NI-9203 Generates Internal Noise When Acquiring Data

Issue Details

• I am experiencing noise on my signal measurements when using the NI-9203 current input module on cDAQ or cRIO. This noise appears to be generated internally, not as a result of external factors. In some cases, the noise frequency seems to be exactly equal to the sampling rate set in LabVIEW or NI MAX.
• Why does my measurements using the NI-9203 appear to be noisy?

Solution

The NI 9203 is a multiplexing card which means a single ADC is used to convert incoming signals from all channels. Each active channel line is internally connected and disconnected from the ADC controlled by the convert clock. This may cause measurable energy spikes at a rate equal to the sampling rate. However, with properly functioning hardware, this noise will be well within the NI 9203 specifications for noise and settling time. 

Verify Internal Noise

Verify the magnitude of noise by doing a zero-signal test and running the following VI. This test should be done with zero-signal. Please wire the analog input that you will be measuring on to the ground of the card in order to only read internal noise.


Note: This image is a LabVIEW snippet, which includes LabVIEW code that you can reuse in your project. To use a snippet, right-click the image, save it to your computer, and drag the file onto your LabVIEW diagram.

Reduce Internal Noise

Once the noise is verified, there are two options to reduce it.
     1. Add a shunt capacitor between 2-10 µF parallel to the input. (Prefered method, especially for low sampling
         rates.)
     2. Oversample the measurement, reduce data afterwards (25000 Hz aquisition, then a reduction to 1000 Hz)

Additional Information

For calculating the theoretical noise in nA:

1. Find the theoretical standard deviation for 9203 in its datasheet.
2. Input noise is: 1 LSBrms RMS and 7 LSB for Peak-to-Peak (meaning any two points should be within 7 LSB's and on average the noise is within 1 LSBrms).
3. To calculate noise (with a 99% confidence level i.e. 3 sigma) we use the formula: Noise = ((2*Range)/(2^Bit))*LSBrms*3.

Ex: The theoretical noise of the NI-9203 in bipolar mode will be around 2000 nA. 


To avoid noise introduced by multiplexing altogether, use a simultaneous sampling card like the NI 9227.