9205 in differential measurment

[nicolasb]

Hi,

I'm trying to read some radiation sensors : PAR (https://www.campbellsci.fr/skp215) and CNR4 (http://www.kippzonen.com/Product/85/CNR4-Net-Radiometer#.WHT2Ru3wnRZ) with NI 9205 on CRIO.

I'm doing a differential measurment but I have some trouble with precision. There is about 300µV noise, which is huge for this kind of measurment (I'd like to have at least around 20µV of precision, which I'm suppose to have in range ±200mV).

Both sensors have a ground reference.

I tried all kind of connection I found for differential measurment without success :

- sensor grounded and COM of NI 9205 grounded

- sensor grounded (or not) and 1MOmh resistor between "sensor -" and COM

- sensor grounded (or not) and 50kOmh resistor between both "sensor -" and COM, "sensor +" and COM

For now the only conclusion I have is that the NI 9205 is not 16 bits precision, or the range I choose is not respected.

Is someone did the same kind of measurment with Ni9205, and could help me with the connection ?

Thanks,

Nicolas

[nicolasb]

Nobody ?

To sum up : Is that possible to get around 10 µV precision with NI 9205 and how ?

Thanks,

Nicolas

[AndyClegg]

Hi,

Have you confirmed that the range is actually set correctly - +/-200mV should give about 10uV noise (per the datasheet) ?

Of course that noise figure is just related to the 9205, and does not include the sensor itself or interference that might be arising due to wiring running through electrically noisy areas. Perhaps try attaching a well conditioned reference voltage to the input, in an electrically clean area, to see if the noise goes down to the levels in the datasheet - you could have a faulty 9205 (rare)

Noise can also arise from mistakes / inappropriate programming - e.g. if you are multiplexing between different channels too fast and not letting the ADC settle - what sample rates are you using ? Are you using the Scan Interface to read into LV RT or are you using FPGA direct. If sample rates you are using are relatively slow then this should not be an issue. But I would look at your software side of things since it is very easy to get corruption of your signals that look like noise but are in fact due to ailaising / sub-sampling / type conversion problems etc etc.

Finally if you do have noise that is real - perhaps due to the environment the sensor is in - then you can look at applying filtering in the software to reduce that (at the expense of dynamic response).

Hope this helps.

Andy Clegg

[nicolasb]

Hi, thanks you for your interest,

I'm using Scan mode, at 10ms.

I tried to let 1s between each measurment without success.

For information I tried the same measurment with a 9219 and it's working fine, and also with a Campbell CR10.

At the end, I will multiplex the sensors myself and do measurment with a channel of the 9219.

Nico