Controlled Envelope Single Sideband CESSB - df8oe/UHSDR GitHub Wiki

Text and figures © DD4WH, under GNU GPLv3

Controlled Envelope Single Sideband is an invention by Dave Hershberger W9GR with the aim to "allow your rig to output more average power while keeping peak envelope power PEP the same". The increase in perceived loudness can be up to 4dB without any audible increase in distortion and without making you sound "processed" (Hershberger 2014, 2016b).

The principle to achieve this is relatively simple. The process involves only audio baseband processing which can be done digitally in software without the need for modifications in the hardware or messing with the RF output of your rig.

Controlled Envelope Single Sideband can be produced using three processing blocks making up a complete CESSB system:

1.) an SSB modulator . This typically involves putting the audio input into a Hilbert transform and low pass filters producing I & Q signals (a complex SSB signal)

2.) a baseband envelope clipper . This takes the modulus of the I & Q signals (also called the magnitude), which is sqrt(I * I + Q * Q) and divides the I & Q signals by the modulus, IF the signal is larger than 1.0. If not, the signal remains untouched. After clipping, the signal is lowpass filtered with a linear phase FIR low pass filter with a stopband frequency of 3.0kHz

3.) an overshoot controller . This does something similar as the envelope clipper: Again, the modulus is calculated (but now on the basis of the current and two preceding and two subsequent samples). If the signals modulus is larger than 1 (clipping), the signals I and Q are divided by the maximum of 1 or (1.9 * signal). That means the clipping is overcompensated by 1.9 which leads to a much better suppression of the overshoots from the first two stages. Finally, the resulting signal is again lowpass-filtered with a linear phase FIR filter with stopband frequency of 3.0khz

It is important that the sample rate is high enough so that the higher frequency components of the output of the modulator, clipper and overshoot controller do not alias back into the desired signal. Also all the filters should be linear phase filters.

This CESSB system can reduce the overshoot of the SSB modulator from 61% to 1.3%, meaning about 2.5 times higher perceived SSB output power (Hershberger 2014).

Technical stuff for the implementation:

  • 24ksps sample rate (decimation-by-two) --> do not decimate further, because the bandwidth of the signal from the SSB modulator can be quite high
  • implement Weaver modulator?? --> then sample rate could be lowered to 12ksps
  • proposed signal path for UHSDR:
  1. Audio from MIC, LINE, USB
  2. decimation to 24ksps
  3. IIR bandpass (Soprano, Tenor, Bass)
  4. IIR biquad (bass, treble)
  5. FIR Hilbert transform filters to produce I & Q (101 taps should be sufficient)
  6. amplitude and phase correction for I & Q
  7. baseband envelope clipper
  8. lowpass filter linear phase
  9. overshoot control
  10. lowpass filter linear phase
  11. interpolation back to 48ksps
  12. transmit !

Thanks to Dave, W9GR for providing a pdf and hints on the method!

References:

Hershberger, D.L. (2014): Controlled Envelope Single Sideband. - QEX November/December 2014: 3-13. - HERE

Hershberger, D.L. (2016a): External Processing for Controlled Envelope Single Sideband. - QEX January/February 2016: 9-12. - HERE

Hershberger, D.L. (2016b): Understanding Controlled Envelope Single Sideband. - QST February 2016: 30-36.

Forum discussion on CESSB on the Flex-Radio forum HERE

GNU Radio processing block for CESSB HERE

WIKIPEDIA HERE