Loading ...
Sorry, an error occurred while loading the content.

Re: [SpectrumLabUsers] Q: screen capture as CSV file?, complex FFT?

Expand Messages
  • Christoph Maurer
    Hi, ... You can export the FFT, see here in general: http://www.qsl.net/dl4yhf/speclab/textexpt.htm and here in particular:
    Message 1 of 9 , May 31, 2012
    • 0 Attachment
      Hi,

      Am Donnerstag, den 31.05.2012, 03:54 +0200 schrieb ehydra:
      >
      > Hi!
      >
      > 1. I need the FFT spectrum as a captured screen file but not as
      > picture.
      > It should be a CSV file or such. Is this possible? I looked in the
      > online help but there is only doc for pictures. Say something like:
      > frequency SPACE linearAmplitude CRLF

      You can export the FFT, see here in general:

      http://www.qsl.net/dl4yhf/speclab/textexpt.htm

      and here in particular:

      http://www.qsl.net/dl4yhf/speclab/textexpt.htm#export_fft_results


      > with decimal point, not comma.
      >
      > But I can convert dB and comma if there is a problem.
      >
      > 2. My second question is: if it is possible to make a complex FFT to
      > measure impedances and getting amplitude and phase back?

      Maybe with the PAM-Function with continuous wave as modulation?

      http://www.qsl.net/dl4yhf/speclab/phase_amplitude_monitor.htm


      > Thank you -
      > Henry
      >
      > --
      > ehydra.dyndns.info

      Best Regards

      Christoph


      --
      Christoph Maurer B.Sc.

      Technische Universität Darmstadt
      Institut für Kernphysik
      AG Strahlen- und Kernphysik
      -Kosmische Plasmaphysik
      Schlossgartenstr. 9
      64289 Darmstadt

      Office: S2|14-107
      Tel.: (+49) 6151 16 4426
    • ehydra
      ... Looks like all is in one line. Hm. I will try this converting it with Perl. ... For example, if one measure a loudspeaker with help of an calibrated R.
      Message 2 of 9 , May 31, 2012
      • 0 Attachment
        wolf_dl4yhf schrieb:
        >> 1. I need the FFT spectrum as a captured screen file but not as picture.
        >> It should be a CSV file or such. Is this possible? I looked in the
        >> online help but there is only doc for pictures. Say something like:
        >> frequency SPACE linearAmplitude CRLF
        >>
        >> with decimal point, not comma.
        >>
        > Not necessarily CSV, but this may be similar to what you are looking for:
        >
        > http://www.qsl.net/dl4yhf/speclab/textexpt.htm#export_fft_results
        >

        Looks like all is in one line. Hm. I will try this converting it with Perl.


        >
        >>
        >>
        >> 2. My second question is: if it is possible to make a complex FFT to
        >> measure impedances and getting amplitude and phase back?
        >>
        > Not sure.. to measure an impedance we'd need something proportional to
        > the current fed into one input, and something proportional to the
        > voltage fed to the other. The complex FFT (i.e. FFT of a complex input
        > signal) has another use: Here, the two inputs are used to distinguish
        > 'positive' and 'negative' frequencies.
        >

        For example, if one measure a loudspeaker with help of an calibrated R.
        Interesting, it looks like that in specific cases there is no need to
        measure the phase. The amplitude vs. frequency is enough to reconstruct
        the whole circuit parameters. Even the correct phase response can be
        reconstructed.


        Thank you!



        - Henry

        --
        ehydra.dyndns.info
      • wolf_dl4yhf
        ... Yes, of course. One line per FFT. Many lines for many FFTs ;-) Cheers, Wolf .
        Message 3 of 9 , Jun 1, 2012
        • 0 Attachment
          Am 31.05.2012 20:08, schrieb ehydra:
           

          wolf_dl4yhf schrieb:
          >> 1. I need the FFT spectrum as a captured screen file but not as picture.
          >> It should be a CSV file or such. Is this possible? I looked in the
          >> online help but there is only doc for pictures. Say something like:
          >> frequency SPACE linearAmplitude CRLF
          >>
          >> with decimal point, not comma.
          >>
          > Not necessarily CSV, but this may be similar to what you are looking for:
          >
          > http://www.qsl.net/dl4yhf/speclab/textexpt.htm#export_fft_results
          >

          Looks like all is in one line. Hm. I will try this converting it with Perl.

