j'utilise une carte d'acquisition ADC USB 6229, 16 bits.
j'utilise les deux premières voies de cette carte pour lire un signal en différentiel. Quand je court circuite ces 2 entrées et que je les connecte à la masse du chassis de la carte j'obtiens un bruit résiduel de 160 uV pp (bruit ~blanc). J'ai configuré la carte pour une pleine échelle de +/- 200 mV. En théorie je m'attends a obtenir un bruit qui oscille entre +/- 6 uV (j'ai considéré 15 bits).
Je peux me tromper d'un facteur 2 ou 4 mais pas d'un facteur 20...
Voyez vous où est la faille?
Merci d'avance pour votre aide
Solved! Go to Solution.
Can you share the VI and settings you used to gather this data?
You have to connect AI to AI GND (as short a possible) and not Chassis GND (I don't know if that even exist for USB devices)
I have updated my tests.
I have shortciruited input a0 and a8. I use the "differential" mode.
I set the fullscale either to 10V and to 200 mV.
I use the NIMAX to display the data you see in the attached file.
I use another script to get the standard deviation of the measurement of 50000 samples at 50 kHz. (This is the line SEPD in the DOS cmd window)
for the 10V Fullscale I got a std dev of , that is dV = 2.2413e-4 V, or dV/V = 22.4 ppm
for the 200mV Fullscale I got a std dev of , that is dV = 1.4532e-5 V or dV/V = 70 ppm
assuming 16 bits I expect the noise to oscillate between 0 and 1 LSB. I consider a LSB equals to 2^-15, that means a dV/V = 30 ppm .
I would appreciate to have also 30 ppm when I use the full scale at 200 mV
Could you tell whether alll this is making sense?
Many thanks for your help
Is your calculation RMS or peak?
As per specifications, you can expect SD 13uVrms noise in the 200mV range and you're measuring SD 14.5uV - which is almost the same
Similarly, you can expect SD 244uVrms in the 10V range and you're measuring SD 224uV - within the limits
I would say your measurements are well within the specifications of the instrument - just your had different expectation.
BTW, you need to short circuit AI0, AI8 and then AIGND altogether, you're effectively nulling the differential signal but not the common mode.
Indeed it fits the spec.
I have short circuited a0/8 and Gnd
the Std Dev is rms.
Still I am wondering why the spec are less with the low full scale (200 mV), is this connected to the amplifier noise?
Thanks again for your feedabck
Yes, the system noise would be dependent on a lot of factors and a critical one would be the amplifiers and attenuators used in the Analog front-end used for the ADC.
I would also assume that at the 200mV the input signal is attenuated (noise also attenuated) whereas at other high ranges, along with the signal, the noise is also amplified to an extent.
what is strange to me is why NI did not size their amplifier (mostly required at 200 mV, to extend 200 mV up to 10 V) to get the associated noise still compliant with 16 bits.
Should not be a big deal.
I would be interested to get what others manufacturer deal with that issue