Measuring Ruggedness in High Power RF Amplifier Designs
By Brian Battaglia, Dave Rice and Phuong Le, HVVi Semiconductor, Inc.

Ruggedness is a crucial metric in high power RF amplifier designs. Today’s RF amplifiers are expected to not only provide excellent performance characteristics in terms of output power, gain and efficiency, but also to sustain that performance for more than 20 years in the field. Traditionally, designers have built amplifiers using components fabricated in cost-effective silicon-based technologies such as bipolar or LDMOS. While these technologies can operate at high voltages, they suffer from low ruggedness ratings. For example, LDMOS has an inherent destructive mechanism (i.e. a parasitic bipolar transistor) built into the transistor structure which limits the ability of the device to work reliably at high operating voltage.

More recently, engineers at HVVi Semiconductors have developed a new vertical silicon MOSFET specifically for high power RF amplifier applications. Called the High Voltage Vertical Field Effect Transistor (HVVFET™), this transistor uses a novel structure that is inherently more rugged than competing RF power transistors. This paper will take a look at how designers measure ruggedness in RF power amplifiers and how they determine if a specific power amplifier can sustain high levels of RF performance in a severe environment.

What is Ruggedness?
Ruggedness refers to the ability of the RF power transistor to withstand load mismatch conditions under high output power conditions without experiencing device failure or measurable long-term degradation in device performance. Under mismatched load conditions, a large amount of power can be fed back into the active device where it is dissipated in the semiconductor. The ability to handle the large power dissipation internally in the active area without altering the performance is indicative of a reliable device.

Although there are no standard metrics to define ruggedness, the ruggedness of a specific transistor is typically a function of the magnitude and phase of the mismatch, the output power level conditions, and the thermal dissipation properties of the amplifier. Unfortunately, semiconductor device manufacturers use different criteria to measure ruggedness. Some define the failure criteria for the ruggedness test as no degradation in output power. This technique is not an adequate definition for failure, however, since some technologies, like bipolar, can deliver full rated output power even when the ruggedness test damages some of the cells. The remaining cells simply provide the full power at a higher temperature, which results in a decrease in the MTBF of the device.

Other power device manufacturers specify failure criteria as a more than 20% shift in one or more of the DC test parameters. A few semiconductor power device vendors do not specify a ruggedness rating at all. Some LDMOS manufacturers, on the other hand, measure ruggedness during what they call a “burn-in” test which acts as an accelerated life test, causing a huge change in the DC parameters after a single trial. This minimizes the performance drift over the operating lifetime of the active device in the field.

Figure 1 depicts a typical RF transmitter chain. In a communication system, the information to be transmitted is processed in the baseband using DSP and analog logic circuits. In order to transmit this information in an efficient manner, the data is modulated with an RF carrier frequency. The local oscillator is used to mix the required information up to the RF frequency. The power amplifier (or power amplifier chain) transmits this signal into the air through the antenna, with the output power of the final stage amplifier and the antenna gain determining the range that the information is broadcast. A power amplifier output matching circuit transforms the output impedance of the transistor to the antenna impedance in order to optimize the performance of the active device. The antenna component, however, is exposed to uncontrolled environmental conditions so that the antenna impedance will vary. Similar issues affect radar transmitter ruggedness.

Test Basics
Figure 2 shows a standard RF bench and the equipment needed to measure the RF performance of a power amplifier. During device characterization, an industry standard matched load of 50 ohms is presented to the output of the power amplifier. When performing the ruggedness stress test, the RF switch changes the load from the matched case to that of a shorted load connected to a line stretcher. The line stretcher component varies the phase of the shorted load over a full 360 degrees, simulating any mismatch condition that could occur in real world applications.

There are three controlling electrical factors presented to the power amplifier in a ruggedness stress test. They are the amount of input power, the dc bias supply voltage and the load presented to the device. Temperature can also affect ruggedness, but it is first order independent of the three electrical factors. Any of the electrical factors can be varied, although many semiconductor vendors only vary the mismatch seen by the load and use the nominal value for RF power and bias voltage. The semiconductor may indeed prove rugged under these set conditions. However, a true test of ruggedness will vary all of these factors at the same time to accurately mimic real world conditions and see if the device still maintains performance after the ruggedness test.

Under linear conditions, a perfectly matched load is able to convert the entire applied RF signal into transmitted RF power. A mismatched load will transmit some of the RF signal, but will reflect the remaining power back into the device. This reflected power must be absorbed internally in the transistor. The voltage reflected is measured by the reflection coefficient gamma where the magnitude of gamma normalized to unity relates the magnitude of reflected voltage (i.e. gamma of zero implies no reflection and perfect transmission, while a gamma of one implies total reflection). Another parameter used to express the amount of reflection and characterize the type of mismatch is the voltage standing wave ratio (VSWR). It is directly related to the magnitude of the reflection coefficient as seen in Equation 1.