          Yes, of course. One line per FFT. Many lines for many FFTs ;-)


          Cheers,
            Wolf .
        • ehydra
          Hi Wolf - I made the experiment: Capturing a DCF77 ferrite-antenna FFT into a file. There are two problems: 1. The FFT bin frequency is not in the list 2. The
          Message 4 of 9 , Jun 7, 2012
          • 0 Attachment
            Hi Wolf -

            I made the experiment:
            Capturing a DCF77 ferrite-antenna FFT into a file.

            There are two problems:
            1. The FFT bin frequency is not in the list
            2. The FFT bins width are not synchronized between capturing file and
            screen measurement

            Look here:
            http://ehydra.dyndns.info/NG/SpectrumLabUsers/curve-fitter/SL_FFT_bins.png

            The first diagram is the actual data file. The second diagram shows a
            'perfect antenna' (actual with a lower peak frequency because I want to
            run the fitter in LTspice thereafter). The peak at 77.5KHz in the
            captured data is completely lost in the FFT! I think because the bins
            are smaller than captured in file.

            I set:
            1. Number of exported FFT bins: 64
            2. Frequency of first FFT bin [Hz]: 50000
            3. Running EMU0202 at 192KHz

            Is there a way to make SL more compatible to LTspice input format?
            I hand-formatted the FFT capture file to look like this:
            http://ehydra.dyndns.info/NG/SpectrumLabUsers/curve-fitter/DCF77.sub (It
            is a text file)

            This is the origin capture file:
            http://ehydra.dyndns.info/NG/SpectrumLabUsers/curve-fitter/DCF77.csv


            I think the constant 0.3662 is just wrong.

            Any idea to make things simpler?

            Thanks!

            - Henry




            wolf_dl4yhf schrieb:
            > Am 31.05.2012 20:08, schrieb ehydra:
            >>
            >> wolf_dl4yhf schrieb:
            >> >> 1. I need the FFT spectrum as a captured screen file but not as
            >> picture.
            >> >> It should be a CSV file or such. Is this possible? I looked in the
            >> >> online help but there is only doc for pictures. Say something like:
            >> >> frequency SPACE linearAmplitude CRLF
            >> >>
            >> >> with decimal point, not comma.
            >> >>
            >> > Not necessarily CSV, but this may be similar to what you are looking
            >> for:
            >> >
            >> > http://www.qsl.net/dl4yhf/speclab/textexpt.htm#export_fft_results
            >> >
            >>
            >> Looks like all is in one line. Hm. I will try this converting it with
            >> Perl.
            >>
            > Yes, of course. One line per FFT. Many lines for many FFTs ;-)
            >
            >
            > Cheers,
            > Wolf .
            >

            --
            ehydra.dyndns.info
          • wolf_dl4yhf
            Hi Henry, I just made a quick check for the FFT export function, and -so far- couldn t find an error in the indicated frequency bin width. But I exported *all*
            Message 5 of 9 , Jun 7, 2012
            • 0 Attachment
              Hi Henry,

              I just made a quick check for the FFT export function, and -so far- couldn't find an error in the indicated frequency bin width. But I exported *all* frequency bins, not just a range. Anyway this doesn't affect the width of a single bin.

              Here's what I did:
              - Sample rate 11025 Hz (my E-MU is currently on duty for another project..)
              - 8192-point FFT

              Indicated on the FFT control panel:
              >
              Effect of FFT settings with fs= 11.0250 kHz:
              Width of one FFT-bin: 1.34583 Hz
              <

              Resulting FFT logfile (first line with real data, not the headline):

              2012-06-07 20:31:58.7    1.3458    0.0000    -81.6    -103.1    -79.6    -66.7

              So we've got date, time, bin width in Hertz, and frequency of the 1st bin in Hz (here: 0 Hz for the 'DC' component).

              I will try this again with the E-MU at 192 kS/sec and DCF77 as soon as time permits.


              All the best,
                Wolf .


              Am 07.06.2012 21:31, schrieb ehydra:
               

              Hi Wolf -

              I made the experiment:
              Capturing a DCF77 ferrite-antenna FFT into a file.

              There are two problems:
              1. The FFT bin frequency is not in the list
              2. The FFT bins width are not synchronized between capturing file and
              screen measurement

              Look here:
              http://ehydra.dyndns.info/NG/SpectrumLabUsers/curve-fitter/SL_FFT_bins.png

              The first diagram is the actual data file. The second diagram shows a
              'perfect antenna' (actual with a lower peak frequency because I want to
              run the fitter in LTspice thereafter). The peak at 77.5KHz in the
              captured data is completely lost in the FFT! I think because the bins
              are smaller than captured in file.

