Engineering Ambition Can Be Scary - Development of High Power GaN Transistors
A heavy tension hung in the air as the RFHIC test group prepared to switch on their new 600 watt transistor. Engineers in the room huddled behind chairs and under desks to avoid any arcs or sparks as the test began. A familiar hum broke the silence as the test ran and data came pouring in, the team relaxing back into business as usual.
Organizations in the RF space are constantly predicting, adapting and responding to the shifting challenges and demands of their customers, designing and manufacturing solutions that continue to push the envelope of what our technology has to offer. It can be easy to overlook moments like these, where decisions that range from flipping a switch on a high-power transistor to completely changing an approach to manufacturing test are inherently daunting. But that does not stop the teams at RFHIC and NI from facing them.
In this article, the two companies weigh in on the journey to bring GaN on diamond to market, their partnership and how they’re working to address what comes next for the RF power amplifier space.
“For GaN on diamond [transistors], we were the first to do it, so we had to figure out everything from scratch.”
RFHIC is a key player in the GaN on silicon carbide transistor market, serving both telecommunications and defense. The benefits of GaN on silicon carbide are well known and mature but still present limitations for high-power applications. These limitations and the constant demand for improvements in speed, power density and thermal conductivity led RFHIC to develop and showcase the world’s first commercialized GaN on diamond transistor. Boasting higher power density, significant improvements to heat dissipation leading to a longer lifetime and a smaller footprint, these transistors bring a new innovative foundation to the RF power amplifier world. RFHIC’s next daunting challenge was how to effectively scale manufacturing of this solution.
“The scary part was the unknown unknown vector,” explained Sam Kim, vice president and general manager at RFHIC. “When we commercialized GaN on silicon carbide, there were a lot of unknowns, but we had various foundry partners who had proven technology for the GaN on silicon carbide infrastructure. For GaN on diamond [transistors], we were the first company trying to make four-inch wafer hemp devices, so we had to figure out everything from scratch.”
Building the Partnership
“It’s human nature whenever you try something new to have some hesitation.”
“Our relationship started when RFHIC needed to find a way to scale their manufacturing efforts,” said David Hall, go-to-market director and head of semiconductor marketing at NI.
The automated test and measurement company partnered with RFHIC to tackle several challenges around increasing capacity and throughput of their production lines. The two companies agreed on modernizing RFHIC’s entire production test line with PXI and LabVIEW™, pointing to a few key indicators of success, including test time, number of stations and operator hours. Switching over to something so different meant having to get buy-in, so first teams had to successfully recreate existing processes with the new hardware and software.
“It’s human nature whenever you try something new to have some hesitation,” said Kim, noting the difficulty of both validating a new test approach and meeting aggressive go-to-market deadlines.
Switching to PXI and LabVIEW immediately reduced test time from 50 seconds to 8 seconds, increased daily capacity from 30,000 to 80,000 devices per station and decreased the number of stations needed to run tests from 18 to 12. Operators also saw a dramatic reduction in time spent setting up and running tests from customized user-friendly HMIs and training.
Scaling for the Future
Optimizing manufacturing lines to increase throughput and capacity have a huge impact on reducing cost for RFHIC and their customers. The smaller, higher efficiency GaN on diamond transistors also help to drive down operating costs. In the base station industry, customers are charged per unit volume and weight, so the smaller and lighter the components, the lower the cost. As production for these transistors ramp, other benefits will begin to manifest.
“One of the things we see as a pretty consistent challenge in this industry is the desire to test devices across broad enough specifications that you can guarantee those specifications to your customers,” said Hall. “One of the benefits of having higher throughput is you can increase test coverage and incorporate more specifications like error vector magnitude without actually increasing cost.”
The incorporation of a software connected modular solution also helps RFHIC stay agile. Kim explained that in the RF space, customer demands can change from quarter to quarter and they need to be able to adapt just as quickly to changes in test requirements. This is already being seen in the reuse of test systems from GaN on silicon carbide lines for GaN on diamond.
Today’s challenges in producing smaller, more efficient components for RF base stations in telecommunications and defense range from modifying steps during production test to introducing new innovative technology to the market. In all of these, organizations need to be prepared to address the unknown unknowns that threaten progress.
“There’s not just a fear in executing a single test, there’s also fear associated with completely changing your approach to manufacturing,” noted Hall.
In RFHIC’s case, there was a significant amount of complexity in their existing production lines. From pick and place robotics, drivers and attenuators conditioning signals to be read by VSTs to the software required to automate measurements, there’s a tremendous business risk in rearchitecting a system that’s already working.
“It’s the nature of the RFHIC team to not be scared of trying something new,” said Kim.
From LDMS to GaN on silicon to GaN on silicon carbide and now GaN on diamond, the company is familiar with exploring new ways to continue bringing value to their customers. The goal is to find the tools, teams and partners that provide a level of expertise and certainty to combat unknowns when they arise.
Listen to more on this project in this exclusive podcast: