EDGE Power Amplifier Distortion Measurements in a Large Signal Polar Modulation (LSPM) System
By David Zhao: Department of Application Engineering & Jason Zhou: Department of Customer Quality Engineering, RFMD®

This paper reviews a practical step-by-step procedure on how to perform a large signal polar modulation (LSPM) EDGE power amplifier distortion test. First, the theory of AM-AM and AM-PM distortion in a LSPM system is discussed in detail. Second, the method on how to test EDGE PA distortion on AM-AM and AM-PM by using a vector network analyzer is analyzed.

This paper will also include an automatic test program on distortion testing. Using this test program, all test data can quickly be processed and analyzed.

Pulsed RF signals and triggered measurements–two methods in simulating GSM transmit signal – are also presented in this paper.

Theoretical Background Large Signal Polar Modulation (LSPM)
Polar modulation is desirable for two primary benefits:

- It provides reduced power consumption by use of a saturated power amplifier for non-constant-envelope modulation (phase and amplitude) — offering approximately a 20—25% reduction in power.

- Low noise outputs eliminate SAW filters in the transmit (TX) path while supporting non-constant-envelope modulation.

The concept is to pre-distort the desired signal to compensate the distortion caused by the non-linearity of the saturated power amplifier. RFMD® introduces LSPM architecture in its POLARIS™ TOTAL RADIO™ product family as shown in Figure 1. In this system, characterizing the power amplifier distortion is a key factor to ensuring proper operation in the transmitters.
Figure 1 shows the transmit architecture used in RFMD’s POLARIS 2 TOTAL RADIO Module, or P2RM.

Polar Modulation PA Distortion
There are two kinds of PA distortion operating in saturated mode in LSPM system. One is AM-PM distortion and the other is AM-AM distortion.

Saturated PAs are very non-linear due to their low bias points and higher efficiencies, which cause AM-PM distortion in the amplifier. AM-PM is described as the S21 phase versus input control voltage of the PA (VRAMP).

Figure 2 shows a graphical concept of AM-PM distortion.

The polar modulation PA exhibits a large variation of output phase, _(out), versus POUT with different input drive levels and temperatures. An important objective for a PA designer is to maximize AM-PM flatness and minimize AM-PM variation over temperatures and voltages. The PA AM-PM distortion over temperature should be checked and verified in a lab setting to determine the transmit performance over temperature.

AM-AM distortion is described as the non-linear transfer function between the output RF voltage versus the input control voltage (VRAMP). Pre-distortion is used to correct for the non-linear characteristics of the power amplifier. Figure 3 shows a graphical concept of AM-AM distortion.

Polar modulation PAs operate within the nonlinear VOUT versus VRAMP area.

In the next section, a way to test AM-PM and AM-AM distortion is discussed.

Measurements and Analysis of Pulsed RF Signals
Most of the GSM/EDGE PAs are not designed to operate at 100 percent duty cycle. Therefore, testing a PA with 100 percent duty cycle signal does not reflect the PA’s real operating conditions. Running a PA in 100 percent duty cycle would cause thermal overstress and burn the device. Therefore, pulsed RF signals to test the DUT is needed. The testing method discussed in this paper is designed to simulate GSM/EDGE PA 1/8 duty cycle operating conditions. For pulsed signal testing, it is important for the vector network analyzer (VNA) to be aligned accurately with the pulsed signals. An external trigger for this testing is needed.

Triggered Measurements
Agilent E5071B is used in this test as the VNA. (E5071B is one of the Agilent ENA series.) There are two event triggers available in the VNA. One is a Triggering on Sweep and the other is Triggering on Point. The network analyzer is usually triggered on sweep mode.

In the sweep mode, VNA is triggered to start a frequency scan immediately. But in the test system discussed here, the VNA cannot finish a thorough scanning for the GSM working band during one burst time, which is approximately 577us. This is the difficulty that needs to be resolved in the testing setup.