              I set:
              1. Number of exported FFT bins: 64
              2. Frequency of first FFT bin [Hz]: 50000
              3. Running EMU0202 at 192KHz

              Is there a way to make SL more compatible to LTspice input format?
              I hand-formatted the FFT capture file to look like this:
              http://ehydra.dyndns.info/NG/SpectrumLabUsers/curve-fitter/DCF77.sub (It
              is a text file)

              This is the origin capture file:
              http://ehydra.dyndns.info/NG/SpectrumLabUsers/curve-fitter/DCF77.csv

              I think the constant 0.3662 is just wrong.

              Any idea to make things simpler?

              Thanks!

              - Henry

              wolf_dl4yhf schrieb:
              > Am 31.05.2012 20:08, schrieb ehydra:
              >>
              >> wolf_dl4yhf schrieb:
              >> >> 1. I need the FFT spectrum as a captured screen file but not as
              >> picture.
              >> >> It should be a CSV file or such. Is this possible? I looked in the
              >> >> online help but there is only doc for pictures. Say something like:
              >> >> frequency SPACE linearAmplitude CRLF
              >> >>
              >> >> with decimal point, not comma.
              >> >>
              >> > Not necessarily CSV, but this may be similar to what you are looking
              >> for:
              >> >
              >> > http://www.qsl.net/dl4yhf/speclab/textexpt.htm#export_fft_results
              >> >
              >>
              >> Looks like all is in one line. Hm. I will try this converting it with
              >> Perl.
              >>
              > Yes, of course. One line per FFT. Many lines for many FFTs ;-)
              >
              >
              > Cheers,
              > Wolf .
              >

              --
              ehydra.dyndns.info


            • ehydra
              Thank you Wolf! It looks to me as a problem if the display FFT bin numbers are not equal to the exported bin numbers. Your example shows both set to the same
              Message 6 of 9 , Jun 7, 2012
              • 0 Attachment
                Thank you Wolf!

                It looks to me as a problem if the display FFT bin numbers are not equal
                to the exported bin numbers. Your example shows both set to the same value.
                If I set the first bin to zero, I get a big file without interesting
                ranges (below 50KHz).

                I tried to reduce the file size to the minimum reasonable size to reduce
                the personal work lacking experience with Perl converting the file
                format. I have to hand-calculate all the frequency values... because SL
                doesn't export this value for every magnitude value in the list.

                Hm. I will check another time.

                Good night -
                Henry



                wolf_dl4yhf schrieb:
                > Hi Henry,
                >
                > I just made a quick check for the FFT export function, and -so far-
                > couldn't find an error in the indicated frequency bin width. But I
                > exported *all* frequency bins, not just a range. Anyway this doesn't
                > affect the width of a single bin.
                >
                > Here's what I did:
                > - Sample rate 11025 Hz (my E-MU is currently on duty for another project..)
                > - 8192-point FFT
                >
                > Indicated on the FFT control panel:
                > >
                > Effect of FFT settings with fs= 11.0250 kHz:
                > Width of one FFT-bin: 1.34583 Hz
                > <
                >
                > Resulting FFT logfile (first line with real data, not the headline):
                >
                > 2012-06-07 20:31:58.7 1.3458 0.0000 -81.6 -103.1 -79.6
                > -66.7
                >
                > So we've got date, time, bin width in Hertz, and frequency of the 1st
                > bin in Hz (here: 0 Hz for the 'DC' component).
                >
                > I will try this again with the E-MU at 192 kS/sec and DCF77 as soon as
                > time permits.
                >
                >
                > All the best,
                > Wolf .
                >
                >
                > Am 07.06.2012 21:31, schrieb ehydra:
                >>
                >> Hi Wolf -
                >>
                >> I made the experiment:
                >> Capturing a DCF77 ferrite-antenna FFT into a file.
                >>
                >> There are two problems:
                >> 1. The FFT bin frequency is not in the list
                >> 2. The FFT bins width are not synchronized between capturing file and
                >> screen measurement
                >>
                >> Look here:
                >> http://ehydra.dyndns.info/NG/SpectrumLabUsers/curve-fitter/SL_FFT_bins.png
                >>
                >>
                >> The first diagram is the actual data file. The second diagram shows a
                >> 'perfect antenna' (actual with a lower peak frequency because I want to
                >> run the fitter in LTspice thereafter). The peak at 77.5KHz in the
                >> captured data is completely lost in the FFT! I think because the bins
                >> are smaller than captured in file.
                >>
                >> I set:
                >> 1. Number of exported FFT bins: 64
                >> 2. Frequency of first FFT bin [Hz]: 50000
                >> 3. Running EMU0202 at 192KHz
                >>
                >> Is there a way to make SL more compatible to LTspice input format?
                >> I hand-formatted the FFT capture file to look like this:
                >> http://ehydra.dyndns.info/NG/SpectrumLabUsers/curve-fitter/DCF77.sub (It
                >> is a text file)
                >>
                >> This is the origin capture file:
                >> http://ehydra.dyndns.info/NG/SpectrumLabUsers/curve-fitter/DCF77.csv
                >>
                >> I think the constant 0.3662 is just wrong.
                >>
                >> Any idea to make things simpler?
                >>
                >> Thanks!
                >>
                >> - Henry
                >>
                >> wolf_dl4yhf schrieb:
                >> > Am 31.05.2012 20:08, schrieb ehydra:
                >> >>
                >> >> wolf_dl4yhf schrieb:
                >> >> >> 1. I need the FFT spectrum as a captured screen file but not as
                >> >> picture.
                >> >> >> It should be a CSV file or such. Is this possible? I looked in the
                >> >> >> online help but there is only doc for pictures. Say something like:
                >> >> >> frequency SPACE linearAmplitude CRLF
                >> >> >>
                >> >> >> with decimal point, not comma.
                >> >> >>
                >> >> > Not necessarily CSV, but this may be similar to what you are looking
                >> >> for:
                >> >> >
                >> >> > http://www.qsl.net/dl4yhf/speclab/textexpt.htm#export_fft_results
                >> >> >
                >> >>
                >> >> Looks like all is in one line. Hm. I will try this converting it with
                >> >> Perl.
                >> >>
                >> > Yes, of course. One line per FFT. Many lines for many FFTs ;-)
                >> >
                >> >
                >> > Cheers,
                >> > Wolf .
                >> >
                >>
                >> --
                >> ehydra.dyndns.info
                >>
                >>
                >
                >

