hatever emerges as the topology for an all-Internet Protocol wireless network, the groups developing the infrastructure and basestations for such networks will face powerful challenges. And one of the biggest of these will be the type of power amplifiers to deploy in next-generation cellular basestations to cope with higher power and frequency ranges.

New standards requiring significantly higher data rates and complex modulation schemes demand amplifiers that handle higher peak-to-average power ratios and offer higher levels of signal purity.

Emerging multichannel power amplifier technology, which can significantly reduce system costs, places its own demands on linearity, since the MCPA needs to handle rapid envelope fluctuations among multiplexed inputs. Such demands land squarely on the output power transistor.

As wireless networks shift toward higher frequencies, the power ratings of available transistors tend to drop off. This is particularly true for silicon-based transistors like laterally diffused MOSFETs (LDMOS).

There are of course major issues with LDMOS as the technology of choice for high-power, high-frequency power amps. Its frequency response is limited by gate-charging effects and the transmit time needed for carriers to make it through the n-drift region. Manufacturers are also striving to get costs down (dollars per watt) and increase power density (watts per square inch): the target for the latter is about 300 W/inch2.

Main suppliers include Motorola’s Semiconductor Products Sector (Phoenix, Ariz.); Philips Semiconductors (Eindhoven, Netherlands); UltraRF (Sunnyvale, Calif.); Ericsson Microelectronics (Richardson, Texas); M/A Com (Lowell,Mass.); and Xemod (Santa Clara, Calif.).

As of today, the group is joined by a new name in the industry, Celiant Corp. (Warren, N.J.), which is a spinout of Lucent Technologies and Bell Labs to sell LDMOS-based power amplifiers and RF subsystems to the merchant market.

While LDMOS certainly has many practical advantages, many in the industry believe it has reached its limits, and will be struggling with the needs of power amplifiers at frequencies much above 2.2 GHz. "We desperately need some practical and cost-effective technologies to emerge. LDMOS is holding the whole thing back. We also need to get back to feedback technologies," said Steve Cripps, an independent consultant based in Somerset, U.K., who specializes in component technologies for wireless-infrastructure equipment.

The latest offerings from the entrenched power amp suppliers are addressing some of the drawbacks with LDMOS, and are emphasizing that improvements mean they can cope with the higher specifications and efficiencies needed for third-generation networks. For instance, Xemod recently demonstrated LDMOS-based power amp modules capable of handling 350 W and aimed specifically at wideband code-division multiple-access applications. And UltraRF has released a part that increases the RF envelope of LDMOS above 3 GHz. The most recent offerings also push the efficiency and linearity performance of the company’s high-power LDMOS transistors.

After extending the cutoff frequency of its second-generation LDMOS power transistors to above 3GHz, Philips Semiconductors is now focusing on slashing the drift characteristics on its latest parts by minimizing hot-carrier injection and shielding the transistor’s gate and drain sections.

And not surprisingly, numerous projects are under way to develop novel LDMOS-based devices that will yield both higher impedance levels and power density.

For instance, researchers in Sweden from Ericsson Microelectronics (Kista) and Uppsala University (Uppsala) described at the recent European Microwave Conference a new LDMOS transistor concept with a dual-layer extended-drain region. This is said to shield the active-gate region from high voltages, and thus overcomes the fundamental incompatibility between short channels and high-voltage operation.

Such enhancements to LDMOS technology will need to meet the challenge from companies working on quite different, wide-bandgap semiconductors as the material of choice for high-frequency/high-power combinations in power amplifiers. The leading short-term candidate here is gallium nitride, and the company leading the attack is Nitronex (Raleigh, N.C.).

The company has started shipping to selected basestation manufacturers GaN-based power amplifiers that, crucially, were made on 4-inch silicon substrates using a proprietary transition layer and patent-pending epitaxial growth and deposition technology. Initial samples of the high-electron-mobility transistors range from 8 W to 35 W of peak power at frequency bands from 2 to 3 GHz. The parts operate from a 28-V drain supply.

At Cree Inc. (Durham, N.C.), there is a three-pronged effort in the development of power amps for basestations. Cree, too, is working on GaN, but also has a foot in the LDMOS camp following the acquisition last November of UltraRF. Cree’s main thrust, though, is with the silicon carbide (SiC) technology it is pioneering. Cree has already produced a Class B FET capable of generating about 14 W of power output at 1.95 GHz, with 11-dB power gain. The part is said to offer outstanding adjacent-channel power ratio in RF power amplifier designs.

John Palmour, director of advanced devices at Cree, said the SiC-based devices hold the most potential for the high-frequency, high-bandwidth needs of power amplifiers in next-generation wireless infrastructure.