What Is CESSB?

Controlled Envelope Single Sideband is not a new on-air emission type and it is not a magic “loud” button. It is a way to control the envelope peaks that a bandwidth-limited SSB generator can create after the speech has already been compressed or limited. Done well, it lets the useful speech waveform sit closer to a declared peak-envelope-power ceiling. Done badly—or followed by a nonlinear amplifier—it can still produce distortion and unwanted emissions.

That distinction matters. A cleaner waveform is not proved by an average-power meter, and a higher average level does not guarantee a pileup win. The engineering evidence is a controlled before-and-after test at the final RF output.

From W9GR to NR0V and WDSP

David L. Hershberger, W9GR, introduced the CESSB method in his QEX paper. The problem he addressed is subtle: even if a speech waveform has already been amplitude-limited, the filters and phase response used to constrain its bandwidth and form SSB can generate new envelope overshoot.

Warren Pratt, NR0V, carried that work into practical SDR engineering through implementations, documentation and tools. His CESSB resource records that implementation lineage, and his WDSP library in the TAPR OpenHPSDR repository makes the wider DSP context inspectable. That W9GR-to-NR0V/WDSP path is an important part of the story: the paper explains the controlled-envelope method, while working SDR code places it inside a complete transmit-audio and modulation chain.

Why a Band-Limited SSB Signal Overshoots

A hard-limited audio waveform contains sharp transitions and spectral components outside the wanted speech band. A following low-pass, band-pass or SSB-generation filter removes some of those components. The recombined time-domain waveform can then have peaks higher than the limiter threshold. Phase error, including imperfections around a Hilbert-transform SSB generator, can add more envelope excursion.

Clipping those new RF or complex-baseband peaks without another bandwidth-control step creates fresh intermodulation products and spectral spreading. Filtering again can create another, usually smaller, overshoot. CESSB therefore treats peak control and bandwidth control as a coupled problem rather than pretending that one audio clipper has finished the job.

Compression, CESSB and ALC Are Different Jobs

Speech compression changes the level distribution of the microphone signal so quieter syllables are raised relative to peaks. It may include AGC, equalisation, soft limiting or other voice processing. In Hershberger’s architecture, that density-producing work happens upstream; the CESSB block has no work to do while the envelope remains below its control threshold.

CESSB overshoot control operates on the complex SSB envelope after the bandwidth-forming process has exposed the peak problem. It aims to hold the final envelope near a known limit without surrendering the required spectral boundary.

ALC is a transmitter gain-control or protection loop. Its detector, timing, control range and sample point are radio-specific. An ALC indication is not proof that CESSB is active, and CESSB does not eliminate the need to set drive correctly. If a downstream ALC loop pumps, clips or changes gain dynamically, it can alter the waveform that the DSP produced.

The dB Gain Is a Measurement, Not a Promise

At a fixed PEP ceiling, reducing occasional overshoot can allow the rest of the speech waveform to be scaled upward. In that controlled comparison, mean speech power rises and peak-to-average power ratio falls:

The comparison is meaningful only when the same speech sample, input level, upstream compression, audio and RF bandwidth, PEP reference, transmitter chain and averaging interval are used. Hershberger reported a result of about 3.8 dB for a particular test sample and implementation. That is valuable experimental evidence, not a universal 2–4 dB entitlement for every voice, radio or setting.

At low power, a measured rise in average modulation power may improve intelligibility when the path, interference and receiving conditions permit. It does not guarantee a cleaner signal, a stronger field at every instant or success in a pileup. Those outcomes also depend on propagation, antenna systems, receiver behaviour, operating technique and the other stations present.

The Linear Transmit Chain Still Matters

The controlled envelope must survive every stage after the DSP: digital scaling, DAC, modulator, driver, power amplifier, ALC path, external amplifier, filter, tuner and load. Insufficient headroom, compression, current or voltage limiting, supply droop and thermal drift can all regenerate IMD or flatten peaks.

Higher mean power at the same PEP also changes the hardware duty. The resulting temperature rise is not a fixed CESSB penalty; it depends on waveform statistics, processing level, amplifier class and efficiency, cooling, supply, load, operating duration and the ratings of each component. Watch current and temperature, respect the manufacturer’s duty-cycle limits, and leave enough headroom to keep the final chain linear.

IMD and Occupied Bandwidth Need RF Evidence

For SSB, a two-tone test is useful because its changing envelope creates predictable odd-order intermodulation products. Record the two audio tones, their spacing and relative levels, output PEP, processing and ALC settings, analyser reference plane, resolution bandwidth and the reference used for dBc. With two equal tones, a result referenced to either tone differs by 6 dB from one referenced to the combined PEP, so the reference must be stated.

A repeated speech recording is the better complement when the question is talk-power density and real occupied spectrum. Compare CESSB on and off with identical audio and final PEP. Record mean power over a declared interval, peak distribution or complementary cumulative distribution if available, occupied bandwidth, adjacent spectral power and any visible regrowth. The ITU occupied-bandwidth definition encloses the central 99% of total mean emission power; local rules may require additional masks, reference bandwidths or measurement procedures.

CESSB’s objective is simultaneous envelope and bandwidth control. It cannot certify the emissions of a complete transmitter by itself. A nonlinear stage after the processor can recreate splatter, while a narrow screenshot can hide it. Scan enough frequency range, use suitable sampling and attenuation, verify that the measurement chain is not overloading, and apply the law of the station’s jurisdiction.

Support Is Firmware-Specific

“Does this radio have CESSB?” is a manual-and-version question, not a permanent model-table question. Current official documentation explicitly names W9GR-style or controlled-envelope SSB processing in FlexRadio SmartSDR, the Elecraft K4 manuals and QRP Labs QMX firmware documentation. The OpenHPSDR lineage is documented in the WDSP repository.

That is a documentation snapshot, not a comprehensive support list. Availability, defaults and interaction with compressor controls can change with model, firmware, software build and operating mode. Check the current official manual for the exact combination in use, then verify the RF output. Silence in a manual is not evidence that an undocumented algorithm is present or absent.

A Reproducible CESSB Check

My conclusion is simple: CESSB is good engineering when it recovers headroom lost to filter-generated envelope overshoot and the resulting waveform remains linear and spectrally controlled through the complete station. The useful number is not a promised dB figure. It is the measured improvement under declared conditions.

Primary Technical References

• David L. Hershberger, W9GR: Controlled Envelope Single Sideband, QEX
• Warren Pratt, NR0V: CESSB implementation resources
• TAPR OpenHPSDR WDSP repository
• ITU-R reference definitions for peak-envelope and mean power
• ITU-R SM.1446: Definition and Measurement of Intermodulation Products
• ITU-R SM.1541: Unwanted Emissions in the Out-of-Band Domain
• ITU-R SM.328: Spectra and Bandwidth of Emissions

Questions, antenna-factor records or height trials to share? Contact RF.Guru.