[CW] CW compared with SSB

[pa0wv]

The "detection of weak CW signals in noise" on eham.net lcwo.net and 
CW-archives (here) resulted in the enormous reply of 4 answers.  The 
download on my website however was in the hundreds; reactions on the CW 
files are still very welcome.


The result was up to now:


Average  percentage errors  of the   4 participants
[code]
-10 dB SSB filter errors    0,75%
         after CW filter     2,50%
-11 dB SSB filter           12,5%
         na CW filter        16,25 %
-12 dB SSB filter           32,5%
         na CW filter        21,25 %
-13 dB SSB filter           65%
         na CW filter        52.5 %
[/code]

So CW with PEP during key down of -10 dB below white noise with 
bandwidth 300-2800 Hz is sufficient decodable. When you narrow the 
bandwidth to 150 Hz the S/N is 12.2 dB better BUT the reliability of 
decoding is hardly different.

That is obviously due to the human brain that is able to listen 
selective to a tone present or not.

Now the question is: what is the difference with SSB

Weak SSB you have to detect with spelling words: alfa bravo charley ...
It turns out when you spell a thru z with minimal spacing it takes 17,7 
seconds
When you sent in Morse code a thru z at 19 wpm it takes the same time.

The spelling words are per word adjusted to all having the same PEP 
value.
Random sequences of 10 characters are generated in white noise, band 
limited to 300-2800 Hz

As far as I can see the difference with CW is about 14 dB so a PEP 
difference at the antenna of the transmitter of a factor 25 makes it 
equally decodable.

Please try to decode the files
http://pa0wv.home.xs4all.nl/eham/SSB6.WAV  ( 6.0 S/N ratio)

http://pa0wv.home.xs4all.nl/eham/SSB5.WAV   ( 5.0 S/N ratio)

down to

http://pa0wv.home.xs4all.nl/eham/SSB0.WAV  (0.0 dB S/N ratio)


Each file contains 10 random spelled out characters.
Your result can be sent to my email pa0wv at amsat.org

73 Wim PA0WV