CW Sensor
How CW Power Sensors Detect RF Energy
A CW sensor is the detector head of an RF power measurement system; it terminates the line in a precision 50 Ω load and produces an output proportional to absorbed power. Two detection technologies dominate. Thermocouple sensors absorb the RF in a thin-film resistive load and read the resulting temperature rise as a thermoelectric voltage. Because heat is intrinsically proportional to true average power, a thermocouple head responds correctly to any waveform, but its sensitivity is low, limiting the bottom of its range to roughly −30 dBm (1 μW). Diode sensors rectify the RF directly, giving far greater sensitivity down to −70 dBm, but they are only true-average responding while operating in the square-law region.
The square-law region is the heart of why these heads are labeled "CW." Below about −20 dBm a Schottky diode's rectified output is proportional to input power, so the reading is correct for any signal. Above −20 dBm the diode transitions toward its linear (envelope-detecting) region, where output tracks voltage rather than power. To recover an accurate reading at higher levels, the power meter applies a correction curve that was measured with a pure CW (single-tone) source. Apply that same CW-calibrated head to a modulated signal in this region and the correction is wrong by an amount that scales with the signal's peak-to-average ratio.
Every reading is scaled by the sensor's calibration factor, which folds together effective efficiency and mismatch loss at the test frequency. Modern sensors store a full cal-factor table in EEPROM so the meter reads the table automatically; older heads required the operator to enter the factor by hand from a printed chart. Selecting the wrong frequency point applies the wrong correction and biases the result directly.
Square-Law Region and Calibration Factor
PdBm = 10 × log10(PmW / 1 mW) (0 dBm = 1 mW)
Square-law diode output:
Vout ≈ β × Pin (valid for Pin < −20 dBm)
Corrected (true) power via calibration factor:
Ptrue = Pindicated / Kcal (Kcal = effective efficiency × mismatch)
Mismatch uncertainty bound:
Umm = 20 × log10(1 ± |ρs| × |ρl|) dB
Where β = diode sensitivity, Pin = incident power, Kcal = calibration factor, ρs = source reflection coefficient, ρl = sensor reflection coefficient. Example: ρs = 0.20, ρl = 0.091 → Umm ≈ ±0.16 dB.
CW Sensor Technology Comparison
| Sensor Type | Dynamic Range | Frequency | True Avg? | Best Application |
|---|---|---|---|---|
| Thermocouple CW | −30 to +20 dBm | 100 kHz to 50 GHz | Yes (all waveforms) | General CW & pulsed average power |
| Schottky diode CW | −70 to −20 dBm | 10 MHz to 110 GHz | Square law only | Low-level CW signals, receivers |
| Wide-dynamic-range diode | −70 to +20 dBm | 9 kHz to 6 GHz | CW-corrected | Single-tone, ATE power leveling |
| Thermistor (bolometer) | −20 to +10 dBm | 10 MHz to 26.5 GHz | Yes (all waveforms) | Transfer / reference standard |
| True RMS / average sensor | −60 to +20 dBm | 50 MHz to 18 GHz | Yes (modulated) | Modulated & multitone signals |
Frequently Asked Questions
Can a CW sensor measure the power of a pulsed or modulated signal?
A thermocouple CW head responds to true average power for any waveform within its range and bandwidth, so it reads pulsed or modulated average power correctly. A diode CW head is only true-average (square law) below about −20 dBm; above that the meter applies a sinewave-based CW correction, so a modulated signal produces an error that grows with peak-to-average ratio. For high-level modulated signals use a wide-dynamic-range square-law diode, a thermocouple, or a dedicated true RMS sensor.
What is the calibration factor on a CW power sensor and why does it change with frequency?
The calibration factor corrects for effective efficiency (power actually dissipated in the sensing element) and input mismatch at the test frequency. It runs 95 to 99 percent at low frequency and falls to 85 to 92 percent above 50 GHz as connector loss, skin effect, and mismatch worsen. The meter divides the measured power by the cal factor for the entered frequency, so selecting the wrong frequency point biases the reading directly.
How much measurement uncertainty does source and sensor mismatch add?
Mismatch is usually the dominant uncertainty term. The bound is Umm = 20 log(1 ± |ρs| × |ρl|) dB. For a source VSWR of 1.5 (ρ = 0.20) and sensor VSWR of 1.2 (ρ = 0.091) the bound is about ±0.16 dB, or roughly ±3.7 percent. Precision CW sensors specify low input VSWR (1.10 to 1.18) and pads or tuners are added to cut source reflection when sub-percent accuracy is needed.