Simple FFT Frequency PLOT (NO PSD o Autospectrum)
55 views
Skip to first unread message
pdvmipv
Jun 18, 2009, 5:33:23 AM6/18/09
to IADS
Hello,
is there any way to plot the result of the FFT computation putting on
X Axis the frequencies and Y Axis the Amplitudes with an Envelope
curve ?
is there any way to plot the result of the FFT computation putting on
X Axis the frequencies and Y Axis the Amplitudes with an Envelope
curve ?
pdvmipv
Jun 18, 2009, 5:50:30 AM6/18/09
to IADS
Uploaded to the file section the file Agusta_Simple FFT Frequency
PLOT.jpg as reference.
On Jun 18, 11:33 am, pdvmipv <luigino.pagan...@agustawestland.com>
wrote:
PLOT.jpg as reference.
On Jun 18, 11:33 am, pdvmipv <luigino.pagan...@agustawestland.com>
wrote:
Mike Jones
Jun 18, 2009, 1:53:04 PM6/18/09
to IADS
You can create a Reference Envelope for the Y Access using the
Envelopes Table
The help topic for the IADS Help System is "Envelopes, Creating in
Table"
I'll summarize the process for a Frequency Plot here:
On the Dashboard, click Configuration.
Open the Data folder and click on Envelopes.
Enter the name of the X Axis parameter in the Parameter 1 field.
In the Parameter 2 field. If you are creating an envelope in a
Frequency display enter Magnitude (reference curve in a magnitude
plot), Phase (reference curve in a phase plot), or FreqPlot (reference
curve in a Frequency Plot).
Enter an Envelope Name. Envelope names should be unique for each
envelope group and envelope subgroup.
Enter the Envelope Group and Envelope Subgroup names (if applicable).
Enter the X and Y values of the envelope into the ParameterValuePairs
column. You can define a closed polygon, entering the points in a
clockwise manner. To create a closed envelope, the first and last pair
of points entered must be the same. For example: -58482 -58482 -58482
54957 549575495754957 -58482 -58482 -58482. You can also define a
line, curve, square, or a polygon with an open side. You can also
copy/paste directly from Excel columns - see instructions below.
Select an envelope Color from the Drop-down.
Enter an envelope line Width from 1-10.
Select a Line Style from the drop-down menu.
Select a Calculation Type from the drop-down menu (Load Limit
envelopes only; for reference envelopes enter None).
Select a Transparency % if desired.
Click a save option.
Create a Frequency Plot or Frequency Response Plot.
Drop the parameter into the display
On Jun 18, 2:50 am, pdvmipv <luigino.pagan...@agustawestland.com>
wrote:
> > curve ?- Hide quoted text -
>
> - Show quoted text -
Envelopes Table
The help topic for the IADS Help System is "Envelopes, Creating in
Table"
I'll summarize the process for a Frequency Plot here:
On the Dashboard, click Configuration.
Open the Data folder and click on Envelopes.
Enter the name of the X Axis parameter in the Parameter 1 field.
In the Parameter 2 field. If you are creating an envelope in a
Frequency display enter Magnitude (reference curve in a magnitude
plot), Phase (reference curve in a phase plot), or FreqPlot (reference
curve in a Frequency Plot).
Enter an Envelope Name. Envelope names should be unique for each
envelope group and envelope subgroup.
Enter the Envelope Group and Envelope Subgroup names (if applicable).
Enter the X and Y values of the envelope into the ParameterValuePairs
column. You can define a closed polygon, entering the points in a
clockwise manner. To create a closed envelope, the first and last pair
of points entered must be the same. For example: -58482 -58482 -58482
54957 549575495754957 -58482 -58482 -58482. You can also define a
line, curve, square, or a polygon with an open side. You can also
copy/paste directly from Excel columns - see instructions below.
Select an envelope Color from the Drop-down.
Enter an envelope line Width from 1-10.
Select a Line Style from the drop-down menu.
Select a Calculation Type from the drop-down menu (Load Limit
envelopes only; for reference envelopes enter None).
Select a Transparency % if desired.
Click a save option.
Create a Frequency Plot or Frequency Response Plot.
