NI USB-6210 Noise issue (Voltage analog input)

[MichaelCKW]

I recently purchased the NI-USB-6210 to replace my old voltmeter, Keithley 197A. After installation and initial testings, I found that the baseline noise was significant increased. 

For the same setting, my old Keithley 197A meter measured 5.0000+/-0.0001 V. However, the new NI-USB-6210 measured 5.00+/-0.01V. The noise is 100 times higher. 

At a lower voltage, it's 0.500+/-0.002V VS 0.5+/-0.4V. The noise is too high. According to the spec of 6210, it measures 10 V in 16 bit, and this translates to 0.1 mV. So I suspect the card may be defective.

Does anyone have the same issue? Would this be defective or could it be some settings to increase the resolution?

[johnsold]

Several things: (this would probably be better in the Multifunction DAQ Board as it is unrelated to Digital I/O)

1. The resolution of the USB-6210 on the +/-10 V range is 20 V/(2^16) = 0.305 mV.  The resolution is calculated on the full scale range.

2. The noise level is specified on the AI Absolute Accuracy Table (page 4 of the Specifications document) as 229 uV rms. That means the peak to peak noise is probably about a mV.

3. The Keithley takes three readings per second and probably does some smoothing or averaging of the voltage during that time.

4. The USB-6210 aperature time is not specified but is probably on the order of the reciprocal of the maximum sampling rate or ~ 4us. The input bandwidth is 450 kHz.  The magnitude of the noise on a signal is proportional to the square root fo the bandwidth.  The bandwidth of the 197A is not specified for DC measurements but may be as low as a few Hz. and is quite likely less than the power line frequency. That could produce a noise reduction by almost a factor of 100.

What noise do you get with the inputs to the USB-6210 shorted to ground using very short wires?

You may want to try taking multiple samples and averaging them.  This reduces the noise, assumed to be random. Sampling at 1 kHz and averaging 333 samples would give the same reading rate as the Keithley. Also consider an external low pass filter between your signal source and the USB-6210 input.  A simple RC filter may be useful, although the details depend on your signal source and reading rates.

Lynn

[SnowMule]

How are you wiring everything?  Using differential or RSE measurements? 

What range are you using?

Lots of things can affect accuracy when you're looking for a couple of LSB's. 

[Charles.K]

I'm responding to this old post. I have experienced an unexpected noise using a DC current shunt with a USB-6210 to measure low level DC voltages on the USB-6210 0 to 200 millivolt scale. I couldn't get rig of the noise without very significant filtering but that affected the dynamic response. The noise turned out to be noise generated from within the USB-6210 itself! Obviously nothing can be done about that at least realistically. That said the solution is to use a front end DC signal conditioning amplifier and this will "quietly" amplify the signal to a much higher level such as 0 to 10 VDC and then use the 10 VDC range on the USB-6210. This vastly increases the signal to noise ratio to the point where the USB-6210 noise is so insignificant in the measurement it doesn't matter. There are at least a couple of companies out there making excellent DC instrument amplifiers that will do this nicely. Hope this helps anyone.