Power measurements are a must in the development of any RF or microwave product whether it's a mobile phone or a sophisticated radar system. The choice of an RF or microwave power measurement system is more complex than ever with power sensors/meters offering functionality that was only found in higher-end analyzers. With the large variance in product offerings and specifications on manufacturer's data sheets, it's helpful to have an understanding of the most important factors when evaluating USB power sensors/meters.
Choosing a USB power sensor involves many of the same criteria as traditional RF power meters and sensors. Factors like frequency range, dynamic range, accuracy, zero and calibration, speed of measurements, and triggering continue to be critical to the selection process.
Power sensors cover frequencies from several kHz to 110 GHz. The most commonly used ranges are through 6 GHz to 20 GHz. Since power sensors are broadband detectors, they detect all RF power at their input across the entire frequency range. Variations in the frequency response of the sensor are accounted for in a calibration table stored within the sensor.
Dynamic range depends on the type of sensor technology used. Diode-based sensors have the widest dynamic range usually ranging from -60 dBm to +20 dBm or more. Their wide dynamic range coupled with their quick response time make diodes the preferred solution in most applications. A diode sensor achieves a wide dynamic range by extending the useful range of the diodes beyond their square law region through the use of correction factors, and the use of multiple diode paths.
When using multiple paths, the method used to switch between these paths can have an effect on linearity. Most sensors measure one path at a time and switch at some threshold. The transition point can be a point of discontinuity or hysteresis leading to non-linearity or measurement delays. The latest power sensors continuously digitize both paths simultaneously and use a weighted average over the transition point.
Compared to diode-based sensors, thermistor-based sensors have a limited dynamic range from -20 dBm to +10 dBm, whereas thermocouple sensors typically have a dynamic range from -35 dBm to +20 dBm. The typical maximum input power value for most power sensors is +20 to +23 dBm. Power attenuators and couplers can be used to reduce the maximum power at the input of a power sensor, but their use introduces added reflections between sensor and attenuator. These reflections decrease measurement accuracy and require proper matching and more set-up time to calibrate out VSWR mismatches.
Overall accuracy is a combination of several error sources and is typically calculated by combining the errors in a standardized way. These error sources include: sensor to device under test (DUT) mismatch, calibration factors, linearity, noise, temperature, and zero-offset. Most manufacturers follow the ISO Guide (ISO/IEC Guide 98) to the Expression of Uncertainty in Measurement which explains in detail how uncertainty factors combine. Overall accuracies for power sensors range from 2% to 5%. Calibrating a power sensor requires connecting it to an external reference source. Zeroing a sensor usually requires disconnecting it from the device under test. Zero and calibration requirements can increase test times and cost, especially in automated test systems. If a power sensor requires periodic zeroing or calibration, the ATE system must be designed to accommodate these procedures. This usually requires some combination of costly switches, manual setup procedures, or dedicated software. Some newer sensors have eliminated the need for zero and calibration.
Power sensors typically specify several parameters that relate to measurement speed and the vocabulary varies between manufacturers. Some typical terms include sample rate, reading rate, and measurement rate. Sample rate is the rate at which analog-to-digital conversion takes place. Reading rate tells how fast the meter can convert raw samples into measurements. These are important specifications, but don't address the fundamental question of the time required to obtain a settled measurement.
The sample rate of a sensor helps determine a sensor's ability to measure pulse characteristics, but a high sample rate does not directly translate into fast, settled measurements. Reading rate has a more direct impact on measurement speed, but it may not accurately reflect the rate at which an instrument delivers settled power measurements. Settled measurements not only depend on sampling rate, but also on signal noise, signal amplitude, sensor architecture, and the integration time required for a stable measurement. When evaluating power sensors for measurement speed, it is best to evaluate the units side by side, rather than relying solely on datasheets.
For most basic power measurements, triggering is not a critical capability. However, if measurements on a specific portion of a pulsed signal are needed, or if there is a need to reduce test time in high-throughput ATE systems, triggering can be an important consideration. Basic power sensor triggering usually consists of an external TTL input. This can be useful for synchronizing power measurements with other instruments like signal generators, network analyzers, oscilloscopes, or additional power sensors. In automated test applications, the ability to externally synchronize measurements can be critical to reducing test times and maximizing throughput. More advanced power level triggering is also available in newer sensors that can synchronize measurements with the incoming RF signals.
Choosing the right power sensor for an application depends on the signal characteristics, but determining what can be measured cost-effectively also plays a major role. Measurement capabilities of sensors range from basic average power measurements to detailed pulse characteristics more typically found in a vector signal analyzer or dedicated pulse analyzer.
Most power meters are capable of delivering accurate average power measurements on continuous wave (CW) signals. These common signals are of constant amplitude and frequency. They are relatively uncomplicated and the capability of performing average power measurements is common to most power sensors. There is also a subset whose average power measurements are referred to as true average or true RMS.
