Measurements, Testing, and Calibration Power and Signal Measurement Informational

What is the difference between a thermocouple, diode, and thermistor power sensor?

Q: How does mismatch affect power measurement?

Mismatch between the power sensor and the DUT output creates standing waves that cause the measured power to differ from the actual available or delivered power. Mismatch uncertainty: U_mismatch = ±20×log10(1 ± |Gamma_source| × |Gamma_sensor|). For source |Gamma| = 0.2 (14 dB RL) and sensor |Gamma| = 0.05 (26 dB RL): U_mismatch = ±20×log10(1 ± 0.01) = ±0.087 dB. For source |Gamma| = 0.5 (6 dB RL) and sensor |Gamma| = 0.1 (20 dB RL): U_mismatch = ±20×log10(1 ± 0.05) = ±0.42 dB. Mismatch is often the largest single error source in power measurement. To minimize: use a well-matched sensor (RL > 25 dB). Add an attenuator pad (6-10 dB) between the DUT and sensor: this reduces both reflection coefficients by 2× the attenuation.

The three main RF power sensor technologies differ in operating principle, dynamic range, and accuracy: (1) Thermocouple sensors: operate by converting RF power to heat, which generates a DC voltage via the thermoelectric (Seebeck) effect. Power range: -30 to +20 dBm (1 uW to 100 mW). Very broadband (DC to 50+ GHz). True RMS measurement (measures total power regardless of waveform or modulation). Accuracy: ±0.2-0.5 dB with calibration. Slow response (10-100 ms settling time). Best for: CW power measurement, calibration reference, wideband signals. (2) Diode sensors: use a Schottky diode (or diode array) to rectify the RF signal and produce a DC voltage proportional to the RF power. Two ranges: square-law region (P < -20 dBm): the diode DC output is proportional to P_RF (true RMS). Linear region (P > -20 dBm): the diode output is proportional to voltage (peak detection, not RMS; depends on waveform). Extended range: -70 to +20 dBm (with multiple diode stacks and signal conditioning). Very fast response (< 1 us for wideband diode sensors, enabling modulated signal analysis). Best for: average power measurement over a wide dynamic range, peak power measurement, and modulated signal power. (3) Thermistor sensors: use a temperature-dependent resistor in a DC-substitution bridge. RF power heats the thermistor; the bridge circuit reduces the DC bias to maintain constant total power dissipation. The DC power reduction equals the RF power. Power range: -20 to +10 dBm (10 uW to 10 mW). The most accurate sensor type (±0.1 dB, directly traceable to DC power standards). Very slow (1-10 s settling). Used for: primary calibration standards, metrology, and verifying other sensor types.

Category: Measurements, Testing, and Calibration

Updated: April 2026

Product Tie-In: Power Meters, Spectrum Analyzers, Signal Generators

RF Power Sensor Technologies

Selecting the right power sensor technology is essential for accurate RF power measurement. Each type has strengths and weaknesses that make it best suited for specific applications.

Detailed Comparison

(1) Thermocouple: a thin-film thermocouple (typically Ta-N/Ta or N-type/P-type thin film) absorbs the RF power in a matched 50-ohm resistive element. The heat flows through a thermal path to a reference junction. The temperature difference generates a Seebeck voltage proportional to the absorbed power. Modern thermocouple sensors use multiple thermocouples in series (thermopile) for higher sensitivity. Key characteristics: true RMS (responds to total power, independent of waveform). Frequency range: DC to 110 GHz (sensor-dependent). Sensitivity: -30 dBm minimum (1 uW). Maximum power: +20 to +30 dBm (with attenuator). Response time: 5-100 ms. Temperature coefficient: 0.01-0.05 dB/°C. (2) Diode: a zero-bias or low-bias Schottky diode rectifies the RF signal. In the square-law region (V_RF << V_T ≈ 25 mV): I_DC ∝ V_RF^2 ∝ P_RF (true RMS). In the linear region (V_RF >> V_T): I_DC ∝ V_RF ∝ sqrt(P_RF) (peak detection). Dual-path wideband sensors (e.g., Keysight U2000 series): automatically switch between square-law (low power) and corrected linear (high power) detection, providing wide dynamic range with modulation correction. Key characteristics: wide dynamic range: -70 to +20 dBm (90 dB). Fast response: < 10 us (enables peak and time-gated measurements). Modulation bandwidth: 100 MHz (for diode sensors with video output). Temperature coefficient: 0.02-0.1 dB/°C. (3) Thermistor: the feedback loop (DC substitution) makes the thermistor measurement independent of the thermistor characteristics (nonlinearity, aging): absolute accuracy depends only on the DC measurement accuracy and the RF-to-DC efficiency of the mount. Key characteristics: accuracy: ±0.1 dB (highest of all sensor types). Traceable to national DC power standards (NIST fundamental power measurements use calorimetric methods or DC-substitution thermistor mounts). Dynamic range: -20 to +10 dBm (30 dB) — very limited. Response time: 1-35 seconds (very slow). Used almost exclusively for calibration and metrology, not for production testing.

