Why voltage is so noisy through DAQ device although the value from multimeter is very stable?

[dhm04042]

As title said, this is my serious concern.

I'm using NI PCI X-Series connecting to BNC 2120, and I want to acquire stable voltage with this. The value through device is differ from that of multimeter(Keithely 2000).

Why is this happening even I'm using shielded cable? Don't tell me about environment, since multimeter is in same condition too and there's no noisy source around this.

The noise at device is about 10mV, and from multimeter, it's 1mV or less.

[Henrik_Volkers]

Main reason: The low pass filter in the Keithley 🙂

How does the signal from your X-Daq look like if you sample with 1kSPS for one or more periodes of your line frequency (or 1s) and look at the mean value?

The Keithley is a 6.5 digit device, what does the spec of your DAQ tells about noise and  accuracy?

[dhm04042]

Thx for the reply.

So if the reason is "physical' low pass filter, what should I do for the terminal block(bnc 2120)?

I mean, before signal is connected to terminal, everything is fine, and the noisy something is happening in the terminal block, right?

Then does it mean I have no solution unless I open the block?

I can't upload the signal right now, but one thing is clear that it can't be use since I need relatively precise result...

[Henrik_Volkers]

The Keithley 2000 is a 6.5 digit DMM with integrating ADC ..

Your DAQ (btw. what type?) has another type of ADC, so you will need to some signal processing (filter , wether in hard or software) to get similar results. Don't expect the same accuracy!

So: the joy of engineering is to do what you need with the stuff you can get.

What do you have?

What do you need?

(You want to meaure a voltage... how fast is it changing?  What is the source? ...)

Finally: You can use the Keithley to read the voltage 😉 

[paw510]

---

Then does it mean I have no solution unless I open the block?

It might be simpler, faster and more efficient to implement the filter in the software. 

The point is that at 10mV level your noise is certainly picked up somewhere and is not random (the true thermal noise will depend on the source impedance and the DAQ bandwidth but I would expect  to be in the microvolt range at most, cf. https://www.daycounter.com/Calculators/Thermal-Noise-Calculator.phtml).  And at this level you will be able to see it with an oscilloscope and determine the frequency. Most often it will be the 50/60Hz line frequency and harmonics.  If there is such a dominant noise frequency  you can filter it (together with its harmonics) wery well if you configure your DAQ to take measurements during a period (burst mode of something like that) corresponding to an integer number of cycles of the "noise" and then average them. In general I always filter the DAQ inputs averaging data during at least one 20 ms interval (50Hz period). Of course you then sacrify bandwidth but there is no free accuracy 🙂

[Dom5]

Perhaps a silly assumption, but are you sure you don't have any grounding issues?