GaN or GaAs, TWT or Klystron – Testing High Power Amplifiers for RADAR Signals Using Peak Power Meters
Before we look at the peak power meter and its capabilities, let’s look at different technologies used in high power amplifiers (HPA) for RADAR systems, particularly GaN on SiC, and why it has grabbed the attention over the past decade.
The second part of the 20th century has seen the rise and dominance of semiconductor technology in electronics, and the ushering out of the Vacuum Electron Devices (VEDs) commonly referred to as tubes. Aside from rock bands love for the “warm” sound of the tube amplifier, one area in which VEDs maintained their dominance has been PAs that operate up to mega watts of output power at very high frequencies (1 GHz to 100 GHz) found in RADAR systems. Most commonly used VED technologies in RADAR applications today are Traveling Wave Tubes (TWT), Klystrons, Magnetrons and Gyrotrons. TWT amplifiers (TWTA) provide what the RADAR systems need the most multi-octave bandwidths, multi-kilowatt peak power output, support of high frequencies, as well as ruggedness and reliability. Compared to TWTAs, Klystron power amplifiers (KPA) offer better efficiency.
About three decades ago semiconductor based PA solutions called solid state power amplifiers (SSPA) started making modest inroads as an alternative technology for certain RADAR applications. Silicon-based laterally diffused metal oxide (LDMOS) offered a few hundred watts of output power, ruggedness and reliability. Its upper frequency limited to about 3 GHz (S-Band) at high output powers curbed its applicability in RADAR. Gallium arsenide (GaAs), a wide-bandgap semiconductor, overcame the high frequency deficiency of LDMOS, reaching above 100 GHz, yet its lower operating voltage has limited its output power capability. High power GaAs amplifiers have often required paralleling of multiple devices to reach desired power levels at the expense of loss in efficiency. GaAs found common use in RADAR but still was not a viable alternative in most high power RADAR applications. About a decade ago when GaN was introduced it created much excitement since it offered hundreds of watts of output power at very high frequencies; it also provided the ruggedness required in RADAR applications. In the earlier years of GaN the concerns over reliability due to lack of a suitable substrate to get the heat out of the die were addressed by placing GaN on silicon-carbide (SiC) wafer substrate. While GaN provided the heat capacity, SiC provided three times better thermal conductivity. The enhanced thermal performance improved the reliability and ruggedness.
While TWTAs and KPAs are here to stay for the foreseeable future, GaN has established itself as a suitable lower cost alternative for the long run.
Peak Power Meter for Pulsed RADAR Measurements
The most critical analysis of the pulsed RF signal takes place in the time domain. Since peak power meters measure, analyze and display the power envelope of a RF signal in the time domain, they are an essential tool for measuring, analyzing and characterizing anomalies and behavior of high power amplifiers used in pulsed RADAR applications. Figure 1 shows a simplified block diagram of a benchtop peak power meter.
The front end of a power meter is the RF envelope detector housed in a power sensor. The detector removes the RF carrier and generates an analog waveform representing the envelope of the RF input signal.
The most critical specification of the detector is its response time to a pulsed RF signal or its risetime. If the detector does not have the bandwidth to track the envelope of the signal, accuracy of all measurements including peak, pulse and average power is compromised (Figure 2).
The detector output is then digitized by an analog-to-digital converter (ADC). The digitized samples are processed by the digital signal processor for linearization and measurement analysis. The processed waveform is displayed in the time domain as power-versus-time along with automated pulse and marker measurements. In Boonton’s peak power meters, on the vertical axis the power (or voltage) is displayed in watts, volts or dBm with ability to change the scaling and vertical center. On the horizontal axis timebase can be set as low as 5 ns per division (50 ns span) to zoom in to a specific portion of the waveform such as rising or falling edge to observe fine details of the waveform. Boonton’s peak power meters use Random Interleave Sampling (RIS) technique which yields 100 ps resolution on repetitive waveforms. Figure 3a shows a time domain diagram of the conventional sampling and interpolation method, Figure 3b shows the RIS method. Figure 3c shows how the resolution is improved going from conventional to the RIS based method on a Boonton 4500B Peak Power Analyzer.
Peak power meters can be triggered by the incoming RF signal or by an external gating (baseband) trigger signal applied to auxiliary inputs. Real Time Power Processing™ featured in the 55 series USB Peak Power Sensors allow the sensor to trigger on 100,000 pulses per second, capturing each pulse and glitch events. Advanced triggering features of Boonton peak power meters, like trigger holdoff, delays the re-arming of the trigger which is useful when working with interrogation friend or foe (IFF) RADAR signals.
In terms of measurement capabilities, peak power meters perform numerous manual, automated marker and automated pulse measurements. Automated pulse measurements (Figure 4) provide measurement values of a number of critical parameters that help characterize the performance of the power amplifier and the RADAR system. Rise and fall time indicate the amplifier’s ability to output a pulsed RF signal. Overshoot pinpoints potential ringing problems. Droop shows the amplifier’s power supply limitations with prolonged pulse widths. Pulse Width, period, pulse repetition rate and duty cycle measurements provide other time domain characteristics of the signal. There are also a number of automated marker measurements (Figure 4 on the right) that enable time gated measurements. These measurements are performed between two markers and provide average, peak, min and max power readings, as well as peak to average ratio, delta marker of power level and delta time marker measurements.
A significant advantage power meters have over other measurement instruments is the size of the power sensor. It is small and light enough to be directly connected to the measurement port without the need for a RF cable which can degrade measurement accuracy due to impedance mismatch and cable loss, especially at higher frequencies.
