Académique Documents
Professionnel Documents
Culture Documents
The following article will attempt to explain the basic theory of the frequency response function.
This basic theory will then be used to calculate the frequency response function between two
points on a structure using an accelerometer to measure the response and a force gauge hammer
to measure the excitation.
Fundamentally a frequency response function is a mathematical representation of the relationship
between the input and the output of a system.
So for example the frequency response function between two points on a structure. It would be
possible to attach an accelerometer at a particular point and excite the structure at another point
with a force gauge instrumented hammer. Then by measuring the excitation force and the
response acceleration the resulting frequency response function would describe as a function of
frequency the relationship between those two points on the structure.
The basic formula for a frequency response function is
Where
And
And where
Frequency response functions are most commonly used for single input and single output analysis,
normally for the calculation of the
or
frequency response functions. These are used
extensively for hammer impact analysis or resonance analysis.
The
frequency response function is used in situations where the output to the system is
expected to be noisy when compared to the input.
The
frequency response function is used in situations where the input to the system is
expected to be noisy when compared to the output.
Additionally there are other possibilities, but they are outside of the scope of this article.
or
can be used for resonance analysis or hammer impact analysis.
commonly used with random excitation.
The breakdown of
is as follows,
is most
Where
And
And where
and
The breakdown of
Where
And
therefore is as follows,
And where
and
In the following example we will discuss and show the calculation of the
response function.
frequency
The excitation or input would be the force gauge instrumented hammer, as shown in Figure 1 as a
time history.
Figure 1: X(t)
In this case the response or output would be the accelerometer, as shown in Figure 2.
Figure 2: Y(t)
However as discussed earlier the frequency response function is a frequency domain analysis,
therefore the input and the output to the system must also be frequency spectra. So the force and
acceleration must be first converted into spectra.
The first part of the analysis requires the Cross Spectral Density of the input and output, this
is
. This is calculated using the response as the first input and the excitation as the second
input to the Cross Spectral Density Analysis in DATS the result is shown in Figure 3. Were
being calculated for use with
for example, then the excitation would be the first input and
the response the second input to the Cross Spectral Density Analysis in DATS.
Figure 3: Sxy(f)
Next the Auto Spectral Density of the input, or excitation signal is required. This is calculated
using the Auto Spectral Density Analysis in DATS, this analysis is sometimes known as Auto Power,
the result of which is shown in Figure 4, this is
Figure 4: Sxx(f)
The Cross Spectrum is then divided by the Auto Spectrum and the resulting frequency response
function is shown in Figure 5.
Figure 5: H1(f)
The response function would normally be shown in modulus & phase form as shown in Figure 6.
The entire analysis as used in DATS.toolbox is shown in Figure 7, the data flow from the original
input and output, force and response, can be seen through to the frequency response function.
TheDATS software does, of course, provide a single step transfer function analysis. We have
deliberately used the long-hand form below to illustrate the steps in this article.
It is necessary to understand that for the purposes of understanding and clarity in this article
some important steps have been glossed over, windowing of the input for example, to allow the
basic understanding of what makes up the frequency response function.
The following two tabs change content below.
Bio
Latest Posts
James Wren
Application Engineer & Sales Manager at Prosig
James Wren is an Application Engineer and the Sales Manager for Prosig Limited. James graduated
from Portsmouth University in 2001, with a Masters degree in Electronic Engineering. He is a
member of the Institution of Engineering and Technology. He has been involved with motorsport
from a very early age with special interest in data acquisition. James is a founder member of the
Dalmeny Racing team.
Share:
inShare5
More
Related
2.
Hi Ken,
3.
4.
Hello Eapen,
Im not sure anything can be inferred from Figure 5, it simply shows the Frequency Response
Function for this particular test.
Reply
5.
6.
Hi Kiran,
7.
8.
used to derive the cross spectra between the vibration references and the measured sound
response. These cross spectra are then used to calculate the Reference Related Auto spectra
at the response position. Each Reference Related Auto spectrum is related to the coherent
contributions from the particular references and source input.
If you would like to discuss this further please feel free to contact us directly.
Reply
9.
10.
Hi Andy,
11.
