Wednesday was the APEC social event at the Aquarium of the Pacific. Most people consider it only a social event and stayed in the large lobby area and skipped the best ocean displays in the Aquarium. I have to admit there was a very impressive three-foot, giant crab that I would not want to meet free diving.
Maybe petting the sting rays and eating food at the same time was a bad idea. Speaking of food, why were they serving kid's food, like mac and cheese, and chicken fingers at an event where half of the attendees are PhDs? At least the Aquarium was a short six-minute walk from other restaurants so you could get something to eat.
Yesterday was the final day of APEC. In the session “Technologies that Advance Power Design” IS1.4.3, I watched the SiC/GaN marketing presentation by Richard Eden from IMS Research. The packed audience was probably hoping for a free research report telling them when they can start making money and selling SiC/GaN in quantities. But Richard Eden was giving away just enough information to sell his full report and I think he did spark some interest. The most important point he made and then clarified in the Q&A session was that customers value the performance of the technology enough, and are willing to pay 1.5x over silicon for GaN and 2x for SiC. I am still concerned that he said that GaN prices should match Si in 2019. I was counting on GaN prices lowering a bit sooner than that. Eden also reported that GaN will dominate up to 900V, then SiC above that. His final point was that GaN must prove its reliability before it moves into the industrial market. I would correct that some by saying that the technology is very close to proving its reliability for the industrial markets, but several years away for the automotive market.
SiC/GaN marketing presentation by Richard Eden from IMS Research.
In my Monday post I promised to talk about 5 Mhz -10 MHz inductors that would work well with GaN. It seems that Dartmouth can create CoZrO, 20-nM thick laminations with good Q and 5-nM thick oxides between lamination. The bad news is that it takes 40 minutes to make 40 laminations which is about 1 micron thick. The good news is that they have 18u-thick Cu that could handle the higher frequencies, perhaps even up to 3A. All that said, this thin film method is not yet cost effective.
There was a bit of hallway talk about a 1 MHz iron nano-particle inductor that was created in China by cooling (quenching) the material quickly from 2,000 degrees. It sounds like a good method of raising the useful frequency of powdered iron and lowering the cost by removing nickel from the recipe. This could be used for 600V GaN switchers since they still are <1MHz.
Just a final note: Attendees filled the final sessions despite the fact the sessions ended at 5:30 pm. That does not always happen.
--Jeff