CW Source
Spectral Purity of a Single-Tone Carrier
The defining attribute of a CW source is that it puts essentially all of its energy into one spectral line. In the ideal case the output is a pure sinusoid v(t) = A·cos(2πf0t), a delta function in the frequency domain. Real sources depart from that ideal in three measurable ways: harmonic distortion (energy at integer multiples 2f0, 3f0, and so on), non-harmonic spurious tones from reference feedthrough and synthesizer fractional-N artifacts, and a noise pedestal around the carrier that is quantified as phase noise. A high-quality benchtop unit holds harmonics at −30 dBc when running its output amplifier near compression, improving to −60 dBc or better when backed off, and keeps spurious below −60 dBc across the band.
Frequency stability comes from the reference oscillator rather than the microwave hardware itself. A synthesized CW source phase-locks a tunable oscillator, often a YIG-tuned oscillator or a fractional-N VCO loop, to an internal oven-controlled crystal or to an external 10 MHz house standard. Because frequency multiplication scales phase noise by 20·log(N), a source that synthesizes 40 GHz from a 100 MHz reference inherits about 52 dB of added phase noise relative to the reference, which is why millimeter-wave CW sources are far harder to keep clean than HF or VHF units.
Output power must be both accurate and flat. Internal automatic level control (ALC) loops sense forward power with a diode detector and adjust a PIN-diode or vector attenuator to hold the set level within roughly ±1 dB across a multi-octave span, with finer leveling possible against an external power sensor. For receiver and component testing the absolute level matters because it sets the calibration of insertion loss, gain, and compression measurements.
Power, Level, and Phase-Noise Relations
v(t) = A × cos(2πf0t) → single line at f0
Power in dBm (50 Ω system):
PdBm = 10 × log10(PmW) → 0 dBm = 1 mW; +10 dBm ≈ 0.71 Vrms
Multiplied phase noise:
Lout(f) ≈ Lref(f) + 20 × log10(N) dBc/Hz
Harmonic distortion (relative):
Hn = 10 × log10(Pnf0 / Pf0) dBc
Where A = peak amplitude, f0 = carrier frequency, N = multiplication ratio, L = single-sideband phase noise. Example: a 100 MHz reference at −150 dBc/Hz multiplied to 10 GHz (N = 100) yields ≈ −110 dBc/Hz, ignoring loop noise.
CW Source vs. Other RF Stimulus Types
| Source Type | Output | Modulation | Typical SSB Phase Noise (10 GHz, 10 kHz) | Best Use |
|---|---|---|---|---|
| CW source | Single tone, fixed f, fixed power | None | −110 to −130 dBc/Hz | Compression, LO substitution, spectral purity tests |
| Swept (CW sweep) source | Stepped or ramped frequency | None | −105 to −125 dBc/Hz | Filter and network response, return loss |
| Pulsed source | Gated CW envelope, low duty cycle | Pulse | Carrier dependent | Radar, high-power GaN load-pull |
| Vector signal generator | Modulated I/Q waveform | AM/FM/φ/I-Q | −100 to −120 dBc/Hz | 5G, Wi-Fi, EVM and BER testing |
| Free-running oscillator | Single tone, unlocked | None | Worse, drifts with temperature | Low-cost embedded LO, beacons |
Frequently Asked Questions
What is the difference between a CW source and a signal generator?
A CW source is the simplest signal generator: one unmodulated tone at a fixed frequency and power. A full signal generator adds AM, FM, phase, pulse, and vector I/Q modulation, plus sweep and list modes. Any signal generator can run as a CW source by turning modulation off, but a dedicated CW source omits that hardware, making it cheaper and often lower in phase noise, ideal for LO substitution and continuous amplifier or filter stimulus.
How much harmonic and phase noise does a typical CW source have?
A good microwave CW source holds harmonics near −30 to −60 dBc and non-harmonic spurious below −60 dBc. SSB phase noise at 10 GHz is roughly −110 to −130 dBc/Hz at 10 kHz offset, degrading about 6 dB per octave of carrier frequency because multiplication adds 20·log(N). Power flatness is within ±1 dB, and frequency accuracy is 1 part in 108 or better when locked to a 10 MHz reference.
Why use a CW source instead of a pulsed source for amplifier testing?
CW applies continuous power, the worst-case thermal load, which is correct for measuring saturated output power, gain compression, steady-state efficiency, and thermal qualification of amplifiers. A pulsed source duty-cycles the RF so the device stays near its quiescent temperature, which suits high-power GaN load-pull where CW would overheat the part. CW also gives the cleanest single-tone spectrum for harmonic, spurious, and phase-noise measurements.