For automated test environments all Boonton peak power meters can be remotely accessed via various interfaces such as USB, LAN (TCP/IP), RS232 and GPIB, depending on the model.
More reading materials on the topic of power measurements are provided at the end of this article with links to articles and application notes as well as Boonton’s newly revised RF Power Measurement Guide that provides practical insights to making measurements for RADAR signals.
Test Set-ups for Making Pulsed RADAR Measurements
Although there are numerous pulsed RF amplifier architectures, two types are considered in this article. The first is a modulating amplifier; the input is CW and the output is pulsed RF where a gating signal modulates the incoming signal to achieve the desired pulse RF signal (Figure 6).
In Figure 6 a dual channel meter measures the input signal power at P1 and the reflected power at P2 in time domain for return loss calculation, as well as monitoring anomalies of the reflected signal. At P3 the output of the amplifier is monitored and measured. Key amplifier parameters that were mentioned in the previous section are measured at P3 and P4 monitors the reflected power of the load. The gating signal that modulates the RF input signal can also trigger the power meter enabling delay and latency measurements. Boonton’s highest performance benchtop peak power analyzer, model 4500B, is equipped with waveform math capabilities and is capable of displaying gain and return loss in time domain. The instrument is also equipped with two scope channels. When the gating signal to the amplifier triggers the peak power meter, both the gating signal and the output of the amplifier can be displayed on the peak power meter enabling critical timing measurements, as well as detecting if any amplifier anomalies are caused by the gating signal.
While benchtop models support one (single) or two (dual) channels, the 55 Series USB peak power meters can measure and display up to eight channels on a single GUI window. Using three or four USB sensors, amplifier input and output power, reflected input power and reflected load power can all be measured and displayed on the same trace window or in an automated test set-up via remote programming (up to 16 or 32 channels).
The second type is a gain only amplifier where the output is an amplified (and often distorted) version of the pulsed RF input signal (Figure 7) with no gating signal supplied to the amplifier. The set-up is ideal for analyzing not only fully assembled amplifiers but also a subassembly such as the driver stage or the final stage of the amplifier or even a semiconductor power transistor like GaN. The set-up uses three peak power meters and a directional coupler to make scalar like gain and return loss measurements of the amplifier.
When evaluating new technologies like GaN, monitoring power droop across the pulse width is critical as it can be an indicator of the limitations of the thermal properties of the GaN device and its package. Time domain peak power measurements can be taken at the output of the amplifier (Figure 9).
Since the power rating of a typical sensor is about +20 dBm, the output of the PA is attenuated to protect the power sensor while making the output power measurement. Before measurements can be taken a thorough calibration procedure is required at the frequencies in which the amplifier is going to be tested to account for losses in the signal path. The losses that need to be calibrated out in the test set-up of Figure 7 and calculations required to compute gain and return loss are provided below.
L1: Loss from Signal Generator output to the FWD port of the directional coupler.
L2: Loss from Signal Generator output to the power amplifier input.
L3: Loss from amplifier output to the 40 dB attenuator output.
L4: Loss from amplifier input into the REV port of the directional coupler.
Once the losses are measured, input, output and reflected power measurements can be made:
P1: Power reading at FWD port of the directional coupler.
P2: Power measured at the 40 dB attenuator output.
P3: Power measured at REV port of the directional coupler.
PA input power = P1+L1-L2
PA output power = P2+L3
PA input reflected power = P3+L4
All Boonton peak power meters are capable of adding an offset to the measurements so the math above can be done by the meter once the losses are measured and entered to each channel as an offset.
The input, output and reflected power measurements can be used to compute gain (S21) and input return loss (S11).
PA Gain (dB) = PA output power (dBm) – PA input power (dBm)
PA input return loss (dB) = PA input power (dBm) – PA input reflected power (dBm)
The 4500B can perform these measurements in time domain using waveform math.
It is important to note that in both test set-ups, Figure 6 and Figure 7, the directional couplers need to have excellent directivity in order to make accurate power measurements, especially for return loss calculations. Unused ports of the couplers must be terminated with 50 ohms during measurements.
Figure 8 shows three waveforms measured using the test set-up in Figure 7 using the 55 Series Peak Power Sensors. The input waveform is displayed on CH1, reflected waveform on CH3 and the output on CH2. Note that automated measurements performed on all three channels are displayed to the left of the trace display window. The measurements can be transferred to a spreadsheet to perform the necessary gain and return loss calculations, as well as other parameters of interest. In an automated test environment the same measurements can be accessed through remote programming to perform gain and return loss computations as well. Droop measurement capabilities are shown in Figure 9 using the Boonton 55318 USB Peak Power Sensor. Power droop can be measured either using the automated pulse measurements or using automated marker measurements as well as horizontal markers. The automated marker measurements can display the droop placing markers at the desired points on the waveform and using MkRatio. Alternatively reference lines can be placed on the vertical axis at the desired high and low points of the pulse to measure the droop. Automated Pulse measurements are computed automatically based on pulse definition irrespective of marker or reference line placements.
VED based amplifiers have dominated the PAs used in the aviation and warfare RADAR systems for the past seven decades. However, new semiconductor based SSPAs, have made inroads to various RADAR applications, especially GaN based ones. Regardless of technology used in the RADAR PA, high resolution highly accurate time domain power measurements are critical to understand the amplifier performance and behavior. Peak power meters are an essential measurement tool for time domain power analysis to test RADAR power amplifiers for R&D, quality, manufacturing, field support and system calibration.