Sorry, perhaps I took my analagous situation too far. This is actually exactly the same
question as Pauls, but we have a slightly different conception of the nature of the data. I
should perhaps have chosen a better analogy, as the trigger pulses occur at 20-150Hz which
would be a pretty inhuman rate of drumstick operation.
Reply
12.
Hi Andy,
Thanks for the additional information, but I still dont understand your application and cant
comment on it unless you can explain what your trying to do. Perhaps you could contact us
directly to discuss?
Reply
13.
Ashujc
Reply
A.
So if a shaft is rotating at 100 times per second you would have a fundamental frequency
of 100Hz. If there were two blades on opposite sides of the shaft somewhere long its
length, this shaft they would be causing an excitation or noise at 200Hz because there is
two of them. You could say the noise from the blades is a 2nd order noise of the main
shaft.
The same goes for any other number. An order is the relationship to the main
fundamental frequency that occurs in multiples of the fundamental.
Reply
14.
15.
16.
Your example is very instructive. However as a new comer in FRF, I could not figure out how
you obtain the phase for FRF from the steps mentioned in figure 6. Obviously the division can
only give a real number, which is the amplitude of FRF. Thanks.
Reply
17.
18.
segments will all be zero except for the one window that contains the actual impact. Moreover
the response in the subequent windows has nothing to do with the corresponding force
window leading to potentially significant bias errors in the transfer function.
If you must use segment averaging to obtain transfer functions for tap test data then I think
you need to make it very clear that this is for multiple taps and that the window length used
must exactly correspond to each tap and response time history (i.e. there must be no
segmentation within each time history pair) and the window length must be sufficient to
capture the full decay of the vibration following the tap. Your readers should also know that
the highlighted inappropriateness of segment avergaing for tap test data cannot be overcome
by multiple taps at random intervals that only compounds the errors introducing rippling in
the estimated transfer functions because of the [Fourier transform properties of the] inherent
similarity between one tap and the next.
Reply
19.
Hello Stuart,
Thank you once again for your comments.
First of all I can only comment on the software we produce at Prosig and the methods we
would recommend.
This article is intended to give a basic understanding of the concept of what we call Hammer
Impact tests or you refer to as Tap tests.
For Hammer Impact Analysis we do not use or suggest overlapping segments, we would
indeed suggest this is an incorrect method.
So I agree with your points, it is just you have assumed we use a method we do not use or
recommend.
Our Hammer Impact software uses a Wizard to set-up the sample rates and durations to
match exactly. The data is then processed as one entire block, including a pre-trigger. A force
block and response block, having had force and response windows applied.
Thanks again for your comments.
Reply
20.
21.
Hello Umar,
Thank you for asking a question on our blog.
You question is quite straight forward.
You should have a frequency response curve from each of your accelerometers, these should
give you a basic idea of how the structure is behaving.
If you want to understand further then I would recommend the output over input method as
detailed in the article.
If you have further questions, please ask.
Reply
22.
23.
The Frequency Response Function is a measurement of motion per unit area. It is a Complex
quantity and importantly has phase. The main point for the FRF is that it is related to the input
force that is exciting the structure.
The Auto-spectrum Response is a measurement of the motion only, there is no phase
component. Additionally the Auto-spectrum Response is not related to the input force that is
exciting the structure, but only to the response of the structure.
Reply
24.
25.
Hello Krishna,
Thank you for asking a question on our blog.
A frequency response function is a measure of an output of a system in response to a given
input.
So in short no, you can not use only responses to create an FRF as detailed above. You can
however calculate the Transmissibility between several accelerometers if you define one of
them as a reference.
But it depends on your objectives and w hat your trying to achieve.
Reply
26.
where the accelerometers are located). Now, my question is: Can the resonance frequencies
of the test article be identified by examining the modulus & phase plots of the FRFs?. If yes,
then how to identify those resonance frequencies?. Thanks.
Reply
A.
Hi Nikk,
Thank you for asking a question on our blog, it is always good to talk to people who are
performing tests, at the coal face we say.
Yes your correct, if your using uni-axis accelerometers and you do as described you will
have 20 frequency response functions. I would suggest that the impact point be near or
on one of the accelerometers, but you do not have to do this.
Yes, your right from the Modulus and Phase plots of the FRFs you will be able to identify
the frequencies of the resonances.
