LabVIEW

cancel
Showing results for 
Search instead for 
Did you mean: 

Spike noise reduction with DAQmx

Hi all,

 

I am not a specialist for this signal processing or DAQ measurement. 

I just collected data with NI-USB-DAQ in 10 kHz sample rate.

The data, displacement information, was coming from a control unit of length measurement by laser.

However, I found some spike noises in the sinusoidal data as shown in the figure.

(I enclosed the raw data, x-axis (time in sec) and y-axis (voltage).)

 

Can I remove the irregular data with some filters in LabVIEW?

and is there some techniques to collect data with USB-DAQ? 

For example, Do I need to give some ground for other usued channels to prevent crosstalk in DAQ?

 

My final goal is to analyze FFT of the signal to find the right frequency of sinusoidal data.

 

Labmaster.

 

wave test.PNG

 

 

0 Kudos
Message 1 of 4
(1,481 Views)

Are you recording single ended or differential?  Is your equipment properly grounded?  Are your input wires shielded?  Is the shield properly grounded?

 

Sounds like more like a question for a good EE or someone who has built analog circuits (and knows about noise and shielding).

 

Bob Schor

0 Kudos
Message 2 of 4
(1,424 Views)

If your signal source is relatively low impedance you shouldn't be too concerned about cross-talk from neighboring channels, but of course if you can ground these it wont hurt.

 

As Bob mentioned, the first thing to do is to check (and if possible improve) your measurement setup integrity. However if your pop noise is really part of your signal you'll have to deal with it with signal processing.

 

If you take an FFT approach then applying a low-pass or band-pass filter will likely not help. You can try to use the Extract Single Tone Information.vi on your signal (converted to a waveform). Since your signal looks like an exponential decaying sine tone the result will not be totally accurate (but maybe good enough for your application). I would recommend to only analyze a small subset of your signal (3-10 periods) to minimize the error caused by the decaying amplitute.

 

A better approach may be to use a parametric curve fitting algorithm (returning amplitude, frequency, phase and decay time constant for your model). Pre-filtering your signal may improve accuracy though your pop noise events are relatively rare so most likely wont affect the fitting in a significant way.

0 Kudos
Message 3 of 4
(1,403 Views)

Hi labmaster,

 

Can I remove the irregular data with some filters in LabVIEW?

With a running median filter the noise is effectivly removed:

check.png

(I used Excel…)

Best regards,
GerdW

using LV2011SP1 + LV2017 (+LV2020 sometimes) on Win10+cRIO
0 Kudos
Message 4 of 4
(1,399 Views)