Global Sources
EE Times-Asia
Stay in touch with EE Times Asia
EE Times-Asia > RF/Microwave

Quenching the thirst of RF power amps

Posted: 16 Nov 2005 ?? ?Print Version ?Bookmark and Share

Keywords:wireless? rf? power? management? mobile?

Battery life or runtime is crucial in portable wireless systems such as cellphones, PDAs and laptop computers because it ultimately defines the devices mobility. With decreasing form factors and increasing functional densities, battery life is in short supply, decreasing the practical nature of these mobile electronic solutions. Within a wireless system, the RF power amplifier (PA), in fact, consumes most of the energy because it must drive the antenna with sufficient power to transmit signals across space to remote receivers. Hence, the battery life of a wireless system is strongly dependent on its PA and the efficiency of the PA should be as high as possible, but not at the cost of linearity.

To achieve high efficiency and linearity in RF PAs, two approaches can be undertaken:

Linearize an efficient, nonlinear (switching) PA with feedback control;

Increase the efficiency of an inherently linear PA.

In the first approach, the bandwidth of the control loop must be significantly higher than the linearizing signal, be it the envelope or the complete RF signal, to process the information fast enough to truly follow the signal. Understanding that technology is already being pushed to its limits and HF operation necessarily implies higher switching losses, linearizing a PA inevitably decreases the maximum bandwidth of the PA to well within the limits of a given process technology. Consequently, increasing the efficiency of linear PAs may be a more tenable approach for improving battery life without significantly degrading linearity.

Several wireless technologies have been developed and are widely used nowadays, such as GSM, CDMA IS-95, W-CDMA and ieee 802.11a/b/g. The demand for local wireless applications has grown and 802.11a/g signals have gained popularity. Unfortunately, higher spectral density and consequently higher linearity are the side-effects of this trend, as seen by the high bandwidth, high peak-to-average power ratios (PAPR) and high propensity for above-average power levels required by 802.11a/g signals. As a result, the average efficiency of conventional and good-quality PAs in this environment degrades because not only are linearity requirements increased, but so are signal and envelop bandwidths, all of which incur efficiency trade-off losses.

Before blindly working on designing adjustable supplies, the power probability distribution of the RF signal must be considered because it defines how battery life can best be improved. The figure shows an example of the power output probability distributions for 802.11g signals, illustrating the tendency for portable systems to mostly operate in light-to-moderate power-level conditions, not high-power modes. The maximum output power may be 25dBm, but the most probable value is approximately 16dBm, which is where the PAs efficiency is significantly lower, and this is similar for 802.11a/g, CDMA IS-95 and W-CDMA signals. Consequently, the efficiency of PAs in light-to-moderate loading conditions is critical for operation life.

There have been several techniques proposed to improve the average efficiency of linear PAs and most notable are the dual-PA (Doherty), envelope-following and power-tracking schemes. As justified earlier, linearizing switching PAs with negative feedback is discounted in this foregoing discussion because of its inherent bandwidth (speed) limitations.

In the dual-PA scheme, one amplifier sources the low-to-moderate load power levels, while the other sources the above-average power range. Only one PA is thus active and optimally designed for the critical light-to-moderate load region, where efficiency is normally lost. At higher power levels, both PAs can continue to operate in their gain-compression regions, thus maintaining peak power efficiency. Unfortunately, the power divider and combiner required are lossy, especially when integrated on-chip, and designing two PAs may not be entirely attractive in terms of cost.

Avoiding the dual-PA paradigm implies adding intelligence to the supply voltage and, in the case of the envelope follower, forcing the supply to follow the envelope of the transmitted RF signal. The envelope-following supply is thus kept slightly above the actual peak of the signal to allow the PA to fully process the RF signal, envelop and phase, which is how linearity is maintained. Since the difference in supply and signal peak voltage is kept low, minimal power losses are incurred by the PA.

To add the intelligence to the supply is to design a dynamically adaptive supply circuit, which unfortunately also incurs power losses, but hopefully less than the PA drain losses just saved by the envelope-following scheme. Switching power supplies are therefore viable solutions, given their propensity for high efficiency. Their efficiency, however, is ultimately limited by their switching frequencies, which is why RF signals with low envelope bandwidths like CDMA and W-CDMA signals benefit the most from this scheme, unlike higher spectral-density signals such as those of 802.11a/b/g.

As envelope bandwidths increase, another dynamic scheme must be implemented, which is how average power tracking finds its niche. The bandwidth required to track the average power of high-bandwidth envelopes is lower, given the nature of the averaging function, which is attractive for switching power-supply circuits within the context of power efficiency. Since the supply is now at an average level, the PA incurs the power losses the envelope-following scheme saves for low output power levels, but significant savings are still achieved over the conventional fixed-supply scheme.

In tracking the average power, high PAPR signals are clipped and distorted, which is why this method is most attractive for low PAPR and infrequent above-average power level applications, such as those of CDMA, W-CDMA and 802.11 b, but not 802.11a/g applications. The few and low power signals that clip ultimately translate to acceptable bit-error rates and, in the case of 802.11a/g signals, bit-error rates would increase significantly.

Following the envelope is not the most attractive solution because of its bandwidth requirements and supply-efficiency limitations. Tracking average power, meanwhile, seems to be the best solution, scoring high on integration and complexity. In the case of highly probable, high PAR, spectrally-dense signals like 802.11a/g, the dual-PA approach may be the only viable option, in spite of its relative complexity and integration limits. Note that completely adjusting the biasing conditions of any dynamically adaptive supply scheme improves efficiency, be it an envelope or power-tracking scheme.

Gabriel A. Rincsn-Mora, Sr. Member

Hsuan-I Pan, Student Member

IEEE georgia tech Analog and Power SoC IC Design Lab

Article Comments - Quenching the thirst of RF power amp...
*? You can enter [0] more charecters.
*Verify code:


Visit Asia Webinars to learn about the latest in technology and get practical design tips.

Back to Top