Hello,
I'm developing a high precision weighing scale based on electromagnetic force compensation. The main task is to measure a DC current with as much resolution and precision as possible.
Here you can find the detailed documentation of the project (it is in German only):
http://www.zeilhofer.co.at/wiki/doku.php?id=praezisionswaage
Schematic: http://zeilhofer.co.at/documents/waage/Schematic_V1.2.0_20140214.pdf
In the table in chapter "Emulation der Wägezelle" (http://www.zeilhofer.co.at/wiki/doku.php?id=praezisionswaage#emulation_der_waegezelle) you can find measurements made with the ADC with different offset-voltages (right column). As you can see, the noise (written in braces) increases with the DC-Voltage.
Of course you can say, why do I know, that the noise of the voltage-source doesn't increase. I am not 100% sure, but I did some research with a voltage reference supplied with an extra battery and therefore galvanic isolated, which i connected directly to the pins of the evaluation modules input pins (and of course disconnected my measurement circuit).
The reference has 1.23V and I built 2 identical circuits of them.
In the first scenario I connected one of them directly to the inputs, and measured the voltage and the noise with the ADC.
In the second scenario i connected the two references anti-serial, so that the DC Voltage cancels nearly to zero voltage, but the noise of the references sums up geometrically (factor 1.41). Connected it again tho the eval module and measured DC and noise.
Results: first case: 1.23V DC + 16uV noise; second case: 0.005V DC + 2.2uV noise.
(measurements i'm taking about here in this post were made with 250S/s and downsampled to 31.25S/s, noise was calculated over 20s of recording (about 600 samples), with linear drift removed before RMS-noise calculation).
Do you have any explanation for the results? Is this phenomenon an inherent property of delta sigma ADCs? What kind of ADC would you recommend instead? I want to get a resolution of 20.000.000 LSBs at a samplerate of about 1S/s.
Because it is for research only, the price of the ADC doesn't really matter. As I was looking for a high precision ADC, I thought I found one of the best ADCs one can get - but it seems not to be true.
kind regards,
Karl Zeilhofer