Douglas T. Smith Editorial Services

m On the Occupied Bandwidth of CW Emissions
Site Contents
- Home
- Editorial Services
- Book and articles by Doug Smith
- Downloads
- About Doug Smith
- Book and article errata
- Links
- Photo Albums

On the Occupied Bandwidth of CW Emissions
2003, Douglas T. Smith Editorial Services

Where the ARRL Handbook and some manufacturers went wrong.


Much ado has been made lately about the occupied bandwidth of SSB phone emissions. 3 kHz or so of bandwidth, accompanied by typical levels of IMD products, has been the acceptable norm for a long time now. Much more than that on a crowded band tends to raise objections. What is the acceptable norm for CW emissions?

Part 97 of the FCC rules states that emissions " shall not occupy more bandwidth than necessary for the information rate and emission type being transmitted, in accordance with good amateur practice." (47 CFR 97.307a) A reasonable definition of good amateur practice includes the avoidance of waveforms that produce objectionable levels of interference to other users. Why then do we put up with commercial transmitters that consistently demonstrate such waveforms?

I postulate that the production of offensive CW signals is caused by one of two things, or both: 1) inconsiderate operators, or 2) ignorance on the part of equipment designers about the optimal CW waveform. The first thing must be left to our intrepid federal law-enforcement officials; I attempt to address the second in what follows.

One Wrong Way

For many years, the ARRL Handbook has published a figure in Chapter 12 with the caption, "Optimal CW waveform." In recent editions, it is Fig 12.20, whose shape is reproduced here as Fig 1. Even a cursory analysis reveals it to be far from optimal.

Granted, such an envelope can be produced by a simple R-C network. Rise and fall times may be controlled by the time constant of the network. The trouble is that such an envelope produces significant and unnecessary keying sidebands. It does that because it contains amplitude discontinuities; its amplitude does not change smoothly as it begins rising or falling. It has abrupt changes in its slope at those points.

Fig 2 is a spectral analysis of the waveform of Fig 1. Spectral occupancy is chiefly determined by the envelope shape and not by the keying speed. To be sure, keying such a waveform at high speed puts more energy into adjacent frequencies than at low speed; but the instantaneous amplitude of the keying sidebands is constant during the rise and fall times, regardless of keying speed.

Well, the Ten-Tec Orion does it better!

The Right Way

Fig 3 is a depiction of the ideal CW keying envelope. It rises uniformly from zero amplitude to its maximum; it also falls smoothly from there to zero. Envelope shape is sinusoidal; it is called "raised-cosine" keying, which is very close to the shape occupying minimum bandwidth.

Fig 4 is a spectral analysis of the waveform of Fig 3. By comparison with Fig 2, note that modulation sidebands have decreased in amplitude by a large factor. Digital signal processing (DSP) readily allows designers to implement such a keying envelope. Admittedly, it is difficult to achieve with traditional analog electronics; but for DSP-based rigs, it is relatively simple. Why then don't we see such waveforms from modern production radios generally?

ALC Effects

One reason is that automatic level control or ALC in a transmitter may modify the transmitted envelope. The job of ALC is to reduce drive to the final power amplifier in a transmitter so that it is not overdriven. In other words, it sets the maximum power output.

It follows that more drive than necessary to reach that maximum output power is initially applied. ALC reduces drive rapidly when the maximum is reached. That effectively shortens the rise time of CW elements because the attack time of ALC is fast. ALC decay time is usually slow, so the fall time is not usually affected by ALC. The result is an envelope resembling Fig 5 that has a sharp change of slope at the power set point.

Transmit gain control or TGC is an algorithm that largely avoids envelope distortion. It does that by reducing drive over time so that ALC does not have to work so hard. Drive is adjusted slowly-- more slowly than in ALC-- such that excess drive does not exceed a dB or so. Sharp slope changes caused by ALC are thus mitigated. More information about TGC can be found at


Certain production rigs could produce better CW envelopes than they do. If we are serious about 47 CFR 307a, let us apply it consistently across the variety of different modes, including CW. Our publications ought to indicate what is best without ambiguity. Perhaps then equipment designers would wake up and smell the coffee-- Doug Smith, KF6DX.