Synchronous Sampling & FFT Windows

Signal Processing Concepts in Machine Condition Monitoring

Patrick Carle
Product Manager
1997

PUBLIC INFORMATION

Rev 5058-CO900E Copyright © 2014 Rockwell Automation, Inc. All Rights Reserved.
 Agenda

Asynchronousvs.
Asynchronous vs.Synchronous
SynchronousSampling
Sampling

FFTWindows
FFT Windows

PUBLIC INFORMATION Copyright © 2014 Rockwell Automation, Inc. All Rights Reserved. 2
 Asynchronous vs. Synchronous
Sampling

Asynchronous Sampling
The “sample rate” (how often a value is measured by the ADC) is determined by the Internal
Clock of the hardware. Also referred to as “asynchronous sampling”.

Synchronous Sampling
The “sample rate” is based on the speed of the shaft such that a consistent
number of samples are measured during each revolution, regardless of speed.

Most modern systems sample at a constant, high rate, then apply software algorithms that “decimate” and “resample”
the signal to simulate the required sample rate.
Resampling means to sample from the digital waveform at a new sample rate. Resampled points that fall between
actual sample points are typically interpolated from adjacent actual sample points.

PUBLIC INFORMATION Copyright © 2014 Rockwell Automation, Inc. All Rights Reserved. 3
 Why Synchronously Sample
Synchronous Sampling provides a high degree of confidence in the
magnitude and phase of the integer multiple order vibrations by assuring
that each integer order is centered within a bin of the FFT…

Centered in the bin means that the wave form ends

on a complete cycle.
PUBLIC INFORMATION Copyright © 2014 Rockwell Automation, Inc. All Rights Reserved. 4
 Synchronous FFT
Synchronous sample of a 60 Hz, 3600 RPM signal. 1x Peak is centered in
the bin. The signal is a pure sine wave with no noise.
Note that all of the energy of the signal is in the 1x bin – compare the 1x value (High:) with
the overall.
(60 Hz / 3600 cpm signal is centered in the 1x Order bin)

PUBLIC INFORMATION Copyright © 2014 Rockwell Automation, Inc. All Rights Reserved. 5
 Asynchronous FFT
Can produce similar results…
(plot X axis reduced to 0.1 seconds for clarity).
Asynchronous sampling will present
the same results if the signal happens
to fall centered on a spectral bin.

60 Hz signal fixed sampled at 400 Hz with a 1024 block size

PUBLIC INFORMATION Copyright © 2014 Rockwell Automation, Inc. All Rights Reserved. 6
 Asynchronous FFT
…but it isn’t likely to

If the signal falls between bins then energy will “leak” into adjacent bins and
the FFT will produce incorrect or misleading results.
Bin Leakage is when the energy of a signal “leaks” into adjacent bins.
59.5 Hz signal fixed sampled at 400 Hz with a 1024 block size

PUBLIC INFORMATION Copyright © 2014 Rockwell Automation, Inc. All Rights Reserved. 7
 Synchronous Measurements can leak to
Even when Synchronous Sampling signals can fall between bins! Such as
any:
• signals at non-synchronous frequencies
• signals at non-integer multiples of running speed
1x Centered in bin plus non-synchronous frequency at 0.97 x Running Speed.

PUBLIC INFORMATION Copyright © 2014 Rockwell Automation, Inc. All Rights Reserved. 8
 Agenda

Asynchronousvs.
Asynchronous vs.Synchronous
SynchronousSampling
Sampling

FFTWindows
FFT Windows

PUBLIC INFORMATION Copyright © 2014 Rockwell Automation, Inc. All Rights Reserved. 9
 FFT Windows
FFT Windows are algorithms that manipulate the data to reduce the effect of
side lobes (bin leakage) on the data.
An FFT Window can be applied in either the time domain, using a weighting algorithm, or in
the frequency domain using a convolution algorithm.

Rectangular, Square and

Uniform are all terms that mean
“no window”.

PUBLIC INFORMATION Copyright © 2014 Rockwell Automation, Inc. All Rights Reserved. 10
 Worst Case
The worst case, for bin leakage, is when the signal of
interest traverses spectral bins.

Signal traversing across spectral bins – no window applied

Data illustrates the
peak value of a
signal, as read
from the FFT, as it
is swept across
bins. The input
signal is at a
constant
amplitude.

PUBLIC INFORMATION Copyright © 2014 Rockwell Automation, Inc. All Rights Reserved. 11
 Effect of Windows
Amplitude

FFT Windows can alleviate the problem but different

windows provide different results.
Signal traversing across spectral bins – several windows applied
The Flat Top
Window is
clearly the best,
or is it?

PUBLIC INFORMATION Copyright © 2014 Rockwell Automation, Inc. All Rights Reserved. 12
 Effect of Windows
Shape

Each window will affect the shape of the spectral peak differently.

The shape of the peak can be Signal traversing from 59 to 61 Hz.

very important when analyzing

the spectra data especially
when multiple signals are in
close proximity or, perhaps,
crossing one another.

PUBLIC INFORMATION Copyright © 2014 Rockwell Automation, Inc. All Rights Reserved. 13
 Effect of Windows
Distinguishing Signal Frequencies

The shape of the signal affects our ability to distinguish between peaks in
close proximity, and to identify the precise frequency of individual signals.

When diagnosing spectral Signal traversing from 59 to 61 Hz.

content, usually frequency

accuracy is much more
important than magnitude
accuracy.

PUBLIC INFORMATION Copyright © 2014 Rockwell Automation, Inc. All Rights Reserved. 14
 Bin Leakage in Synchronous Data
Bin leakage is often thought a Synchronous data colliding with non-synchronous frequencies .
concern of Asynchronous Sample
data only.

However, the phenomena can occur

in Synchronous data as well - with
just as confusing results.

PUBLIC INFORMATION Copyright © 2014 Rockwell Automation, Inc. All Rights Reserved. 15
 FFT Windows
Summary

FFT Windows:
• Can be applied to both synchronous and asynchronous sampled data
• Should be applied based on the expected content of the signal
• Should be chosen based on the type of analysis required
To perform the most accurate analysis (especially of transient data), the
best solution may be to apply multiple windows to the same data to
accurately resolve the characteristics and behavior of all of the signal
components.

PUBLIC INFORMATION Copyright © 2014 Rockwell Automation, Inc. All Rights Reserved. 16
Questions?

PUBLIC INFORMATION

Follow ROKAutomation on Facebook & Twitter.

Connect with us on LinkedIn.

www.rockwellautomation.com

Rev 5058-CO900F Copyright © 2014 Rockwell Automation, Inc. All Rights Reserved.