Definition: Aliasing 3

[exploengineer]

Hi All,

I know what aliasing is and many times I need to explain it to a layman or engineer for that matter that doesn't understand vibration, etc.

Does anyone have a good explanation in layman's terms for aliasing?

Thanks in advance for your help!


Frank Lucca M.I.Exp.E.

http://www.terradinamica.com

 

[hacksaw]

it is a problem that arises when the data is collected at regular intervals and it is converted into amplitude(and phase) format as a function of frequency. it shows up as spurious frequency content when the rate of sampling is comparable to the rate that the data itself is changing.

 

[IRstuff]

Aliasing is the consequence of inadequate sampling. Nyquist theorized that a sampling 2 times the highest frequency of a signal was the minimum required to accurately recover all frequencies.

When sampled below the Nyquist frequency, a sampled signal will exhibit "aliased" frequencies that were ambiguously recovered as lower values than they were in the original signal.

Mathematically, sampling results in multiple copies of the signal spectrum in frequency spaced at the sampling frequency. If spectra overlap, as with undersampled signals, the higher frequency content is "folded over" into a lower frequency range, thereby causing "aliasing"

TTFN

 

[electricpete]

Since you asked for example to a layman, I like this approach:

Draw a sinusoid, then look at the location of samples for:
5 samples per period of the sinusoid
2.1 samples per period of the sinusoid
1.9 samples per period of the sinusoid

At 2.1 samples per period if you connect the dots you will see the original sinusoid.
At 1.9 samples per period the distance between samples is more than 1/2 a period. Connect the dots of your samples and it forms a sinusoid at lower frequency than the one you were sampling. That is aliasing.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[byrdj]

As Electricpete stated, drawing the sine wave and then putting points on the line is how I explain to non engineering managers about data sampling rates.

 

[MachineryWatch]

Or, you could use the video/film analogy where video captured at 30 frames per second of a spoked wagon wheel or car wheel while it is turning is not capturing enough samples to truly reproduce the motion. When the video or film is played it looks like the wheel is turning backwards because the inadequate number of frames per second (sample rate) has created an alias! It was particularly obvious in the old westerns where they showed the various horse drawn wagons and "Stage Coaches".

Skip Hartman

http://www.machinerywatch.com

 

[exploengineer]

Electricpete and MachineryWatch,

Thanks guys, that's what I was looking for. Sometimes its hard not to be an engineer. When we are speaking to layman, this is especially true.

In rock blasting, if we set up monitoring stations to close to a blast, aliasing occurs and screws up all the data. I'm continually trying to explain this to other engineers and regulators that do not understand much about vibration.

Best regards,

Frank Lucca M.I.Exp.E.

http://www.terradinamica.com

 

[electricpete]

In the context of rock blast, I can think of another explanation. I am not sure if this descirbes your situation.

Imagine that the analog time waveform consists of very short impacts with lots of lower-magnitude ringing-down or “dead” time between impacts.

http://reliability-magazine.com/pub/CARS12TWF.jpg

The sharp rise time of the peaks represents high frequency content. Therefore we need high sample rate.
The picture here has high sample rate and shows the peaks.

If the sample rate was lower you would likely not sample on the peak. The peak won’t look as high in relation to the other stuff. The character of the waveform changes.


=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[IRstuff]

To expand on EP comments, the fact that aliasing only occurs when the sensors are close to the blast is consistent with the notion that there is higher frequency content in the blast transient.

It is harder to propagatehigh frequencies in air, so the farther you are from the blast, the fewer high frequencies encountered

TTFN