Why LDMOS is the best technology for RF energy
There are many possible applications, including lighting, medical, cooking, heating, drying, defrosting and automotive. The decisive factor in the early adoption and later mass adoption of solid-state power transistors in this RF energy market is the latest progress in the design of highly efficient, low-cost solid-state power amplifiers, made possible by key improvements in LDMOS and GaN technologies on the semiconductor component side. Looking at both these semiconductor technologies available in the market, end users might want to know which is the best technology to adopt at this moment in time for the emerging RF energy market. This article looks at the pros and cons of each.
Maturity and reliability
LDMOS has been and still is the dominant RF power device technology in the cellular communications infrastructure market, having successfully displaced the silicon bipolar transistors in the nineties and held out against other technologies such as GaAs. The same trend has been seen in adjacent RF power markets such as broadcast, radar and ISM, where it has dominated for the past twenty years. The RF energy market has similar requirements, and for the same reasons will also be drawn to LDMOS. On the other hand, GaN technology began and is continuing to progress in markets ranging from cellular base stations to two-way communications and radar. However, even though GaN is now considered a proven technology, it still needs to prove its long-term performance and reliability in the mass market. The widespread adoption of LDMOS in these other RF power applications has happened thanks to the continuous improvement of application-specific reliability and customization of LDMOS device structures within its different series of development and node generations (see figure 1), but also thanks to the ongoing ability to leverage low-cost silicon manufacturing technologies.
In terms of reliability during operation, LDMOS is a more reliable technology for high-power continuous wave(CW)-type applications than GaN. Because of the greater power density of GaN, it is grown on SiC substrates rather than Si for high-power transistors, since SiC has a much better thermal conductivity than Si. With LDMOS the dissipated power per square millimetre is much less compared to GaN and therefore there is less mechanical stress and fatigue on the silicon, solder joints and so on, improving reliability in operation for CW-type applications.
Performance and ruggedness
GaN devices have potential size, power, bandwidth and efficiency advantages over LDMOS for very high-frequency applications. GaN-on-SiC is focused on improving performance at an acceptable cost premium, and it is mostly a performance-driven solution. Alternatively, GaN-on-Si is a clear candidate for high-volume, cost-sensitive applications with a high-performance requirement. However, when looking at the list of available products in the market for both technologies and making a direct comparison, GaN-on-Si does not bring a performance improvement compared to LDMOS to justify a higher cost. In fact over the past five years, LDMOS has shown a huge improvement and made continuous steps towards higher power and increased efficiency levels, generation after generation, further reducing the justification for using GaN-on-Si (see figure 2).
In CW applications like the ones used for RF energy, the thermal resistance of GaN at the transistor level is higher compared to LDMOS due to its inherent higher power density. Because of that, in CW operation LDMOS is thermally superior to GaN. Additionally LDMOS does not need to sacrifice a number of bond wires to lower the cost, making it the technology of choice in the most rugged devices.
Furthermore, during CW operation the device gets hotter and the performance from a GaN amplifier solution is de-rated significantly more than a LDMOS amplifier, which again makes LDMOS a better candidate for CW applications. The same applies when cooling possibilities are limited because of cost and size restrictions, since the temperature of the baseplate of the amplifier in the final application is significantly higher. This can be seen in the performance data characteristics of the devices offered from different manufacturers, and technologies used when showing the de-rated RF performance over baseplate temperature curves.
Cost is a key factor in the introduction of solid-state devices to the RF energy market to replace magnetron tubes. If the price differential between solid-state and magnetrons remains, magnetrons will still be the preferred option for the cost-sensitive end of the market, however superior and smart the performance of the solid-state alternative turns out to be. It is well known that LDMOS is currently cheaper than any type of GaN, and LDMOS manufacturers are still improving the manufacturing costs associated with the process. Furthermore, when designing with a GaN at the final stage, the whole line-up, pre-driver and driver, needs to be designed with GaN to keep the same supply voltage, which makes the overall solution more expensive than any possible LDMOS equivalent. In addition, GaN requires dual-sequenced supply voltage – negative and positive – to operate, and this increases the cost and lowers the overall system reliability. Finally, due to the much higher dissipated power per square millimetre of silicon, GaN solutions must use bigger heat spreading structures and/or more expensive packages, which all adds to the bill of materials.
There is no one-size-fits-all approach and, as with any technology, the critical factor in a successful product introduction lies in understanding the customer and application-specific requirements. Choosing a solution requires more than just a comparison of product datasheet parameters. Only careful examination of providers’ manufacturing process technology and product reliability in real-application conditions can ensure the selection of proper technology solutions that meet demanding customer and infrastructure requirements. Because of its favorable performance/cost trade-off, LDMOS is a very attractive technology for RF energy frequencies at 433 Mhz, 915 MHz and 2.4 GHz. Meanwhile GaN has clear advantages in power density, at very high frequencies and to cover wide bandwidths, but for RF energy applications with CW application signal and which are mainly narrowband this benefit is not applicable. On the contrary, the higher power density of GaN is a drawback thermally, and does not justify the cost premium. Alternatively, GaN-on-Si, another flavour of GaN technology, tries to cover an area where GaN-on-SiC cannot compete, specifically at 2.4 GHz, claiming to bring the performance of GaN at the cost of LDMOS. However, actual RF performance results show that with LDMOS it is possible to achieve comparable performance results at lower component and system prices and at a higher level of ruggedness and lower thermals. For the RF energy market performance, reliability and overall line-up cost targets must be considered together, and LDMOS technology is therefore a clear winner with proven long-term reliability.