Line Frequency Noise

[Gurdas]

Hi!

I am acquiring data from a digital readout unit using AI on my PCI-6034 card (with CB-68LP).
This signal is VDC (±10mV)

Lets say the output from the digital readout unit is solid 3mV. When I display the acquired signal, I see a random noise signal varying a few mV either side of 3mV (instead of a straight line at 3mV).

What could be the cause?
How can I find out if this noise is line induced voltage noise and if yes, how can I remove it?
Do I need to use averaging/filtering?

My system is Win2000, LV-PDS 6.1 and NI-DAQ 6.9.2

Thanks!,
Gurdas
gurdas@qagetech.com

[LocalDSP]

There can be many reasons for seing what you see. There could be noise issues in the signal you are acquiring (bad shielding or common mode noise) or it could be quantization noise generated by your PCI-6034E board. If your input range is set to +/- 10V, the resolution is only 0.3 mV. Make sure your input range is set to +/- 100 mV.

You can analyze your signal using the many measurement functions in your LabVIEW Analyze palette. Use for example the Power Spectrum.vi to see the actual frequency distribution of your "noise" (check the dB on parameter). Is the spectrum pretty much flat, then you have a noise issue. Do you see peaks around 50 or 60 Hz, then you have a Line Frequency noise issue.

If you are using the Basic DC-RMS.vi, the DC value of your signa
l will be the result of an averaging on your entire record. Use for example a Hanning window if you want to reduce the problems caused by line frequency.

You can also use the Extract Single Tone Information to look for specific spurious tones in your signal.

Once you have identified the source of your noise, you can either filter it out (Filter.vi) if there are tones or obvious high-frequency noise issues, or experiment with the DC-RMS averaging types and windows until you have optimized your measurement.

But always remember that the best place to remove a problem is at its source, so eventually improve your acquisition conditions (differential instead of single ended, shield cabling, pre-conditionning etc...)

[Gurdas]

LD (don't know how to address you ... LD sounds better than LocalDSP!!),

Thanks for your answer. I drifted from this issue due to other pressing engagements. Re-visited it last week.

The problem persists. We tried some software tricks but the results have not been as good as we'd like.

The possiblity of the card itself being faulty arises. We tested signal integrity in MAX using numeric channels 21210 and 21290. I have attached screen shots of what we observed.

We see is noise of the order of 2mV instead of a smooth 0V (in case of 21210) or 5V (in case of 21290).

What could be causing this?

Looking forward to your reply,
Gurdas
http://www.qagetech.com

[LocalDSP]

Gurdas,

On your screen shots I can see that your input range is set to +/- 10V corresponding, as mentioned in my original answer, to a quantization resolution of 0.3 mV (you can clearly see the different discrete values spaced by 0.3 mV). But since your expected signal is as small as 3 mV, you should reduce your input range to +/- 0.1V. This will increase your resolution by a factor of 100 and (hopefully) improve your dynamic range enough to perform reliable measurements.

Again, before using any software tricks to extract information out of your acquired signal, you should focus on optimizing the acquisition itself by optimizing the dynamic range, reducing the bandwidth (if possible) and eventually look at your signal connections (differe
ntial or not, shielded, twisted pair etc...).

Then use some of the described processing tools to optimize your measurement and eventually identify the source of your problem.

[Gurdas]

LD,

Thanks. I tried what you suggested. The resulting screen shots are attached to this post.
As in the previous case, I have used 21210 numeric channel in MAX.

There is improvement but honestly I am not clear what is happening here. The signal still looks 'noisy' although the peak to peak levels have decreased.

Q) Shouldn't I see a solid zero line?

Also, there is NO difference (for the input ranges ±10V, ±1V, ±0.1V) when I see data in my VI. I've used AI config, AI start and AI read.

Feeling somewhat lost .....

Best regards,
Gurdas

[LocalDSP]

The noise seems to have decreased considerably so you are on the right track. There will always be some amount of noise remaining in your signal (and actually this is good since it helps averaging your dc result correctly. This is called dithering).

Your VI should of course give you the same results as MAX. Make sure you enter your input range correctly in your AI Config. Use some of the DAQ examples if needed.