As a rough rule a resonance will show itself as a peak in the Modulus plot and as a flip of
the angle in the Phase plot.
Reply
i.
a.
Then you can clearly see a resonance and the corresponding change in phase.
Do be aware there is no magic or easy way to do this, you have to understand
the structure your testing and have some ideas yourself before you even do any
testing, testing should validate your ideas not the other way round.
2,
Figures 5 & 6 show the same data.
Figure 5 shows the data in complex form, sometimes known as real and
imaginary form. By default in our DATS software the real part is shown as blue
and the imaginary part is shown in red.
Figure 6 shows the same data in modulus and phase form.
The signals are exactly the same just shown in different ways.
As an engineer you should consider the modulus and phase form.
Reply
27.
A.
Hello Rummy,
Thank you for asking a question on our blog.
To compute the FRF (Frequency Response Function) in your case is quite simple,
Step 1
You convert your time series signals to the frequency domain using an FFT algorithm
Step 2
You divide the response signal by the excitation signal
You then have your resultant FRF
You mention Auto Spectra and Cross Spectra, you could use the following method,
Step 1
Calculate the Cross Spectra of the Excitation and Response
Step 2
Calculate the Auto Spectra of the Excitation
Step 3
Divide the Cross Spectra by the Auto Spectra
You then have your resultant FRF
In our Prosig DATS software we have a simple function that takes a time series or a
frequency spectra excitation and response and produces the FRF for you.
Reply
28.
29.
I am new to signal processing and french speaking; so I hope that these two drawbacks will
not make my questions become too stupid.
My problem is about structures on which kind of wind tunnel test has been made. In the
case were a single excitation point is used, say A, then at each frequency w, a transfer
function from the input towards an output quantity B exists, Hab(w) .
Then the power spectral density of the ouput quantity B is said to be Sb(w)=|Hab(w)|.Sa(w)
(Sa being the power spectral density of the input quantity A)
Now what if the input is divided in four points ?
If Saiaj(w) is a cross spectral density of
input j versus input i, can I consider |Sa1a1(w) Sa2a1(w) |
an input cross spestral matrix, being that : [SIN]=|Sa1a2(w) Sa2a2(w) |
and then how can I use it | a4a4 |
to obtain the output responses ? And How ?
Thanks a lot to explain me these basic things,
David
Reply
A.
Hi David,
First of all can I ask if you are measuring forces or just responses? Also do you know if the
input signals(forces) are independent from one another? (If they are independent then the
coherence spectrum between any pair of signals will be approximately zero.) If the input
forces are indeed independent then the output will be a simple summation of the
responses to the individual inputs. However, if the inputs are not completely independent
then the predicted response is not so easy to calculate as it requires the computation of
the partial coherences between the inputs. If you can tell me a little more about your test
situation then I can give you more appropriate advice.
Adrian
Reply
30.
I am very glad to be able to write such a text as above, as I feel a bit more clever about
signal processing. Please continue telling me if what I wrote above is plenty of sense or not,
and feel free to ask anything if you feel that some questions or answers may make me
progress in that domain.
Thanks again,
David
Reply
31.
A.
Hi Jay,
Thanks for asking a question on our blog, it sounds like you have an interesting
application.
We understand the restraints your working too, things in life are never simple and never
have easy to follow guidelines.
1,
Yes, you could drop the steel ball on to the accelerometer, but I thought you said you
could not attach an accelerometer to the structure? I would have thought if you could
attach an accelerometer and drop a ball on it, you could use a very small and light force
hammer. But the accelerometer method is fine if that is all you can do, maybe there is
some practical restrictions that mean you cant perform the test exactly how you would
like to.
2,
There is no issue with the type of data, however acceleration as an input and
displacement as an output would be a Transmissibility function, not a transfer function,
transfer functions should have force in some way. In short, it is not really an issue.
Please feel free to pose further questions if you desire.
Reply
i.
a.
Hi Jay,
I would suggest the best solution for you might be OMA or Operational Modal
Analysis. Often when people are discussing hammer impact testing they
incorrectly refer to hammer impact tests as modal analysis, it is not.
In any case you could place the structure on a shaker table and then use that
shaker control signal as your input, then use the selected sensor to measure the
displacement as a number of points on the structure.
Please let us know if you have further questions at all.
Reply
32.
A.