VSWR is the most commonly used figure of merit when discussing ruggedness testing. The higher the VSWR, the more power that is reflected back to the load of the amplifier. That power must be absorbed by the active device without damage. A VSWR of 5:1 implies that nearly half of the desired output power is actually reflected back to the amplifier. A VSWR of 10:1 corresponds to more than 2/3 of the reflected power. Most semiconductor companies specify the ruggedness of their devices with one of these two VSWR conditions. A VSWR of 20:1, which reflects more than 80% of the output power back into the device under test, would represent a truly superior level of ruggedness.

Modifying VSWR Values
During a ruggedness stress test, a load with a certain VSWR is presented to the device and the phase of this load is varied over an entire period. The phase of the load is modified by a line stretcher (see Figure 2). Different values of VSWR are obtained by using a short circuit, having a magnitude of reflection of 1, and an attenuator. A lossy coaxial length of line acts as the attenuator. An infinite VSWR is achieved by terminating the load with a short and no attenuation. Real world losses in the output coupler, RF switch and line stretcher attenuate the magnitude of the reflection coefficient by a few percent, generating a VSWR greater than 20:1 but less than the theoretical value of infinity. Varying the length of coaxial cable will vary the VSWR presented to the load. The shorter the coaxial cable length, the smaller the attenuation, producing higher VSWR. The length of the cable is determined by the properties of the cable, the desired VSWR and the frequency of interest.

Test engineers can achieve different VSWR values using different line lengths. Theoretically, the VSWR will present a perfect circle on a Smith Chart as it is rotated through all 360 degrees of phase. In the real world, however, there are losses in the coaxial cable and therefore, we expect the VSWR to change as the line stretcher varies the phase of the device through all 360 degrees.

Figure 3 depicts the impedance presented to the drain lead of the device as the phase is swept from 0 to 360 degrees for each VSWR condition. As the phase is varied, the device sees the impedance indicated by the VSWR circle. Generally, when the impedance presented is less than the nominal impedance the device will operate in a low voltage/high current condition. If the high current generates enough heat, thermal failure can occur due to the silicon melting. If the impedance presented to the device is greater than the nominal impedance, the device will operate in a high voltage/low current situation. When the high voltage condition exceeds the breakdown voltage rating, the device goes into avalanche breakdown.

DC and RF performance
To test the ruggedness of a device, it is internally matched to bring the impedances up to a practical level, then further matched to 50 ohms on the PCB board. All device performance data is generated in an evaluation circuit which is at the 50 ohm reference plane and includes all external matching circuitry losses. This environment closely resembles the actual losses experienced in a matched, real-world amplifier application.

Using the standard RF bench set-up described in Figure 2, the HVVFET device was baseline characterized, then subjected to a ruggedness stress test. If the ruggedness stress test was damaging the device, one would expect the DC parameters to shift. During the stress test, the input power is set to the amount required to achieve rated output power, the bias is set at a nominal voltage, such as 48 volts, and the load is set to a mismatch with VSWR of 5:1 and swept through all 360 degrees of phase. The device is retested under nominal conditions for input power, voltage and load and compared to the baseline data. Any measurable shift in performance indicates damage and constitutes a failure. The load is changed to the 20:1 VSWR cable length and re-tested with the ideal load for the baseline data. The device is then subjected to an over-voltage condition where the nominal bias supply voltage is increased by 10% to 53 volts. Next, it is subjected to nominal input power drive. If the device passes the over-voltage condition and the high VSWR mismatch over all phase angles, it is next subjected to twice the amount of input power needed to attain rated output power, which is 3dB worth of overdrive.

To ascertain the impact of each test on performance, DC and RF data is measured prior to testing to create a data baseline. After each test condition in Table 1, DC and RF performance is measured to determine the impact of the ruggedness test. Engineers can set up simple pass/fail criteria specified as no measurable performance change in the data. Historically, ruggedness testing increases device channel resistance (Rds(on)), decreases minimum gate voltage (Vgs(th)) and increases drain leakage current (Idss). Ideally DC parameters are stable and unchanged both before and after the ruggedness testing. The passing criteria for these devices is no change which is measured as less than 1% difference from the pre-test and post-test DC and RF data. The HVVFET device is truly rugged, able to simultaneously withstand over-voltage, input over-drive and high mismatch load conditions with no change in any measured DC or RF parameter ensuring a long life cycle for every RF power application.

Conclusion
Today’s RF power amplifiers are expected to deliver high levels of performance over increasingly lengthy life cycles. As arguably the single most important indicator of device reliability, comprehensive ruggedness testing promises to play a key role in determining the true long term performance capabilities of these important devices.

HVVi Semiconductor, Inc.
www.hvvi.com
TXTLINX.COM91
Email this article to a friend!