cRIO 9205 Analog Reading Vary 100 mV with voltage divider

[russellb]

Hello,

I am using the NI cRIO 9205. The analog input reading of a 8V DC power supply varies by 100mV when a voltage divider is in front of the analog input channel. Some of the channels do not have voltage dividers those channels reading are more stable.  Attached is a drawing of the voltage divider, source and the cRIO module input.

We are just using +8V for testing. The actual signal could be -28 to +28 V that is why we have the voltage divider.  The reason the voltage divider resistance is 1Mohms is because the actual source impedance is 10 Kohms. 

I upgraded my cRIO version as the following link suggested, but it did not improve my results:

http://digital.ni.com/public.nsf/allkb/364B8C91E3D410B086257229007B4783

Please let me know if you have suggestions,

Regards,

Russell

[russellb]

Here is the attachment of the drawing.

[Nick_CH]

Hello Russell,

I just need more precision about your question. The 100mV is an offset or noise (varies continously of several mV)?

Best regards,
Nick_CH

[russellb]

The 100 mV is more like noise.   The analog reading varies by 100 mV depending on when the measurement is taken.

Thank you for your help,

Russell

[Nick_CH]

Dear Russell,

Are you sure that it is not an issue of the calibration values. Do you take in consideration the "LSB weight" and the "Offset" values.

You can have a look in the example: Toolkits and Modules -> FPGA -> CompactRio -> Module Specific -> Analog Input -> NI 9205 Basic IO.lvproj

Best regards,
Nicolas

[rcetrulo]

Dear Russel,

The 9205 has an input multiplexer, which is used to scan across the channels. Every time it moves to a new channel, like every other multiplexed device, it will dump some charge on the input terminal. Increasing the source impedance by using a voltage divider will increase the noise due to this phenomenon, called charge injection. Can you please send a scope screenshot of the signal after the voltage divider? Make sure the module is acquiring data when you take the measurement.

You could try reducing the values of your resistors at the expense of some error due to your source impedance, or add a cap across your low side resistor on the divider (in the order of 1nF) in order to reduce the voltage error due to charge injection, at the expense of reducing your input bandwidth.

Let me know what you see.

RC

[Camell]

As I came here via Google, the solution  can probably be found here:

http://digital.ni.com/public.nsf/allkb/CF83426BC3AC514A86256C10005A4771 

[christian_w]

If you use the module in differential mode check if you have also connected the COM terminal. Without a connected module ground I saw what you see.

Hope it helps

Christian