Sign up Calendar Latest Topics Donate

  Author   Comment  

Posts: 57
Reply with quote  #1 
I've come across the term 'picket fencing' in the last couple of days in relation to spectral data and frequencies which fall off a bin.

I've attached the article, refer to page 2 of the pdf (p.19 of the article).

Has anyone used this term before? I'm assuming that is why the 'Z' key is used in MHM?

Also I found the article quite useful, it highlights the usefulness of the PeakVue Time Waveform and relying less on the spectrum for diagnosing bearing faults.

Attached Files
pdf Pitfalls in the Analysis of Mach Vibration.pdf (173.14 KB, 55 views)

Walt Strong

Sr. Member
Posts: 889
Reply with quote  #2 
If you want to know more about it:

Search: picket fence signal analysis

This term and many others related to signal analysis have been around for my 50-year career. I am glad you showed curiosity, interest, and initiative that seams to be in short supply these days![smile]



Sr. Member
Posts: 647
Reply with quote  #3 

I'm assuming that is why the 'Z' key is used in MHM?

I'd say yes.  Using the Z key invokes frequency interpolation algorithm to overcome the discrete frequency step limitations which some people call picket fence effect. 

The meaining of picket fence effect from is as follows:

Picket Fence Effect


The FFT spectrum is a discrete spectrum, containing information only at the specific frequencies that are decided upon by setting the FFT analyzer analysis parameters. The true spectrum of the signal being analyzed may have peaks at frequencies between the lines of the FFT spectrum, and the peaks in the FFT spectrum will not be at exactly the correct frequencies. This is called Resolution Bias Error, or the Picket Fence Effect. The name arises because looking at an FFT spectrum is something like looking at a mountain range through a picket fence.

By a process of interpolation, it is possible to increase the apparent resolution and amplitude accuracy of the FFT spectrum by a factor of ten.

Note the last piece, by a process of interpolation it is possible to increase the apparent frequency resolution… by a factor of 10.   Personally I would use the phrase "effective resolution" instead "apparent resolution". In other words, we are generally getting a MUCH better estimate of the true frequency of a peak by applying frequency interpolation rather than using the bin center.   And pressing Z does that. 

Summary: Pressing Z applies that frequency interpolation algorithm to give you a better estimate of the frequency of the peak.

I would say this aspect is widely misunderstood or underappreciated in the vibration community.    My own personal journey in trying to figure out and explore this mysterious thing is recorded in this meandering thread (I'll see if I can dig up some of the spreadsheets).  It really impressed me to see how much accuracy in estimating a peak frequency is gained with frequency interpolation. If you are trying to get the best info from previously collected data to help what a particular peak really represents, it can really help to have the best frequency estimate available (using Z).  Of course it cannot overcome the limitations of the original bin width selection set up when the data was collected.  In other words it won't help use discriminate two close-together peaks whose actual spacing is a small number of bin widths.  (that requires using more lines and/or  lower Fmax during data collection to increase resolution / reduce the bin width ...which makes those peaks a larger number of bin widths apart).

Previous Topic | Next Topic

Quick Navigation:

Easily create a Forum Website with Website Toolbox.