Everyone who uses communications frequencies is aware of the problems of channel crowding and interference. In many parts of the United States, repeater coordinating councils have declared the 2-meter Amateur Radio sub bands "full." The CB bands are hopelessly cluttered, and commercial users of VHF/UHF frequencies are searching for a solution to channel overcrowding in the land mobile bands. Obviously, better spectrum management, re-use of existing channels, or new spectrum saving techniques will be necessary if these problems are to be solved.

   One spectrum saving technique just beginning to be used in commercial communications is ACSSB or Amplitude Compandored Single Sideband. This system consists of a compressor at the transmit end of a communication path which brings weaker syllables up nearer to the levels of the stronger syllables.  At the receiver an expander is used to  restore the original dynamics which also pushes down any noise or interference picked up between the transmitter and receiver.

Unfortunately, present commercial implementations of this technology are quite expensive by Amateur standards (over $1200) so that I do not know of anyone using an ACSSB system on Amateur Bands.

   ACSSB systems provide the following:

• Up to 4:1 compression of speech during transmission; allows up to 40 dB of  speech dynamics to ride above 10 dB S/N.

 
• Up to 1:4 expansion during reception;  expands 10 dB incoming dynamics and S/N to restore 40 dB of dynamics (and, of course, effective S/N)

 
• Pilot tone can be used for a reference for AGC control: handles up to 50 Hz fading.

 
• Pilot tone can be used for automatic frequency control (phase locked) and Doppler correction.

 
• Pilot tone can be used to provide reliable carrier squelch operation.

Figure 1a shows time versus amplitude of the word "JUG".  Typically in European languages the dynamic range of speech is about 40 dB. 

Figure 1b shows the same signal compressed.  The scale shows that the original dynamic range has been compressed 2:1 from 50 dB to about 25 dB of dynamics.   A received signal that is at least 25 dB above the noise will have most of the speech at or above the noise level but any background noise at the transmitter end of the circuit will also be heard 25 dB stronger.

 In the case shown, however, S/N is only about 16 dB above receiver or rf environment noise so that some speech is buried.  With no compression, however, a large portion of the speech would have been buried below the noise.  This is why you can often hear a weak station but can't understand them since the intelligence is in the weaker consonants (e.g. "J" and "G") but most of the power is in the vowels (e.g. "U").

Figure 1c shows that even at 16 dB signal-to-noise (S/N), most of the speech will still be at or above the noise level.  When expanded at a 1:2 ratio, the dynamics of the original speech are restored and the noise, in effect, is reduced significantly. 

The S/N isn't really improved, but it sounds like less noise because of the restoration of the dynamics of the speech.  When no one is talking, the noise really does drop down to -32 dB, which is a definite improvement.  Yet at peak speech, noise is still only 16 dB below the desired signal and even less on weaker syllables. 

Those having Real Audio capability can click on the Real Audio Demonstration to hear just how effective this technique can be.  The signals are set to a raw S/N of 10 dB without processing, 10 dB with compandoring, then 16 dB with compandoring for this demonstration.  If you don't have Real Audio, Download it for free.

As I find time, I will upload more of this article with a schematic for an experimental ACSSB adapter.  If there is sufficient interest, I can make PC Boards available at a later date.