Power measurements are not just for the laboratory; they are essential to the operation of many RF/microwave-based systems. Monitoring transmitted power from an antenna, for example, ensures that a communications system remains within its mandated power limits. Measuring received power in some systems can also help calibrate receiver sensitivity or identify the distance to a threat.
Fortunately, the various means of measuring power in RF/microwave systems have improved dramatically over the last several decades, to a point where numerous semiconductor suppliers now offer integrated circuits (ICs) with power-measurement capability comparable to some benchtop instruments. To better appreciate current power-measurement solutions, it may help to review the fundamentals of RF/microwave power measurements.
Power can be expensive to generate, and equally costly to maintain. Any power lost in a communications system—for example, due to loss through a filter—can disrupt reception of a desired signal. Most systems have some means of measuring power to detect problems before they happen, and prevent operational interruptions.
From basic electricity, power is energy per unit time, defined in values of wattage, with 1 watt (W) equal to 1 Joule (J) per second. In terms of voltage, one volt (V) is equal to 1 W per ampere (A), or 1 V = 1 W/A. Power is often treated as the product of voltage and current, or P = V × I, although this quantity tends to vary in most systems as a function of time. Because power can change so much for some signals (such as digitally modulated signals), measuring RF/microwave signal power requires test equipment suited to a specific type of signal.
In the laboratory, the trusted means of measuring the power levels of high-frequency signals is by means of a power meter and the appropriate power sensor. Three types of power sensors are used for RF/microwave power measurements: thermistor, thermocouple, and diode-based sensors. All three sensor types are appropriate for measuring the power levels of continuous-wave (CW) signals, but only diode sensors have the fast response times to accurately measure the time-varying power of pulsed and modulated signals.
Power sensors based on thermistors gauge the level of power from rises in temperature due to the heating effects of applied power on resistors. Thermistors are semiconductors with a negative temperature coefficient. They are a type of bolometer, a device in which resistance changes with temperature changes caused by applied RF/microwave power.
Because thermistors exhibit nonlinear resistance characteristics as a function of applied power, using them for power measurements is anything but straightforward. They are typically used with a bridge and some form of bias (AC or DC) to maintain a thermistor assembly or mount at a constant resistance with applied RF/microwave power. A power meter tuned to the sensor’s changes in bias energy as a function of applied power is then used to calculate the RF/microwave power over a dynamic range enabled by the thermistor mount and its supporting electronics, including video amplifiers.
Thermocouple-based power sensors work on the principle that some dissimilar metals will produce a voltage due to temperature differences at the junctions of two of the metals. Known as the Peltier effect—named for French physicist Jean-Charles Peltier, who discovered the phenomenon in 1834—a thermocouple formed of two appropriate metals can generate a small voltage in response to temperature rises. When current flows through a junction of a thermocouple formed of two of the proper metals, it induces a temperature rise in one of the metals in the junction, with heat absorbed by the other metal in the junction. The resulting flow of electrons or charge can be used as part of a power sensor. A power-sensor thermocouple is usually a loop or circuit made of two different metals, with two junctions. Heat is applied to one junction but not the other, with resulting electron flow towards the cold junction. Modern power sensors based on thermocouple technology generally use thermocouples fabricated as semiconductor chips with the appropriate metal junctions.
Diode-based power sensors make use of a Schottky diode’s capability of rectification, or converting an alternating current (AC) flowing alternately in two directions to a direct current (DC) signal flowing in only one direction. As diodes were fabricated on different materials, from early silicon devices to later gallium arsenide (GaAs) devices, the bandwidth and frequency capabilities of the devices increased, supporting power measurements through RF and microwave frequencies.
Of course, the rectification process has nonlinear voltage-current characteristics, so that various forms of linearization and temperature compensation are needed to perform accurate RF/microwave power measurements with diode-based sensors. But these correction factors can be measured for a specific diode type and stored in memory on board a sensor, such as electronically erasable programmable read-only memory (EEPROM).
