Power measurements are a must in the development of any RF or microwave product, whether itís a mobile phone or a sophisticated radar system. The choice of an RF or microwave power measurement system is more complex than ever with the recent availability of new functions in power meters previously reserved for higher-end analyzers. With the large variance in product offerings and specifications on manufacturerís data sheets, itís helpful to have an understanding of the most important factors when evaluating USB power sensors/meters.
Basic Factors Choosing a USB power sensor involves many of the same criteria as traditional 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 the 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 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 percent.
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 donít address the fundamental question of the time required to obtain a settled measurement.
The sample rate of a sensor helps determine a sensorís 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.
RF Power Measurement over USB is a typical application of metering, the users will be technically sound, so in this case if the USB Metering application by default allows to show the captured signal in raw from then the user can find out what they want. The can switch the sensing elements and find the differences between the sensing devices. I thing this will be best way in the today's highly changing RF Field environment.
Blog Doing Math in FPGAs Tom Burke 23 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...