                --
                ehydra.dyndns.info
              • wolf_dl4yhf
                Hello Henry, I m not an expert in spreadsheets (in fact I hate most of them, especially *xcel) so I cannot say how to produce a nice diagram from the data. But
                Message 7 of 9 , Jun 9, 2012
                • 0 Attachment
                  Hello Henry,

                  I'm not an expert in spreadsheets (in fact I hate most of them, especially *xcel) so I cannot say how to produce a nice diagram from the data. But I'm sure it can be done, somehow. At least, with open office "Calc" I was able to import the FFT-logfile, and validate that the peak of DCF77 was where I expected it to be (in the file written by SL).
                  I have uploaded it as 'FFT_Export_Test_DCF77.txt' to my folder at the SL User group.

                  1024 FFT bins are exported in each line, beginning at approximately 77 kHz (that was entered in SL's FFT-export panel)
                  The first line in the file contains the headline (helpful when viewing the data with a stupid spreadsheet to see what the columns mean). Content, first part:

                  YYYY-MM-DD hh:mm:ss.s    BW_Hz    F1_Hz    dB0    dB1    dB2    dB3    dB4

                  The second line is the first 'real data':

                  2012-06-09 17:15:43.3    1.4648    76999.1082    -108.3    -115.6    -117.1    -116.7

                  1.4648 ("BW_Hz") is the bin-width in Hertz.
                  76999.1082 ("F1_Hz") is the center frequency of the 1st bin in Hertz.
                  -108.3 [dB] is the value from the first exported bin (colum title "dB0") in dB.

                  The peak of DCF77 should be near 77.5 kHz. Calculate the bin-index :

                  (77500 Hz - 76999.1082 Hz) / 1.4648 Hz = 342 (rounded; zero-based index)

                  Thus the column titled "dB342" should contain the DCF77-peak.
                  Import the stuff with open office "Calc" (remember to set the language to English, otherwise the stupid decimal *comma* would be expected in the file, instead of the decimal point), and scroll that column into view.

                  Result:
                  dB341 = -65.5
                  dB342 = -62.1
                  dB343 = -66.6

                  which is the (correct) level of DCF77, as displayed in SL ("dB over full scale", thus negative).

                  A thought: It would be more user-friendly to use the frequency as column titles. But this will inevitably cause more trouble when the bin-width is less than one Hertz. Thus the simple numbering scheme.

                  All the best,
                    Wolf .







                  Am 08.06.2012 00:30, schrieb ehydra:
                   

                  Thank you Wolf!

                  It looks to me as a problem if the display FFT bin numbers are not equal
                  to the exported bin numbers. Your example shows both set to the same value.
                  If I set the first bin to zero, I get a big file without interesting
                  ranges (below 50KHz).

                  I tried to reduce the file size to the minimum reasonable size to reduce
                  the personal work lacking experience with Perl converting the file
                  format. I have to hand-calculate all the frequency values... because SL
                  doesn't export this value for every magnitude value in the list.