Drop the parameter into the display
On Jun 18, 2:50 am, pdvmipv <luigino.pagan...@agustawestland.com>
wrote:
>
> - Show quoted text -
John Bretz
Jun 18, 2009, 3:49:34 PM6/18/09
to IADS
HI -
Right now, the RAW FFT output cannot be displayed directly - is that
part of what you're asking?
John
Right now, the RAW FFT output cannot be displayed directly - is that
part of what you're asking?
John
pdvmipv
Jun 18, 2009, 5:45:59 PM6/18/09
to IADS
Yes it is.
I would like to display the result of the raw fft before any
computationso (so no psd or autospectrum)
Luigi
I would like to display the result of the raw fft before any
computationso (so no psd or autospectrum)
Luigi
John Bretz
Jun 18, 2009, 9:08:27 PM6/18/09
to IADS
We have had the same request a few times in the past. We can't do it
now, but if you can tell me what it is for, maybe I can try and find
another way to help you.
John
now, but if you can tell me what it is for, maybe I can try and find
another way to help you.
John
pdvmipv
Jun 19, 2009, 2:20:25 AM6/19/09
to IADS
Our specilists prefere to see in a chart just the frequency lines
contained in a signal (With optional Envelope).
With our software now we can display the fft in the following ways:
Our signal is analyzed in blocks (with or without overlap) equal to
the sample rate of the parameter producing many FFT spectrum
During the data visualization such a FFT analysis can be displayed (X
axis=Frequency, Y axis=Amplitude) as:
- One single plot containing the Average of all FFT frequencies
- One single plot for each Single FFT Spectrum computed
- One single 3D plot containing all the FFT analysis (Waterfall)
Thanks
contained in a signal (With optional Envelope).
With our software now we can display the fft in the following ways:
Our signal is analyzed in blocks (with or without overlap) equal to
the sample rate of the parameter producing many FFT spectrum
During the data visualization such a FFT analysis can be displayed (X
axis=Frequency, Y axis=Amplitude) as:
- One single plot containing the Average of all FFT frequencies
- One single plot for each Single FFT Spectrum computed
- One single 3D plot containing all the FFT analysis (Waterfall)
Thanks
John Bretz
Jun 19, 2009, 3:41:12 PM6/19/09
to IADS
Can I ask you a couple of questions about your method?
1. Do you display only the "Real" part of the FFT?
2. Does "(G)" label on your Y axis imply some scaling?
It looks to me like what you are doing is very similar to our method
after seeing the jpeg you uploaded.
John
1. Do you display only the "Real" part of the FFT?
2. Does "(G)" label on your Y axis imply some scaling?
It looks to me like what you are doing is very similar to our method
after seeing the jpeg you uploaded.
John
pdvmipv
Jun 22, 2009, 3:49:47 AM6/22/09
to IADS
I am not an expert and I am not the author of our code but I will
try to explain after looked into the code:
- We use to perform the FFT computation a C++ Library called ObjSuite
IMSL
MATH and Signal Processing for C++ (now it is not available anymore
- IMSL dropped it)
- We start from a signal SIG composed of NDAT samples and with sample
rate equal
to CAMP. For example NDAT=10000 and CAMP=2048
- The buffer to compute the FFT is equal to CAMP
- IMSL Float and COmples Vectors Definition
ImslVectorFloat V(NDAT), W(NDAT)
ImslVectorFComplex P(ndat),Q
- Load the SIG float data into the vector V
- Compute a FIRWindow of type Hanning storing the results into the W
vector
- Apply the window to the NDAT data in V
- Perform the DFT storing the results into Q obtaining NDAT/2 data
- Compute the NDAT2 Modulo values storing them in a float array as:
loop i=0 to NDAT/2
AMP[i]=abs(Q(i)) / (NDAT/2)
endloop
Then before representation we perform a process of frequency
reconstruction
using an in house algorithm that produces at the end two float arrays
containg the frequencies and relatives amplitudes reay to be
represented
on a chart. If you need details about the algo I can send the source
code.
About your second question I am not able to answer.