Selection Guide

Application → recommended sensor: General CW measurement (1-20 GHz, moderate power): thermocouple. Widest coverage with good accuracy. Modulated signal measurement (OFDM, QAM, pulsed radar): diode sensor with active video bandwidth. Measures true average power even for signals with high peak-to-average ratio. Low-level power measurement (< -30 dBm): diode sensor (only type with sufficient sensitivity below -30 dBm). Peak power measurement: diode sensor with video output or dedicated peak power sensor. Measures instantaneous power with ns resolution. Calibration lab / primary standard: thermistor (highest accuracy, NIST-traceable). High power (> +30 dBm / 1W): use an attenuator or coupler to reduce the power to the sensor range. High-power thermocouple sensors handle up to +44 dBm (25 W) with an internal attenuator.

Calibration and Traceability

Power sensors require periodic calibration to maintain accuracy: (1) Calibration factor: the ratio of actual RF power to indicated power, specified at each frequency. The power meter applies the calibration factor during measurement. For thermocouple and diode sensors: the manufacturer provides calibration factors at 50-100 frequency points. (2) Reference calibration: compare the sensor against a transfer standard (a calibrated thermistor mount or calorimeter). Calibration interval: 12-24 months. (3) Zero/cal adjustment: before each measurement session, zero the sensor (cap on, no RF) and perform a 50 MHz internal calibration (the power meter has a built-in 1 mW, 50 MHz reference oscillator). This corrects for ambient temperature changes and sensor drift.

Power Sensor Equations

Thermocouple: V_DC ∝ P_RF (true RMS)
Diode sq-law: I_DC ∝ V_RF² ∝ P_RF (P < -20 dBm)
Diode linear: I_DC ∝ V_RF (peak, P > -20 dBm)
Thermistor: P_RF = P_DC_before - P_DC_after
Mismatch Unc: ±20log₁₀(1 ± |Γ_s|·|Γ_L|)

Common Questions

Frequently Asked Questions

Which sensor is best for 5G NR signal measurement?

5G NR uses OFDM signals with high peak-to-average power ratio (PAPR): 8-12 dB for the downlink. A thermocouple sensor measures the true average power correctly (regardless of PAPR) but cannot measure peak power. A diode sensor in the square-law region (< -20 dBm) also measures true average power correctly. Above -20 dBm: the diode enters the linear region where it measures peak-weighted power (not true average). Modern wideband diode sensors with digital correction (e.g., Keysight U2040XA) apply a correction factor based on the signal statistics: the user specifies the modulation type, and the sensor applies the appropriate crest factor correction. For accurate 5G NR power measurement: use a wideband diode sensor with modulation correction enabled, or a thermocouple sensor with the signal level adjusted to the thermocouple range.

How does mismatch affect power measurement?

Mismatch between the power sensor and the DUT output creates standing waves that cause the measured power to differ from the actual available or delivered power. Mismatch uncertainty: U_mismatch = ±20×log10(1 ± |Gamma_source| × |Gamma_sensor|). For source |Gamma| = 0.2 (14 dB RL) and sensor |Gamma| = 0.05 (26 dB RL): U_mismatch = ±20×log10(1 ± 0.01) = ±0.087 dB. For source |Gamma| = 0.5 (6 dB RL) and sensor |Gamma| = 0.1 (20 dB RL): U_mismatch = ±20×log10(1 ± 0.05) = ±0.42 dB. Mismatch is often the largest single error source in power measurement. To minimize: use a well-matched sensor (RL > 25 dB). Add an attenuator pad (6-10 dB) between the DUT and sensor: this reduces both reflection coefficients by 2× the attenuation.

Can I measure pulsed radar power with these sensors?

Thermocouple: measures the average power of the pulsed signal. Average power = peak power × duty cycle. If peak power = 10 kW and duty cycle = 0.001 (1 ms pulse, 1 s PRI): average = 10 W (+40 dBm). The thermocouple measures 10 W. Diode sensor (CW mode): also measures average power. But: if the peak power exceeds the sensor maximum input (typically +20 dBm = 100 mW): the sensor is damaged. Must use an attenuator/coupler to reduce peak power to safe levels. Diode sensor (peak mode): with a video bandwidth > 1/pulse_width: the sensor tracks the pulse envelope. Measures the peak power directly. For a 1 us pulse: need video BW > 1 MHz. Peak power sensors (e.g., Boonton RTP series): designed specifically for pulsed signals with < 10 ns rise time and 195 MHz video bandwidth.

Keep Reading