Different types of RF/microwave power sensors are necessary because of differences in high-frequency signals. When measuring CW signals, all three types of power sensor would yield the same results. But that would not be the case for pulsed or modulated signals, especially signals with advanced digital modulation formats. The need for pulsed power measurements actually goes back as far as World War II, with the development of high-power radar systems and the magnetron and klystron vacuum electronic devices created for those radar systems. The wireless communications explosion that began in the early 1990s brought with it an expansion of complex modulation formats—notably those based on changing relationships of in-phase (I) and quadrature (Q) signal components, such as quadrature amplitude modulation (QAM). The crest factors (the ratio of the peak to average power levels) for digital modulated signals can vary widely, requiring a power measurement solution with an extremely wide dynamic range.
Many modern wireless communications systems transmit and receive signals that are noiselike or pseudorandom in nature, including Long-Term-Evolution (LTE) cellular communications systems, and making power measurements on such signals involves as much statistical analysis as signal capture. Because of the random nature of amplitude peaks for some of these signals, measuring power requires taking a large number of samples for a given communications channel. Next, statistical analysis—such as a complementary cumulative density function (CCDF)—must be performed to calculate the probability that power will be at or below a certain level in that channel during the sampled time period.
When measuring the peak envelope power (PEP) of the types of random or noiselike signals used in modern communications systems, a power-measurement solution must take on many of the traits of a digital sampling oscilloscope (DSO), such as sampling the input signal and making a large number of measurements in a short time in order to capture as many of the peak power levels as possible. Obviously, the greatest accuracy will come from a sensor capable of operating at fast sampling rates and performing as many measurements per second as possible, with many pulsed/peak power sensors characterized much like DSOs.
Traditionally, power sensors have been one-half of a power-measurement solution, working with a compatible power meter. Power meters can be one of two types: a terminating power meter or a directional power meter. A terminating power meter is one in which the power sensor is used to measure the output power from a source but at the same time interrupting or terminating the flow of power. A directional power meter couples some of the power between a source and a load and can measure direct and reflected power levels with the aid of a directional coupler, without necessarily interrupting the power flow through the system under test.
The power meter can contribute to critical measurement linearity by providing a power reference signal for calibrating a connected power sensor. Standards organizations such as the National Institute of Standards and Technologies (NIST) in the United States specify traceable 0-dBm (1-mW) reference signal for calibrating power meter sensors. This reference is usually provided by means of a thermistor mount within the power-meter housing.
Modern power meters, especially when used with appropriate diode sensors for pulsed or PEP measurements on digitally modulated signals, can provide results in terms of detected power in addition to a host of other details about a signal. These include pulse width, pulse repetition frequency (PRF), and pulse repetition interval (PRI) for pulsed signals.
Traditions change, however, and laboratory-grade power measurements are now possible without the power meter, using a new breed of power sensor that uses a Universal-Serial-Bus (USB) connection to a personal computer and the computer for control and signal processing. These are available from a number of test-equipment manufacturers and even components suppliers, such as Mini-Circuits. A USB power sensor/meter typically incorporates a microprocessor-based controller in addition to the power-measurement circuitry, and uses a personal computer (PC) or laptop computer as the graphical user interface (GUI), computational capabilities, and memory storage. USB power sensors/meters are available for both CW and pulsed/modulated power measurements, over frequency ranges as wide as 10 MHz to 26.5 GHz.
One final note: This discussion has focused on laboratory power measurements. In many cases, power measurements may be part of the normal operation of an electronic device or system, and performed by an integrated circuit (IC) incorporating a diode detector, logarithmic-amplifier (logamp) detector, or root-mean-square (RMS) detector. Such ICs are often used for power monitoring and control in modern communications systems. They are capable of power measurements over wide dynamic ranges exceeding 60 dB at microwave frequencies, even on signals with complex modulation such as QAM.