FFTEME Program Description

                            Mike Cook (AF9Y)
                            Tel:  219-637-3399
                            Huntertown, IN

     This program was designed for experimentation with FFT processing
on weak radio signals reflected off the MOON.  For those not familiar
with ham radio moonbounce or EME (Earth-Moon-Earth) communications,
here are a few notes on the operation:

1) Signals are transmitted at 144 Mhz (or higher), normally with high
   power and large antenna arrays.

2) Reflected signals off the moon suffer degradation from several
   factors:

   a) Path loss of 252 db

   b) Background Galactic Noise

   c) Faraday Rotation due radio wave rotation in the ionosphere

   d) Libration Fading due to multipath and the irregular reflecting
      surface of the moon

   e) Doppler Frequency Shift due to the movement of the moon relative
      to the earth.

   f) Path Delay of greater than 2 seconds (but beneficial because
      echoes can be heard by rapid switching from transmit to receive)


     NOTE:  Additional details on the transmitted signal and equipment
            used are discussed in the Appendix.


     The FFTEME program performs a 1024, 2048 or 4096 point FFT on
data retrieved from a file.  The program requirements are a
coprocessor and VGA display.  The source code can be modified for use
on computers without a coprocessor by converting the Single Var to
Real Var but the arrays will have to be reduce from 4096 to 2048. This
will result in a maximum 2048 point FFT.

     The data files are actual digitized recording of transmitted
signals reflected off the moon.   As discussed in the Appendix, the
recorded signal is corrupted by many factors.


      Files Name      Description

       WWV600         AM Recording of WWV 600 hz tone used to
                      calibrate A/D card sampling rate.  Try
                      the following parameters for this and
                      other data files until you get an idea
                      on how the program operates.

                          Samples   = 4096
                          Data Skip =    3
                          Offset    =    0
                          FFT Type  =    N
                          Min Freq  =   50
                          Max Freq  = 2000

                      Note the additional 60 Hz indication in the
                      spectrum.  The maximum freq is shown in the
                      lower right corner of the screen.

       Allones        Test BPSK signal at center frequency of 1030 Hz
                      Shows the Sin(x)/(x) spectrum. Run again with
                      Data Skip of 2 and Squared FFT to get the center
                      frequency.

       KW329N         1500 Watt, 1365 Hz Carrier reflected off the
                      moon with doppler shift.  Very good receive
ÿ                     signal.

       KW329G         1500 Watt, BPSK moonbounce signal.  Use the
                      Data Skip of 2 and FFT Squared selection to see
ÿ                     the center freq.

       MED325B        60 Watt, 1365 Hz Carrier with doppler corrected.
                      This signal could not be heard during the test
                      but does show on the spectrum plot.

       LOW325L        10 Watt, 1365 Hz Carrier with doppler corrected.
                      It's there but too far below the noise.

       MED331U        60 Watt, BPSK signal doppler shifted.  FFT
                      "Squared" will not detected it.  A matched
ÿ                     correlator will detected it if the frequency
                      can be found.  A good challenge.

       KB8331G        An interesting recording of a morse code
                      transmission from a local EME station.  The
                      return moon echoes can be detected between the
ÿ                     dot and dashes.  The spectrum show both signals
                      with a doppler of 142 hz (vs a calculated
                      doppler of 144 hz).  Use a narrow min/max
                      spectrum range around the weaker echos to
                      determine freq.



                               APPENDIX


   My EME station (AF9Y) was used for these experiments.  It consists
   of the following:

   a) Six 42 foot, 22 element antennas custom designed for EME work.
      Gain of the phased array is 25.2 dBi;

   b) Up to 1500 Watts transmit power with 1 5/8 inch feedline;

   c) GaAs Fet Preamp, 0.25 db Noise Figure mounted at the antenna
      combiner;

   d) ICOM-781 Transmitter/Receiver with MMT Transverter;

   e) 386 20Mhz Computer with Co-processor and COVOX Voice Board
      (used for A/D and D/A).

   f) The station is capable of echoes off the moon with up to 16 db
      S/N in a 50 Hz bandwidth.


Encoded Test Signal

A computer generated BPSK signal is converted from digital to analog
by the COVOX board and fed to the transmitter audio.  This signal is
transmitted at 144 Mhz as an upper sideband signal for approximately
1.5 seconds.


The BPSK signal consist of the following:

       8 cycles of 1365 Hz for a "chip"

       31 chips form a maximal code sequence:
          1001101001000010101110110001111
          (1 chip = 0 Phase Shift, 0 chip = 180 deg Phase Shift

          A positive or negative sequence is used to represent
          a 1 or 0 data bit.

        7 positive data bits are transmitted during the 1.5 second tx
          period.

        The data rate is aprox 5.5 bits/sec.


The long term goal is to do both detection and match filter
correlation in real time.  Squaring the data and integrating the FFTs
should provide detection and frequency determination within a few
cycles.  The match filter can then be swept over a narrow range of
frequency and phase until correlation is achieved.

I am looking for another ham with enough of the basic equipment to
allow long transmission experiments.  Please give me a call if
interested.

                                       Mike Cook (AF9Y)