A true average power measurement gives the total power that is incident to the sensor regardless of the modulation bandwidth of the input signal. True average measurements can be made using a sensor with a thermal detection unit or by using a diode detector in its square law region.
Thermal sensors produce a true average measurement based on the heat generated by RF energy. A true average diode detector includes capacitance that integrates the energy received by the detector, resulting in a measurement that closely approximates one from the thermal sensor. True average sensors are well-suited for measurements on broadband modulated signals and can measure all RF energy that is incident on the sensor input, whether that power is pulsed, CW, AM/FM, or in a complex modulated format.
Pulsed Signal Measurements
Using a sensor for analyzing pulsed signals presents a broader set of alternatives and some additional considerations. One approach to measuring pulse power is to assume a constant duty cycle value and a square pulse envelope. For square pulses and well-defined duty cycle, pulse power can be calculated by dividing the average power by the duty cycle. This method is simple, may be used with low cost average power sensors, and will give good results as long as the duty cycle is known and the pulses are square.
Peak power is the highest power point of the signal waveform, which in the case of a pulse, is usually the overshoot on the rising edge, but can occur elsewhere if significant ringing occurs. Peak power measurements are an important tool for evaluating signals on the input of power amplifiers. Peaks outside the amplifier's specifications will cause distortion..
The key to both pulse and peak power measurements lies in the sampling used to capture the pulse data. As shown in Figure 1, analog-to-digital converters are used to convert the sensor detector output into digital form. The faster an instrument samples, the finer the time resolution, resulting in more precise measurements overall and a better ability to capture peaks. The real-time sample rate of faster sensors is 500 kS/s, typically driven by the clock rate of the converters.
Fig. 1: Multiple path block diagram of a USB RF power sensor/meter.
Since the power envelope is being sampled, the likelihood of capturing an accurate peak improves with longer records. Thus, the measurement time must be long enough to capture a few cycles of modulation.
Burst signals, like GSM/Edge as well as other TDMA signals, are pulsed RF waveforms characterized by long pulse widths and long periods. In many cases, the power in a particular part of the burst is of interest. Time gated burst measurements allow users to specify a burst measurement window by setting a delay and sweep time, as shown in Figure 2.
Fig. 2: Diagram of a burst time slot.
The sensors then measure the average power, peak power, and minimum power within the specified gate. Burst measurements may be internally or externally triggered. If available, an automatic internal trigger can determine the trigger level by sampling the incoming signal, examining the data for maximums and minimums, and then setting the trigger level between maximum and minimum points.
In applications like commercial cellular transmission analysis or pulsed radar additional measurements such as overshoot, rise-time and fall-time are helpful for better qualitative analysis of the signal. Pulse profiling capability is relatively rare among USB power sensors.
The ability of a pulse profiling system to measure the frequencies contained in a signal's modulation envelope is referred to as video bandwidth. Higher video bandwidth means that a system can measure higher frequency components and therefore faster rise and fall times.
When visually representing a signal's envelope, the faster the meter's sample rate, the more accurately it can display a signal's characteristics. For repetitive pulsed signals, a sensor can use equivalent-time sampling to effectively sample at a much faster rate, enabling it to recreate a signal envelope with greater fidelity. As shown in Figure 3, equivalent-time sampling allows the sensor to accurately capture signals whose frequency components are higher than the sensor's real-time sample rate. However, the signal must be repetitive.
Fig. 3: Equivalent-time sampling improves fidelity for repetitive signals.
USB Sensor Advantages
For many of the applications described here, a USB-based power measurement system will offer a number of advantages over traditional meter and sensor configurations. These can include cost, convenience, flexibility and integration.
Because USB sensors are connected directly to a PC, no base unit is required lowering the cost of each unit. And since a single computer can be used for multiple sensors, USB sensors can be very economical. Similarly, USB power sensors provide flexibility due to their light weight and compact size, meaning they can be used in applications in which classic power meters are unsuited or cumbersome. Multiple sensors can be embedded in automated test systems or deployed in remote locations.
Since USB power sensors connect to a PC, common Microsoft programming environments are typically supported through dedicated API's or USBTMC. Most USB power meters include an easy to use software suite that simulates a traditional bench meter. With a modern PC, the performance of compact USB power sensor units meets or surpasses the performance of traditional power meters.
Adding to their flexibility, USB power sensors can be integrated with other test devices that have built-in computing power such as signal generators, spectrum analyzers, and oscilloscopes. This eliminates the need for a separate PC, reducing cost, and saving space on the lab bench.
A wide variety of RF and Microwave power measurement options are available today. For USB-based power meters, many of the traditional power sensor/meter specifications still come into play. However, USB sensor/meters can help simplify the chore of selecting a solution by eliminating the need to commit to a particular mainframe. Thanks to the compute power available in modern PCs, new features are available in these compact instruments that have historically required specialized analyzers.
About the Author
Richeal Chen is a product marketing manager at Tektronix with over 10 years of experience in the RF & Microwave field. Mr. Chen graduated in 2003 with a Bachelor's degree from Sichuan University in China.