the AD7684 output.

We are only getting good performance when we connect our oscilloscope to IN+ and IN-, furthermore, we get good results when connecting just the GND of the oscilloscope wherever IN+ or IN-.

We need to measure in both scenarios IN+ > IN- and IN- > IN+, that is why we selected this AD7684 truly differential instead of AD7693

Could you provide some solutions, similar experience... any hint could help!

Best regards and thank you in advance,

attachments.zip
  • WintonM,

      A number of things I would consider, but first

    -- On the two files, what are x and y axis units??  Counts vs. seconds? and Volts? vs. can't read?

    The sigma delta AD7793 (or any S-D ADC) does a digital filter with a lot of averaging.

    SAR types (AD7684) has a nasty dynamic input for IN+, IN- and REF due to the switched caps

    See fig. 20 on d.s.

    -- Try a 1nF cap between the inputs and a 1nF on each input to ground with a DC signal.

       We'll figure out the 1kHz problem later.

    --  What is the min/max current through the one ohm shunt??

    -- What is the voltage top and bottom on the shunt with respect to the same ground that the AD7684 sees?

    Some ADCs that have "differential inputs" require that the voltage swings be symetrical about some voltage.

    E.G. footnote 1, page 3 and table 7, page 7 of AD7684 d.s.

    Instead of 1.2V, you may need 1.25V.

    One meg is high impedance.  Try changing all four resistors to 10k-20k??  This really sounds

    like you are getting noise from a local radio station, etc. when you add a three foot "antenna".

    Or you have a multiple ground or ground loop problem.

    "We are only getting good performance when we connect our oscilloscope to IN+ and IN-,"

    Does this mean two scope probes on IN+ and IN- on your schematic, i.e. across the sense resistor?

    Where were the ground leads on the scope probes?

    PC board??  How many layers?

    Bypass caps on the ADC??  What size? How close??

    Harry

  • Hi Harry,

    Thank you for your reply.

    On the "AD7684 issue.pdf" the X axis is time in miliseconds and Y axis are volts.

    On the 1 Ohm shunt we have to measure currents around mA (ie. 0.80mA or 2.4mA)...

    >>One meg is high impedance.  Try changing all four resistors to 10k-20k??

    We will try that, thank you. I will let you know.


    >>"We are only getting good performance when we connect our oscilloscope to IN+ and IN-,"

    >>Does this mean two scope probes on IN+ and IN- on your schematic, i.e. across the sense resistor?

    That is rigth, when connecting the oscilloscope probe ground to any of them IN+ or IN- the measurements are fine.

    Do you think that replacing with the AD7683 pseudo-differential we could solve this problem?

    Thank you again, regards,

  • WintonM,

    Earlier, I said:

     

    "See fig. 20 on d.s.

    -- Try a 1nF cap between the inputs and a 1nF on each input to ground with a DC signal.

       We'll figure out the 1kHz problem later."

    "-- What is the voltage top and bottom on the shunt with respect to the same ground that the AD7684 sees?"

     

    "PC board??  How many layers?"

    "Bypass caps on the ADC??  What size? How close??"

    Harry

  • 0
    •  Analog Employees 
    on Aug 2, 2018 4:43 PM
    This question has been assumed as answered either offline via email or with a multi-part answer. This question has now been closed out. If you have an inquiry related to this topic please post a new question in the applicable product forum.

    Thank you,
    EZ Admin