Yole Développement interviewed Ray Pengelly, Strategic Business Development Manager, Cree RF Components, that recently announced commercialisation of the highest power and frequency plastic packaged GaN transistors for low cost radar and datalinks.
According to Yole Développement data (detailed in the new RF GaN Technology & Market Analysis report), RF GaN devices continue to challenge silicon’s dominant position. If the GaN-on-Si could be implemented in the 2-5 years, a more optimistic penetration rate- higher than 20% - could be envisaged. In the nominal forecast, RF GaN could reach more than 18% of the overall RF device market by 2020 (i.e. a 9% CAGR from 2013 to 2020). CREE is targeting RF GaN market mainly based on SiC substrates ; let’s discover how it will happen…
Yole Développement: Overall, how does CREE envision the future of wireless telecom in both civilian and defense applications ?
Ray Pengelly: We see the future of wireless telecom to be very healthy both in civilian and defense applications. Higher data rates, carrier aggregation, super-Wi-Fi (802.11ac), interactive applications such as 802.11p, battlefield comms. via “cellular” or satellite services will all contribute to the market.
YD: GaN tends to displace incumbent GaAs HEMT and Si LDMOS in high power RF apps. Is it a simple substitution or does it need a full redesign of the RF chain ?
RP: The answer depends on application but generally the reasons for changing from GaAs and/or Si to GaN are higher instantaneous bandwidths, wider channel bandwidths, higher gain/efficiency, smaller size and weight and lower power consumption. This really translates to redesign of the RF chain given a different voltage rail may also be required in the case of 50V GaN.
YD: What are the main technical and non-technical drivers using Wide-BandGap devices in RF apps ?
RP: The technical drivers depend on application but can include higher power density, higher allowable operating temperatures compared to GaAs or Si (by > 50 deg C), higher gain per device, higher watts/sq.mm both for discretes and MMICs, smaller overall size, higher efficiencies leading to lower power consumption and operating costs, superior reliability.
YD: Which applications fit best with GaN today ? and tomorrow ?
RP: Today, the most popular applications for GaN are in wide-band PA’s covering between octave and decade bandwidths for test instrumentation, military communications, CATV, etc.. GaN is also being heavily used in AESA’s over a range of frequencies from L thru’ X-band. GaN is being used in switch mode PA’s, Class F and Class J amplifiers for telecom applications associated with a range of high efficiency techniques the most popular being Doherty and Envelope Tracking. As gate lengths shrink GaN will be used more tomorrow in applications through Ka band and higher including SATCOMs, back-haul and automotive radars.
YD: Is the extra-cost of GaN balanced by real and proven added-values at system or module level ?
RP: Because GaN (at least from Cree’s perspective) is being used extensively now for telecom applications (mainly LTE), shipments of transistors has now reached millions per year leading to very significant cost reductions in that market-place. This has a knock-on effect in other applications. GaN costs are also dependent on the required specifications of the application particularly for state-of-the-art systems so in those cases the “extra-cost” of GaN is worthwhile to achieve metrics which would be very difficult or impossible to do using alternative approaches.
YD: CREE recently introduces highest power and frequency plastic packaged GaN HEMT for low cost radar and datalinks. Can plastic packages cover all high power RF apps ?
RP: Plastic packages are good for low CW powers up to 45 watts or higher power pulsed (low average power) applications. Where plastic packaging (at least today) has issues is for very high power CW operation. The need for Plastic packaging is usually driven by the need to lower overall product cost just as in the case of Si LDMOS. Plastic packaging does not compromise performance (other than that noted earlier) and can be used up to mm-wave frequencies.
YD: How do you envision recent Nitronex acquisition by MACOM? Any threat with GaN-on-Si vs. GaN-on-SiC ? What are the value-proposition of both technologies?
RP: The Nitronex acquisition by MACOM should provide needed funding to continue the development of GaN-on-Si. Historically, GaN-on-Si use has been primarily confined to fairly low power, lower frequency applications (e.g. Mil COMMs, CATV, etc). Silicon substrate thermal resistance and RF loss have limited its use in higher power and higher frequency applications. Given that GaN-on-SiC is establishing a rapidly growing installed base and its price has dropped so rapidly with the prospect of future price reductions with the eventual migration to 6-in SiC substrates, there will likely not be great incentive to use GaN-on-Si for higher power applications but it is likely the technology will continue to improve for niche segments.
YD: RF industry consolidation leads to bigger players like TriQuint + RFMD. Both had strong involvement in GaN RF developments before merging. How CREE will stay ahead of the competition in such a new playground?
RP: We view consolidation as very healthy for the industry. Cree is a large, vertically integrated company that consumes and sells a large number of SiC substrates and GaN epitaxy. We also have a large and growing SiC Power business that runs on the same line as our GaN RF products. Our RF business is able to leverage this corporate scale to be very competitive in the industry.
Raymond Pengelly has worked for over 40 years in microwave compound semiconductor technologies including gallium arsenide and gallium nitride. He has been employed by Cree Inc. since 1999 and is responsible for strategic business development of RF and microwave wide bandgap devices. He has written over 120 technical papers, 4 books, holds 14 patents and is a Fellow of the IEEE and the IET.