The VNA trigger signal comes from the AWG430 instrument and also works as a PA transmit enabling signal. The first step is to set the VNA trigger source to “external” instead of “internal.” The second step is to set the trigger event on “point” instead of on “sweep.” As stated above, VNA cannot finish the scan through a GSM working band within 577us, so it needs to scan for single frequency on a trigger event. The third step is that a proper trigger delay should be selected.

Due to the fact that the EDGE PA works on burst mode, the VNA should start testing accurately following transmit (TX) burst time. A time alignment between VNA testing and TX EN should be achieved prior to when proper PA transition parameters in the VNA are fetched. The time delay should be adjusted manually to ensure the VNA is triggered in the middle of the burst, before any testing is conducted.

Measurement Setup
In order to validate the theory above, the RFMD® RF3161 polar mode EDGE PA is tested and measured for verification. The test setup is as shown in Figure 4.

The test setup includes an AWG430 as the trigger source and PA logic control. The AWG 430 Port1 outputs PA enable logic and Port2 outputs PA VRAMP logic. The AWG 430 can be programmed to control the VRAMP output amplitude. The VRAMP is set from 0.2V through 1.6V.

Figure 5 shows the control logic for the PA and the trigger source for VNA. The yellow line represents the PA TX Enable and Trigger. The green line represents the VRAMP .

A power meter is employed in the test setup to calibrate VNA output power and to ensure that the PA input power is accurately calibrated. The VNA can be used to test DUT transition parameters, but it is not capable of showing accurate PA absolute output power. PA absolute power testing is based on the theory of PA input power plus PA gain equaling PA output power. In the test setup, VNA performs a frequency scan in the specified span in an external trig mode, such as from 880MHz to 915MHz. All traces are displayed in the ENA and are recorded and transported to an Excel spreadsheet for automatic analysis.

Thus, the computer is the controller in this whole test system. The AWG430 will be controlled to scan at a specified range from the lowest level to the highest level, step by step. At each step, the VNA performs a scan across the specified span. All amplitude response and phase response information is recorded by the VNA, and is then transported by General Purpose Interface Bus (GPIB) in to an Excel spreadsheet.

A 3dB attenuator is employed between the VNA output port and PA input power for stable impedance matching. A 20dB attenuator is also employed at the PA output port to gain better matching and prevent the VNA from being damaged by RF power that is too high.

Distortion Measurements
S21 transition parameters in formats of log magnitude and phase are shown in Figure 6. From log magnitude data, S21 gains of the PA at different frequencies are obtained. From phase data, phase variations of the RF signals at different frequencies are obtained. When adjustments to the value of VRAMP voltage scan from the lowest to the highest, PA output power will change accordingly.

In the meantime, the AM-AM and AM-PM distortion will follow the change. To use GPIB commands to control the AWG430 and transport all VNA displayed traces data to computer to complete an automatic testing. To use the same testing method, other S-parameters can also be achieved automatically, i.e. S11, S22, etc.

Data Analysis and Synthesis
All data obtained from VNA will be transported to an Excel spreadsheet. A concise VBA program or VB program is developed to chat based on collected data. Figure 7 shows the relevancy of VRAMP and POUT. Figure 8 shows the relevancy between POUT and Phase.

So far, all the PA distortion specifications at different frequency points have been measured. The obtained specifications data will instruct transmitter debug or polar PA design. It’s very easy for this test system to get more distortion information if more sweep points within the span in VNA are settled.

Conclusion
The EDGE PA distortion testing method discussed in this paper is an easy way to implement. By using an automatic test program, test data collection under different test conditions combined in one spreadsheet becomes practical. This testing is a simple and powerful basis for insight into the distortion behavior of a LSPM EDGE PA and can give engineers a good reference for engineering analysis.

RFMD®
www.rfmd.com
TXTLINX.COM103
Email this article to a friend!