The G is the Mesaurement's Unit of the Parameter
try to explain after looked into the code:
- We use to perform the FFT computation a C++ Library called ObjSuite
IMSL
MATH and Signal Processing for C++ (now it is not available anymore
- IMSL dropped it)
- We start from a signal SIG composed of NDAT samples and with sample
rate equal
to CAMP. For example NDAT=10000 and CAMP=2048
- The buffer to compute the FFT is equal to CAMP
- IMSL Float and COmples Vectors Definition
ImslVectorFloat V(NDAT), W(NDAT)
ImslVectorFComplex P(ndat),Q
- Load the SIG float data into the vector V
- Compute a FIRWindow of type Hanning storing the results into the W
vector
- Apply the window to the NDAT data in V
- Perform the DFT storing the results into Q obtaining NDAT/2 data
- Compute the NDAT2 Modulo values storing them in a float array as:
loop i=0 to NDAT/2
AMP[i]=abs(Q(i)) / (NDAT/2)
endloop
Then before representation we perform a process of frequency
reconstruction
using an in house algorithm that produces at the end two float arrays
containg the frequencies and relatives amplitudes reay to be
represented
on a chart. If you need details about the algo I can send the source
code.
About your second question I am not able to answer.
The G is the Mesaurement's Unit of the Parameter
pdvmipv
Jun 22, 2009, 3:57:11 AM6/22/09
to IADS
Further info:
(x*x+y*y)^1/2 where Q= x+iy (real+imaginary)
On Jun 22, 9:49 am, pdvmipv <luigino.pagan...@agustawestland.com>
wrote:
> > John- Hide quoted text -
in my previous message i said:
Compute the NDAT2 Modulo values storing them in a float array as:
loop i=0 to NDAT/2
AMP[i]=abs(Q(i)) / (NDAT/2)
endloop
the abs methos of the class ImslVectorFComplex return the modulo of Q
Compute the NDAT2 Modulo values storing them in a float array as:
loop i=0 to NDAT/2
AMP[i]=abs(Q(i)) / (NDAT/2)
endloop
(x*x+y*y)^1/2 where Q= x+iy (real+imaginary)
On Jun 22, 9:49 am, pdvmipv <luigino.pagan...@agustawestland.com>
wrote:
James Bretz
Jun 25, 2009, 4:20:48 PM6/25/09
to IADS
> Further info:
> in my previous message i said:
>
> Compute the NDAT2 Modulo values storing them in a float array as:
> loop i=0 to NDAT/2
> AMP[i]=abs(Q(i)) / (NDAT/2)
> endloop
>
> the abs methos of the class ImslVectorFComplex return the modulo of Q
> (x*x+y*y)^1/2 where Q= x+iy (real+imaginary)
Jim
James Bretz
Jun 25, 2009, 4:24:41 PM6/25/09
to IADS
> Further info:
> in my previous message i said:
> Compute the NDAT2 Modulo values storing them in a float array as:
> loop i=0 to NDAT/2
> AMP[i]=abs(Q(i)) / (NDAT/2)
> endloop
> in my previous message i said:
> Compute the NDAT2 Modulo values storing them in a float array as:
> loop i=0 to NDAT/2
> AMP[i]=abs(Q(i)) / (NDAT/2)
> endloop
> John will be back week to comment. Sorry for the delay,
Forgot to mention. While you're waiting, you might want to go out and
check the Frequency Analysis FAQ.
http://groups.google.com/group/iads/web/iads-frequency-analysis-faq
Most likely it's at too high a level, but it might help to get some
background,
Jim
John Bretz
Jun 29, 2009, 7:12:21 PM6/29/09
to IADS
Hi again -
It looks like your math is very similar to ours. Please try a "G"
keyhit in the display so that our units are unsquared like yours, and
also maybe check the scaling options from the right-click popup menu.
We use an additional scale factor that is based on the window
(Hanning, Hamming...). Please tell us if the amplitudes look better.
John
It looks like your math is very similar to ours. Please try a "G"
keyhit in the display so that our units are unsquared like yours, and
also maybe check the scaling options from the right-click popup menu.
We use an additional scale factor that is based on the window
(Hanning, Hamming...). Please tell us if the amplitudes look better.
John
Reply all
Reply to author
Forward
0 new messages