Application of EDGE power amplifier in mobile phone

In the GSM system, EDGE can be said to further increase the data transmission rate. Through the change of modulation method, coding and multiple transmission time slots, the transmission rate can be tripled. Since the formulation of the EDGE standard in 1999, the EDGE network has been adopted by many countries and their telecom operators. According to the latest statistics of the Global Mobile Suppliers Association (GSA), 307 devices with EDGE functions have been released. According to statistics and estimates by market research organization Strategy Analytics, the EDGE mobile phone market in 2006 was about 160 million units. Between 2005 and 2010, the EDGE/WCDMA mobile phone market will have a significant growth of 51% compound annual growth rate (CAGR).

EDGE RF solution - linear transmission architecture

There are currently three EDGE RF solutions available on the market for mobile phone manufacturers to choose from, which can support 8PSK mode in addition to GMSK mode. These three solutions are Polar Modulation, Polar Loop and Linear EDGE architecture.

As for the linear transmission architecture, the power amplifier used must be able to operate in saturated mode and linear mode. When the mobile phone operates in GSM, it uses GMSK modulation, and GMSK is a constant envelope. The distortion generated by the power amplifier has less impact on it. Therefore, the power amplifier can be operated in the saturation region, that is, the nonlinear region, to improve efficiency. When the mobile phone operates in EDGE mode, it uses a linear modulation method that changes the amplitude and phase, that is, 8PSK modulation. Therefore, the linearity of the power amplifier is extremely required to prevent signal distortion.

Multi-mode operation─GMSK

Mixed with 8PSK transmission

EDGE uses the TDMA time frame structure, so when there are multiple transmission time slots and mixed transmission modes, the power amplifier will have different operating modes, i.e. 8PSK switching to GMSK or GMSK switching to 8PSK. The power generated by the power amplifier must be minimized between time slots (Burst) to avoid deterioration of the output RF spectrum or non-compliance with ETSI specifications. Therefore, the input and control signal timing (Control Timing) and signal size between time slots must be standardized and followed, as shown in Figure 3. The following is an example of using RFMD linear power amplifier RF3158 ​​with 3 transmission time slots, GMSK→8PSK→GMSK.

The waveform of the signal generator output

First, use a signal generator to generate three time slots. Take Agilent Signal Generator E4438C as an example. The settings are as follows:

1. Mode→ EDEG mode.

2. Data Format → Framed.

3. Frame Trigger → Continuous.

4. Configure Timeslots→ Multislot off, TS=TSC0 and set the timeslot. Normal means this timeslot is 8PSK modulation.

5. Output power = 2dBm.

At this time, connect the signal to the spectrum analyzer and observe it with zero span to see the set signal. Since the output signal generated by the signal generator is the input signal of the power amplifier, that is, RFin, in theory, in GMSK mode, the shorter the waveform rise time, the better, while in 8PSK mode, it is required to rise smoothly so as not to affect the output RF spectrum (ORFS-Output Radio Frequency Spectrum). When adjusting the output waveform (burst shape), Agilent E4438C adjusts the rise time, rise delay, fall time, or fall delay based on the wave set to be transmitted within a GSM time frame. The EDGE and GMSK mixed signal

Simulation and Experiment

After setting up the signal generator, connect other devices to the RF3158 ​​evaluation board. Use EVEN 1 of the signal generator as the trigger signal of the arbitrary waveform generator, load the edited waveforms of Tx_Enb, Vramp and Vmode into the arbitrary waveform generator and connect it to the evaluation board. In order to easily observe the timing relationship between signals, that is, to display RFin, RF Out, Tx_Enb, Vramp and VMode on the oscilloscope at the same time, convert the RF Out and RFin radio frequency signals of the power amplifier into electrical signals through the Video Out of the spectrum and display them on the oscilloscope. It is recommended to use VMmode as the external trigger signal of the oscilloscope, or connect VMmode to the Ext Trigger in of the oscilloscope to increase the port usage of the oscilloscope. After completing the signal setting and connecting the instrument, you can turn on the power and signal in sequence.

The timing relationship between power amplifier mode conversion and input signal

When the linear EDGE power amplifier works in GSM mode, the power amplifier works in saturation mode. At this time, Vramp controls the collector voltage of the power amplifier transistor to make the output waveform and power size meet the required requirements and various ETSI specifications. When switched to EDGE mode, the power amplifier works in linear mode. At this time, the collector voltage of the power amplifier transistor is fixed at 3.6V, and Vramp provides the base bias of the power amplifier transistor to control its bias current, so that the power amplifier works in the linear region, like a gain block, and the input RF signal and output power have a linear gain relationship.

The RF3158 ​​supports 50% of the transmission duty cycle of GPRS Class 12, which means that two mixed mode time slots may be transmitted at the same time. Therefore, the power amplifier must complete the mode conversion between the two time slots, which is the guard period. This conversion time can be called the settling time.

When VMode changes from High to Low, it means that the power amplifier switches from the linear mode of 8PSK to the saturation mode of GMSK. At this time, RFin needs to be reduced to the lowest input power below -40dBm (recommended value), and Vramp needs to be reduced to about 0.3V. Turning off Tx_Enb for 1QB (Quarter Bit, 1QB is about 0.92us) helps to shorten the settling time. The settling time is caused by the low-pass filter in the power control loop and the Vramp pin, and turning off Tx_Enb can provide a discharge path.

When we turn on RFin 2QB after VMode turns to Low, we can clearly see a spike in the protection period between 8PSK and GMSK. It can be seen that during the mode conversion, the power amplifier does not complete the stabilization time and does not reduce the RFin signal to <-40dBm or input the RFin signal, which will generate a spike and cause the output RF frequency spectrum (Output Radio Frequency Spectrum-Spectrum due to switching transients) to deteriorate and even fail to pass the specification.

In addition to the timing relationship of the input signal, another factor that affects the transient spectrum of power conversion is the signal size of RFin during the protection period. In this experiment, another signal generator can be used to increase the signal size of RFin during the protection period to test how low RFin should be during the protection period to avoid causing a surge. Figure 20 shows the original signal. The signal size of RFin during the protection period is -74.32dBm. At this time, an additional signal generator generates a continuous signal, which is combined by a combiner and then input into the power amplifier. Figure 21 shows the result of the combination of the two signal generators. Through this experiment, it can be known that the minimum output power generated by the transceiver should not exceed -33dBm.