Post Go back to editing

AD 7193, Noise Count Histogram

Hello community,

Why am i not getting anywhere close the Noise Distribution Histogramm shown on Page 15, Figure 10.
(LT 3042 Reference, Shorted Ain1& Ain2, PMOD Ad5)

my Setup:

i am using the AD7193 (placed on the PMOD AD5) board to measure a load cell.
The Load cell has maximum capacity of approx 100g and has a rating of approx 0.7mv/V.
An LT 3042 chip supplies the excitation voltage (and is wired to the Refin2+/- to provide an external reference) at 3.3 volts which result in a ratiometric measurement.
So far the setup works and has a noise of ~ +/-0.003 gramms (by averaging 1023 measurements, fully differential, which gives me an output data rate on my arduino nano of approx 1 Hz).

If i short Ain1&Ain2 with a jumper (and stop averaging) my noise is way higher than what is shown on Page 15, Figure 10 and on page 17 Figure 19.
If i understand the histogram on page 15 correctly the output rate of 2400 hz means that there is no prior averaging?
What could possible be the reason for such high noise? Since the PCB Layout is done by Digilent, i dont think that the Layout is the reason (in case of shorted out pins).

PS: I am a mechanical engineer who tries to understand the electronic world.

Thanks for your help.

  • Hi Michael,

    Sorry for the delayed response.

    Output rate of 2400hz means that there is no prior averaging. 

    Could you give more info on the setup and load cell being used? 


    Vikas J

  • Hello!
    in the meanwhile I changed my setup to better understand whats going on in my circuit.

    I disconnected the Load cell to check if my ADC without loadcell is doing well.
    I would like to get anywhere close this histogram:

    My values (with Averaging:1, chop dis, sin4, Ain1&Ain2 shorted) look like this:
    (Scale y: 0, 400, 800)

    As you can see (or maybe cannot see because of the bad upload quality) most of the values are between 0 and 400, some are even 800 counts off. So i guess i´m doing something wrong? I measure fully differential (tried pseudo aswell but the problems stays the same) My noise is way to high in comparison the the datasheet histogram.

    Am I naive to think I could reproduce the histogram of the datasheet? Is this some crazy vacuum lab setup which they show in the datasheet?

    As written I am using an Digileng Pmod AD5 with an AD7193 onboard. (Schematic)
    The digilent board has an onboard Vref "generator" ADR441 with 1.2 μV p-p noise.

    Vref (ADR441) of the ADC is supplied to Refin1,
    VDD is supplied via Arduino
    A1&A2 are tied together
    AVDD comes from an LT3042 (3.3v) - my values dont change even if i tie AVDD with VDD.

    In my final application (loadcell) I would like to conduct a Ratiometric measurement:

     i am using this libary to interface Arduino <-> Pmod via SPI

    Thanks for your help in advance 

  • Hi Michael,

    I am seeing the data between 0-800 (negative peak) while the shorted input for differential mode needs to be around midscale (1000...). I would suggest you to first verify that SPI is working fine. Say, you make some changes to the register and read back or  read a non Zero register. It would be helpful if you can share the register map with us.

    Once we are sure the device is configured correctly, then we can debug the noise related issues. 


    Vikas J

  • I made a calibration at the beginning of the programm and then subtracted the following measurements to get values around 0. If I stop doing that, my values are around midscale 
    If i stop doing this my values look like this: ~ 8388700 ..
    My communication is working my Registers with Gain 1 look like this:

    After changing gain to 128 changes to 117:

    My resulting Histogramm looks like this:

    I finally got my setup working with acceptable noise (very close to the noise in the datasheet)
    I had to edit the GIT Libary:

    0x800100 instead of 0x800000

    When conducting a Full scale calibration i had to specify the channel:

    Thanks for your help!

  • Hi Michael,

    Glad to know that it is working as per the requirement. 


    Vikas J

Reply Children
No Data