DESIGNERS of power amplifiers have several challenges they face when designing high-power, broad-bandwidth amplifiers: low load impedances and the presence of performance-impairing parasitics.
The former challenge stems from an output load line impedance that is naturally low, and falls off with increasing power. To address this, designers craft matching circuits that interface to 50 Ω. The bandwidth, efficiency, and complexity of the design are then governed by the ratio of the impedance of the 50 Ω interface to that of the transistor. One solution to this challenge is to use the smallest device with intrinsically higher impedance such as a transistor that operates at higher operating voltage. For example, at 100 W increasing the voltage from 12 V to 48 V results in a significant increase in load impedance, in this case from less than 1 Ω up to 12 Ω. It is a squared relationship, so quadrupling the operating voltage raises the load impedance 16 fold. This higher impedance allows for much simpler matching networks but also allows the potential for much wider bandwidth designs.
The second challenge – parasitics – is particularly problematic when designing wideband power amplifiers. At lower frequencies, resistive elements of the FET-equivalent circuit tend to dominate; but at higher frequencies they are overtaken by the effect of input shunt capacitance between the gate and source and the output shunt capacitance between the drain and source, which together drive down impedance and raise the quality factor, Q. Put another way, the impedance becomes more reactive and rotates around the Smithchart, i.e it varies with frequency.
A very attractive option for addressing both of these challenges is to turn to GaN transistors for power amplifier designs. Compared to the incumbent technologies of GaAs at high frequencies and silicon LDMOS in the lower bands, GaN naturally supports operation at higher voltages, allowing more power at higher frequencies. Additionally, GaN’s higher power density allows for physically smaller devices than their legacy counterparts, which results in both lower intrinsic high-frequency effects, and lower input and output capacitances. The resulting lower Q characteristics of GaN allow for previously unobtainable wideband amplifier designs capable of covering multiple octaves at very high power levels.
Solid-state RF energy offers numerous benefits unavailable via alternate solutions, powering applications like Medical Ablation
Virtues of GaN power amplifiers are not limited to a wider bandwidth. They include an efficiency exceeding 70 percent; a ground breaking power density; and a capability to operate reliably at higher temperatures than other semiconductor technologies, which again allows for physically smaller devices. Combining these assets with an electron mobility that is similar to GaAs enables the construction of transistors delivering high power at high frequencies, while sporting far smaller parasitics than their rivals.
GaN enables a constant output power, zone-controllable heating and a ten-fold hike in oven lifetime.
Due to these strengths, it is of no surprise that GaN is steadily becoming recognized as the optimal solution for mainstream applications in many different markets. Amplifiers made from this wide bandgap semiconductor material are expected to proliferate into every aspect of commercial and consumer RF power markets.
In defence communication and electronic warfare, where the emphasis is on performance, GaN has rapidly displaced LDMOS. This explains why this sector led the way in embracing GaN over the last decade in applications needing semiconductor devices with a high efficiency and a broad bandwidth. Here, one of the appeals is replacing multiple narrowband LDMOS channels with a single wideband one made from GaN that offers comparable or better gain, power, and efficiency. Armed with these attributes, designers can build systems that are smaller and more efficient. In handheld radios this lower battery demand reduces weight, which is everything to a dismounted warfighter.
But the next market that GaN will conquer is much closer to home. Commercial telecommunication systems require highly linear PAs and power efficiency is becoming critical for a host of practical reasons. Infrastructure waveforms use complex OFDM schemes to shoehorn as many bits as possible into a given bandwidth, and the energy has to remain in-channel to prevent interference. Power amplifier designers have employed several linearity enhancing techniques over the last couple of decades, moving from single carrier with simple back-off, to feedforward multi-carrier in GSM networks. Today, the optimal solution is a combination of Doherty combining and digital pre-distortion (DPD). Doherty gives high power efficiency with high peak-to-average ratio (PAR) waveforms, and DPD reduces the vector errors and adjacent channel interference.
So how does GaN perform in this contemporary architecture? The answer starts with an understanding of a key fundamental behavioural difference between GaN and LDMOS. LDMOS has a so-called hard compression characteristic, that is, it is very linear to near P1 dB, but compresses, or limits, just above that level. In other words, PSAT is a hard limit, encountered abruptly and the transition between linear and non-linear regions is almost a point. GaN has a soft compression point. P1 dB and PSAT may be separated by several dB, and if you continue to increase drive, the device will continue to deliver marginally more power.