Hi Rummy,
33.
34.
A.
displacement. The dynamic stiffness of the system is the relationship, or transfer function
between the two parameters listed above.
I would suggest that it may be best to research further in the field of structural dynamics
for further information.
Please feel free to post back what you find.
Reply
35.
followed by interlaced poles ? in terms of stability what does that mean , can you please
clarify things for me clearly.
Reply
A.
Hello Mohammed,
Thank you for posting some questions on our blog.
We will try to answer your questions as best we can.
1,
The sign doesnt make a difference and it doesnt matter where you start, the inversion
will be just that, 180 degrees.
2,
There are resonances and anti-resonances, it sounds like you have found some antiresonances.
3,
Im not sure I understand your question, it seems more like a statement. Your using a
Bode plot and have noted a phase change along with amplitude peaks. You would appear
to have found resonances.
For your final questions I would refer you to a sister article,
http://blog.prosig.com/2014/04/10/how-do-we-design-or-modify-a-system-to-avoidresonance/
This should give you some basic design pointers.
Reply
36.
3) Different response points give different amplitude values but peak frequency could be
same (Correct me if I am wrong). How do we make sure which response point gives us proper
frequency and amplitude values?
What is your suggestion?
Thank you,
Dhruvit
Reply
A.
Reply
37.
A.
Reply
38.
A.
Hi Salman,
The best solution would be check and improve the ground for the test structure and the
front end/sensor.
You should get a positive going pulse similar to that in the article in most if not all cases.
Reply
39.
A.
Hello Lee,
Our software walks the user through a dummy test and if the structure is still ringing then
a response window is applied automatically for the user as required, again the same
applies for the excitation.
Generally you would apply the window to the output or response signal and very rarely
the input or excitation signal.
2,
There are differences for sure, that is why there are different ways of making the
calculations.
The most common types are H1 and H2, but there are many, many others.
For example Prosig DATS supports;
H1, H2, H3, H4, Hs, Hv, Hc, Hr and H0.
You will probably be aware of H1 and H2, (there is no need to go into the others here)
H1(f) = Gxy(f)/Gxx(f)
H2(f) = Gyy(f)/Gyx(f)
Where Gxy(f) is the cross spectral density between the response signal y and the
reference signal x, Gyx(f) is the complex conjugate of Gxy(f), Gxx(f) is the auto spectral
density of the reference signal and Gyy(f) is the auto spectral density of the response
signal.
In almost all cases people will know of or use H1 or H2,
H1 is the classical estimator and is used in situations where the output of the system is
expected to be noisy when compared to the input.
H2 is the inverse of H1 and is used in situations where the input of the system is expected
to be noisy when compared to the output.
But when it comes to my personal preference, I like to keep it simple and when dealing
with transient signals I would use,
H(f) = Y(f)/X(f)
Where the resultant transfer function is H(f) and where Y(f) is the Fourier Transform of the
response signal y(t) and X(f) is the Fourier Transform of the reference signal x(t).
In my experience this is a clean and simple solution to a simple problem.
3,
Using the half power method on the FRF is quite common, I cant say I have compared the
base excitation to an FRF when finding damping factors, my assumption would be that
they would be similar but not the same. Do you have some experience of making
comparisons?
40.
A.
If you want to keep it simple, which I always recommend, excite one axis and measure
the same axis response, then move on to the next axis of excitation and repeat, thus
youll separately excite each axis and measure one response for each. The result will be
three FRFs.
Reply
41.
A.
Hi Jamie,
[latex]H(f) = \frac{Y(f)}{X(f)}[/latex]
In terms of actually creating the ASD and the CSD from the time series data, you would
usually use a signal processing software package such as Prosig DATS for example.
This would take care of all the mathematics for you.
If you wanted to work it out step by the step, the formulas shown above should take you
through.
Please feel free to ask if you have more questions at all.
Reply
42.
A.
The resonance will normally be shown by the largest peak in amplitude. Find that peak
and then check the phase, if the phase has inverted at this point then you have probably
found the resonant frequency.
Often your better off to view the data in a log form.
Do be aware there is no magic or easy way to do this, you have to understand the
structure your testing and have some ideas yourself before you even do any testing,
testing should validate your ideas not the other way round.
Reply
43.
44.
45.
A.