-43dB640.jpg File Description (k3pgp@juno.com (John J. Yurek), Fri 23:47)

File description -43db640.jpg
----------------------------------------

The vertical axis is a LINEAR scale that can be read as a simple
voltmeter meaning that a signal that is 50% out of the noise is at +6 dB.
 The linear scale makes it easier to see weak signals that may be only a
couple of dB out of the noise after integration.

The actual laser signal in this photo is 43 dB below the noise floor. 
Since it is difficult to establish a reference (when the signal to noise
ratio is zero dB) the following was done:

The noise output (No signal) was measured with an HP audio test set with
an audio bandpass running from 500 to 1000 Hz  with the lens capped on
the optical front end.  The audio bandpass filtering was done to prevent
high frequency noise (above the audible hearing range) from influencing
the signal to noise measurements.  This allows the dB readings on the
audio test set to more closely represent what the human ear hears.  If
the audio filter were NOT used the noise floor would be higher making the
laser signal appear considerably more than 43 dB below the noise floor!

A 630 nm signal modulated 100% with an 800 Hz  50% duty cycle square wave
(derived from a 8 MHz quartz crystal oscillator and digital divider
chain) was then fed into the optical detector and adjusted for +10 dB
signal to noise ratio.  The signal was then optically attenuated with
calibrated LIGHT attenuators until it was 43 dB below the noise floor. 
(A total of 53 dB attenuation.)

Please note that there is NO absolute power calibration (reference to 1
mw) here and 3 dB in this case means half laser power.  The only
objective of the optical attenuation was to get the signal below the
noise floor and NOT to determine absolute signal level.  (That will come
later!)  If the 3 dB optical attenuator were placed in front of a 1 mw
laser you would have 0.5 mw coming through the attenuator.

I took the picture at -43 dB because the present software has some kind
of a problem with I push it below -45 dB.  With other software I was able
to achieve a +1.9 dB SNR at -60 dB below the noise floor after 60 seconds
integration.  However the little +1.9 dB spike (although easily seen)
wasn't as impressive so I sent the one captured at -43 dB!

I am told by VE2IQ that signals like this can be solid copy using PSK and
BPSK modulation, especially in the presence of white noise and no QRM,
much as we have with laser QSOs.  (So far :-^)  I will be working on this
in the next couple of weeks after some needed parts get here and new
software is developed.

Basically all that is needed to transmit these modes is an XOR gate
inserted in series with the 50% duty cycle square wave signal so the
phase can be inverted to represent digital data.  Although it's simple to
generate these signals, recovering the signal can be quite a challenge,
especially when it's below the noise floor!  I'll post the results here
as soon as I have something working.

I don't want to waste everyone's time with graphic images since these can
be quite large.  However, if there is sufficient interest I will take
pictures of the system from time to time and post them here.  I will make
every effort to compress these images to as small a size as possible.  I
try to set 40 to 50 K bytes as an upper limit.  Unfortunately the
UUENCODE program adds quite a bit of overhead to the filesize.

John
K3PGP@juno.com