This is ideal for use in DPD linearized power amplifiers as the Pin versus Pout relationship is relatively continuous. The resulting linearity near compression is better than LDMOS and less sensitive to factors like temperature, voltage, or load that tend to move P1dB. The compression point of GaN is continuous, rounded, and in mathematical terms, it is comparatively well behaved across the upper operating region, where power and efficiency are highest. What this means in practical terms is that if the amplifier is properly matched, instantaneous power can well exceed the stated CW capability of the device, which is ideal for reproducing high PAR waveforms.
When digital correction is in place, GaN sets a new benchmark at the system level. Compared to LDMOS, the efficiency associated with a GaN lineup is between 5 percent and 7 percent higher, all other things equal. Comparisons of raw efficiency are very favourable: providing gain at 2.5 GHz, 100 W class GaN devices configured in Class AB can exceed 70 percent, while today’s best LDMOS devices struggle to reach 60 percent. When properly utilized in commercial applications, the superiority in efficiency can deliver a significant impact at the system level.
At MACOM of Lowell, MA, we are enabling commercial adoption with our GaN-on-silicon technology. Now in its fourth generation, it will enable commercial applications and markets to benefit from GaN’s exceptional performance, efficiency and bandwidth at cost structures in line with LDMOS. To deliver cost savings that will accelerate the adoption of GaN in commercial applications, we are scaling the manufacture of our devices to larger wafers.
Another great opportunity for GaN is in providing the source of radiation for the microwave oven. There are emerging magnetron replacement prototypes that utilize LDMOS, but their efficiency falls short of what would be needed for them to be a viable alternative technology. With GaN it is a different story, because devices made from this material can bridge the 10 percent efficiency gap between the desired value and that associated with LDMOS-based modules.
Our technology is ideal for providing a replacement for the magnetron, because it combines the efficiency of GaN − at 2.45 GHz, efficiency is 70 percent − with a cost that is comparable to that of silicon. It is an attractive alternative to the tube based technology that can be traced back to the 1940s, because it enables a constant output power, zone-controllable heating and a ten-fold hike in the oven lifetime.
Plasma lighting offers yet another opportunity for GaN. This form of illumination involves RF power excitation, which is largely serviced with LDMOS technology. Relatively low frequencies are used, such as hundreds of megahertz. Although plasma lighting has been slow to make inroads in the overall lighting market, it has found its best niche in lighting for crop growth. When used for this, it is an ideal lighting source, thanks to a colour temperature that very closely matches that of natural sunlight.
Developments in plasma lighting are now underway that will increase frequencies toward 6 GHz and efficiencies beyond 70 percent; these are specifications that are incredibly challenging for LDMOS, but a natural fit for GaN. With its higher power densities, transistor dimensions can be trimmed, making these devices a very attractive option for vendors that are driving plasma lighting as an alternative to the LED in the indoor light bulb replacement market.
GaN can also aid wireless power transmission technology. Low power, consumer grade wireless power transmission for handsets already exists, while larger scale wireless power generation and harvesting is in the prototype stage of development. With power measured in kilowatts of radiated power efficiency and stringent constraints on the dimensions of the unit − the higher the frequency of transmitted power, the smaller the physical antenna − GaN is the obvious choice for RF energy transfer applications. With devices made from this wide bandgap semiconductor, efficiency can be 10 percent higher than that for LDMOS at 2.45 GHz, an optimal frequency for antenna size.
With opportunities in plasma lighting, microwave ovens, and medical to name but a few, it is clear that GaN can serve many mainstream markets. It is an appealing option, due to the performance it delivers at high levels of efficiency. This enables an overall reduction in costs, allowing it to serve price-sensitive, high-volume applications in the commercial market.
We are firm believers in the great potential of GaN, and we are prepared to support the entire supply chain, in order to drive the adoption of this class of device. Through investments, acquisitions, and manufacturing agreements, we have positioned ourselves to scale and leverage the capabilities of GaN for widespread commercial applications at low cost for our customers.
Developments in plasma lighting are now underway that will require increased frequencies toward 6 GHz and efficiencies beyond 70 percent