Optimize RF PA efficiency to increase your handset's talk-time

A mobile handset isn't just a phone anymore, it's an MP3 player, a fashion statement, a GPS device, a calorie counter, a camera, a data modem, a TV, an email client, and the list goes on. Cellular providers have increased the services bundled with the 3G phones to increase the average revenue per subscriber (ARPU). At the same time, there's an expectation to increase the talk time and battery life with the same or slightly higher capacity batteries. This makes the system design challenging.

System designers must be very cautious and do a power survey of every component on the phone's pcb. The RF power amplifier (PA) powered directly from the battery sticks out like a sore thumb from the power budget perspective.

Cellular and WLAN standards have been evolving with data transmission speeds that started from 14.4 kbits/s in CDMA-1 to 2 Mbits/s in CDMA2000/WCDMA. Modulation schemes used in CDMA and WCDMA result in a modulated signal that exhibits a non-constant amplitude envelope.

The need to utilize the available spectrum efficiently and transmit at high data rates necessitates the use of modulation schemes that use both amplitude and phase changes to encode information. Consequently, PAs must operate with linear gain and phase relationships from input-to-output; these linear PAs inherently operate at lower efficiency as compared to saturated PAs in older/slower transmit schemes such as GSM. This efficiency decrease is primarily due to the quiescent operating point being backed off from the PA's compression point.

The amount of back-off is determined by the complementary cumulative distribution function (CCDF) of the transmitted signal which in turn depends on the wireless standard and the number of active RF channels being transmitted by the handset. When the handset is operating in transmit mode, the RF power section consumes up to 65% of the overall power budget as a result of the PAs intrinsic inefficiencies.

RF PAs used in WCDMA cellular standards have been traditionally powered directly from the battery. This maintains the linearity of the RF PA at the expense of PA (and system) efficiency.

The requirement for linear PAs in systems that have both amplitude- and phase-modulation components require system designers to balance linearity and efficiency requirements in new handset designs. The use of dc-to-dc switching converters to supply the optimal voltage to the RF PA brings about a 50% to 200% improvement in average PA efficiency based on typical RF power transmit probability distributions for WCDMA systems. This improvement has been observed for commercial RF PAs with different architectures such as those with either switched gain stages or conventional Class A/AB multi-stage designs.

For this reason, linear PAs powered by magnetic buck converters can dramatically increase system efficiency. Overall efficiency (OE) or power-added efficiency (PAE) of the PA is a key performance metric of a power amplifier.

OE = RF_Pout /(Pdc+RF_Pin)

The key in using a switcher unit for a PA (SUPA) is to reduce the Pdc factor in the denominator for a given RF_Pout and RF_Pin while maintaining linearity (ACPR/ACLR specs). The supply voltage can be changed with an SUPA to a level that will ensure that the PA's gain doesn't decrease significantly and the amplifier operates in a linear manner in response to the input signal and doesn't compress the output signal.

A quantitative measurement of linearity for a modulated signal is the adjacent-channel-power/leakage ratio. PA gain is typically specified under specific channel and RF_Pout conditions. The change in gain with respect to the amplifier's supply voltage is specific to the PA manufacturer; while PA manufacturers generally don't provide these curves in their datasheet, designers can characterize this or request the data.

When the PA is connected directly to the battery, Pdc = Vbatt×Ibatt and when it's powered by a SUPA, Pdc = Vo×Io. It can be seen that by increasing the PA's overall efficiency, all that's needed is a low Vo compared to Vbatt. This is achieved by lowering the SUPA's output voltage at lower transmitted RF power levels. This in turn reduces Io (the current drawn by the PA) and results in a lower input current drawn from the battery due to the dc-to-dc converter's inherent high efficiency. Note that behaviorally, PAs can be modeled as RF_Pin and Vcc_PA controlled resistors.

It's important to consider the power probability profile of transmit power to understand the impact of savings in powering a PA with a SUPA (Fig. 1). The Tx power probability is dependent on the wireless standard and carrier system design/usage. The profiles are different for urban and rural regions.

1. The PA transmits low power levels for a good percentage of time in a typical handset, which reinforces the savings that are possible with a SUPA. Also shown is the battery power consumed with and without a SUPA.

The dc-to-dc converter's output voltage must be varied as the transmitted power levels are changed to maintain the ACPR specs (Fig. 2). The savings in battery current can be as high as 50 mA in the 0- to 20-dBm power levels. The PA is most often operating in this band of power levels.

2. Shown is the battery current savings when the dc-to-dc converter powers the PA (top). The percentage savings in power or battery current is now shown when the PA is powered by a SUPA (bottom).

A good question to ask is why we have to change the dc-to-dc converter voltage as the transmitted power level is increased. The answer is that this is needed to maintain the ACPR ratios. Adjacent channel power/leakage ratio helps characterizes the PA's distortion and other subsystems for their tendency to cause interference with neighboring radio channels or systems. It's specified as the ratio of the power spectral density (PSD) of the main channel to the PSD measured at several offset frequencies. Violating the ACLR specs will lower the voice quality, as well as decrease wireless system capacity.