                  Hm. I will check another time.

                  Good night -
                  Henry

                  wolf_dl4yhf schrieb:
                  > Hi Henry,
                  >
                  > I just made a quick check for the FFT export function, and -so far-
                  > couldn't find an error in the indicated frequency bin width. But I
                  > exported *all* frequency bins, not just a range. Anyway this doesn't
                  > affect the width of a single bin.
                  >
                  > Here's what I did:
                  > - Sample rate 11025 Hz (my E-MU is currently on duty for another project..)
                  > - 8192-point FFT
                  >
                  > Indicated on the FFT control panel:
                  > >
                  > Effect of FFT settings with fs= 11.0250 kHz:
                  > Width of one FFT-bin: 1.34583 Hz
                  > <
                  >
                  > Resulting FFT logfile (first line with real data, not the headline):
                  >
                  > 2012-06-07 20:31:58.7 1.3458 0.0000 -81.6 -103.1 -79.6
                  > -66.7
                  >
                  > So we've got date, time, bin width in Hertz, and frequency of the 1st
                  > bin in Hz (here: 0 Hz for the 'DC' component).
                  >
                  > I will try this again with the E-MU at 192 kS/sec and DCF77 as soon as
                  > time permits.
                  >
                  >
                  > All the best,
                  > Wolf .
                  >
                  >
                  > Am 07.06.2012 21:31, schrieb ehydra:
                  >>
                  >> Hi Wolf -
                  >>
                  >> I made the experiment:
                  >> Capturing a DCF77 ferrite-antenna FFT into a file.
                  >>
                  >> There are two problems:
                  >> 1. The FFT bin frequency is not in the list
                  >> 2. The FFT bins width are not synchronized between capturing file and
                  >> screen measurement
                  >>
                  >> Look here:
                  >> http://ehydra.dyndns.info/NG/SpectrumLabUsers/curve-fitter/SL_FFT_bins.png
                  >>
                  >>
                  >> The first diagram is the actual data file. The second diagram shows a
                  >> 'perfect antenna' (actual with a lower peak frequency because I want to
                  >> run the fitter in LTspice thereafter). The peak at 77.5KHz in the
                  >> captured data is completely lost in the FFT! I think because the bins
                  >> are smaller than captured in file.
                  >>
                  >> I set:
                  >> 1. Number of exported FFT bins: 64
                  >> 2. Frequency of first FFT bin [Hz]: 50000
                  >> 3. Running EMU0202 at 192KHz
                  >>
                  >> Is there a way to make SL more compatible to LTspice input format?
                  >> I hand-formatted the FFT capture file to look like this:
                  >> http://ehydra.dyndns.info/NG/SpectrumLabUsers/curve-fitter/DCF77.sub (It
                  >> is a text file)
                  >>
                  >> This is the origin capture file:
                  >> http://ehydra.dyndns.info/NG/SpectrumLabUsers/curve-fitter/DCF77.csv
                  >>
                  >> I think the constant 0.3662 is just wrong.
                  >>
                  >> Any idea to make things simpler?
                  >>
                  >> Thanks!
                  >>
                  >> - Henry
                  >>
                  >> wolf_dl4yhf schrieb:
                  >> > Am 31.05.2012 20:08, schrieb ehydra:
                  >> >>
                  >> >> wolf_dl4yhf schrieb:
                  >> >> >> 1. I need the FFT spectrum as a captured screen file but not as
                  >> >> picture.
                  >> >> >> It should be a CSV file or such. Is this possible? I looked in the
                  >> >> >> online help but there is only doc for pictures. Say something like:
                  >> >> >> frequency SPACE linearAmplitude CRLF
                  >> >> >>
                  >> >> >> with decimal point, not comma.
                  >> >> >>
                  >> >> > Not necessarily CSV, but this may be similar to what you are looking
                  >> >> for:
                  >> >> >
                  >> >> > http://www.qsl.net/dl4yhf/speclab/textexpt.htm#export_fft_results
                  >> >> >
                  >> >>
                  >> >> Looks like all is in one line. Hm. I will try this converting it with
                  >> >> Perl.
                  >> >>
                  >> > Yes, of course. One line per FFT. Many lines for many FFTs ;-)
                  >> >
                  >> >
                  >> > Cheers,
                  >> > Wolf .
                  >> >
                  >>
                  >> --
                  >> ehydra.dyndns.info
                  >>
                  >>
                  >
                  >

                  --
                  ehydra.dyndns.info


                Your message has been successfully submitted and would be delivered to recipients shortly.