             Low Power EME Data Collection Experiments

                        de Mike, AF9Y


Introduction

The recent MoonNet discussions on low power EME has renewed my
interest in development of an optimum waveform for the EME channel. 
Before the community settles on a particular waveform, we should first
determine the channel characteristics.  Important characteristics such
as coherence bandwidth, selective fading and multipath are best
determined by direct measurement with averaging over varying
conditions.   W3IWI, KA9Q and others have made astute theoretical
statements concerning the impact of these perturbations on various
modulation techniques.   I believe there is an opportunity for the
larger EME  stations to collect data on the channel characteristics
using their radio equipment and a PC equipped with sound card.


    Data Collection Approach - Phase 1, Initial Experiments

For the first phase, we need recordings with the strongest signal
possible via the EME path.  The idea  is to transmit various waveforms
tailored to allow extraction of key channel characteristics.  A one
minute transmission is sent by one EME station and the resulting
received audio is recorded at the second EME station.  Post analysis
of the recorded data should provide some interesting insight.   

VE7BQH and I ran the first attempt at collecting data on April 28,
1996 using a PC program called TX55a.  This program uses the sound
card to generate audio waveforms that modulate the transmitter in SSB
mode.  I am releasing TX55a and its derivatives as a free public
program.  The program and the data collected will be available from my
webpage at:  http://www.webcom.com/af9y.

In it's current form, TX55a provides the following:

- 1 Khz Tone  (Transmitted every odd 4 second period)

  This allows centering of the signal in the receiver passband and
  will be used as a reference signal during the test period.


- Narrow Band Spread Spectrum BPSK (Transmitted alternating even 4
  second period)

  This modulation is coded with a 63 bit Maximum Length Sequence (MLS)
  to allow extraction of channel delays beyond simple phase shifting. 
  It is useful for multipath and channel coherence analysis.  I
  consider this waveform a candidate for weak signal modulation. 
  TX55a transmits the "R" and "O" characters during the test.  The
  main lobe bandwidth is 250 Hz.


- Multi Tone FSK (Transmitted alternating even 4 second period)

  During this period, three simultaneous tones 1000 Hz, 750 Hz and 1250
  Hz are transmitted. This waveform should provide insight into the
  selective fading aspects of the channel.  Multi Tone FSK is a
  candidate for weak signal modulation and is being considered by W3IWI
  and KA9Q. The implementation of this modulation mode will likely
  require an update after W3IWI and KA9Q have had a chance to review
  it.  I may not have captured their intent.


              Data Collection Procedure

Lionel and I used the following procedure for the test:

1) Establish a 2mtr frequency just above 144.100 since this is SSB
   mode.

2) Using the program generated 1 Khz tone at the transmitter, tune the
   receiver for a 1 Khz receive signal 

3) If available, rotate polarity for maximum signal.  During our test,
   Lionel rotated his array through approximately 60 degrees during a 1
   minute TX period of the 1 Khz tone.  Using the FFTDSP42  program, the
   signal strength bar showed the best polarity position.  This was an
   interesting side test that dramatically showed the advantage of
   polarity rotation.  (I have include a picture of the results	on my
   webpage)

4) Agree to a start time and record 1 to 2 minutes of the test
   transmission from the TX55a program.  You can use FFTDSP42 to record
   the data as a WAV file.  Just make sure the "R" button is red (active)
   and hit "SAVE" at the at the end of the test period before the signal
   image scrolls off the screen.  If you are using another sound program
   for recording, I recommend you use a sampling rate of 8192 samples per
   second.


The stronger the signal, the better the analysis.  I suggest that we
strive for S/N ratios of at least 15 to 20 dB (100 Hz bandwidth). 
Each band will have different characteristics, so we need recordings
for 144, 432 and above.  If you email your recordings to me, I will
post them on my webpage.








