SiGe threatens to weaken GaAs’ grip on automotive radar
Automobiles come with more safety features than ever before. Cars are now equipped with additional impact bars that increase passenger safety when the vehicle is rammed from the side, and air bags that reduced the extent of injuries caused by a head-on collision. However, radar based technologies that can prevent crashes from happening are only just starting to take off. Until very recently, these systems that are based on a set of GaAs chips were only seen in luxury cars.
The lack of penetration into mid and low-priced cars is not because automotive radar is a new technology - it made its commercial debut in topflight Mercedes-Benz models more than 10 years ago, where it offered an advantage over automotive cruise control. Radar detected any changes to the distance of the vehicle infront, and adjusted the speed of the car automatically. Automotive radar systems have evolved in the intervening years, and they now increase driver safety. Today’s drivers receive flashes or audible warnings to make them aware of other vehicles nearby, and if they fail to react, a sequence of seatbelt tightening and harder and harder breaking brings the car to a stop.
The capability of modern automotive radar clearly makes it very attractive. However, expectations of significant commercial success have failed to materialize for several years due to three factors, says Asif Anwar, director of the GaAs and compound semiconductors technologies program at Strategy Analytics: a slow migration of automotive radar from luxury cars to mid-range models; a European outlawing of one of the frequency bands used for this technology in 2013; and competition with alternative technologies.
Up until recently, automotive radar was only fitted as standard in some models of the Mercedes-Benz S class, the BMW 7 Series, and the Lexus 430. “You also had the migration into lesser models, such as from an S class to an E class, as an optional extra,” explains Anwar. Penetration into the far larger mainstream market has been slow, due to a lack of cooperation within the automotive industry. According to Anwar, chipmakers were put under pressure to cut prices, in return they demanded commitments to high volume orders, and it took a long time for the two sides to strike a deal. But Volvo’s recent promotion of the benefits of automotive radar suggests that widespread deployment of this technology into mid-price cars is happening now.
Over the last few years, automotive radar sales have also been hampered by a European Union ruling that will ban deployment of the 24 GHz variant in 2013. This move is to benefit radio astronomers, because signals produced from this form of radar - which emits over a wide band centered on 24 GHz - interfere with those that they are trying to detect.
Cars fitted with the 24 GHz variant of automotive radar are able to maintain an appropriate distance to the vehicle in front at relatively slow speeds in dense traffic. The 77 GHz narrow band cousin performs a similar function at higher speeds in situations where there is less traffic, such as on a motorway. However, efforts are now underway to develop a wideband radar technology replacement for the 24 GHz technology, which will operate over a band centered on 79 GHz. Outlawing 24 GHz radar has hampered the up take of automotive radar, because it has left car manufacturers with the dilemma of whether it is the right decision to deploy this technology before the ban comes into force. This uncertainty aids the producers of alternative technologies, such as those based on cameras or lasers that can take market share from automotive radar manufacturers.
Despite all these problems, Anwar predicts that automotive radar shipments are set to grow substantially, and sales of GaAs MMICs will increase from $54 million in 2008 to more than $130 million by 2013. Many of these sales will be shared between the two leading chipmakers of this technology, TriQuint and United Monolithic Semiconductors. However, WIN Semiconductors could also benefit, because this foundry offers suitable processes for the manufacture of automotive radar chips. And Hittite also operates in this market, following its purchase of this product line from Northrop Grumman. “RF Micro Devices also have the process technologies, since the acquisition of Filtronic,” adds Anwar, “so that’s something to keep an eye on.”
TriQuint’s automotive design effort is based in Dallas, Texas, and originated from the acquisition of Infineon’s GaAs business unit at the beginning of this decade. The first 77GHz GaAs MMICs designed with a fully optical mm-wave process by Infineon were transferred to TriQuint’s low cost and high volume 6-inch fab in Portland, Oregon in 2002, explains Markus Behet, a marketing manger for the company’s automotive radar products. Since then, the company has continued to develop this chip set.
TriQuint is now ramping its 77 GHz chip set for a lead customer that is based in the US. The voltage controlled oscillator, transmitter and mixer are all manufactured with a low-cost HBT process in the company’s Oregon fab. Low noise amplifiers and frequency doublers are fabricated by a fully optical, mm-wave 0.13 μm low-cost PHEMT process, and the switch is manufactured with a vertical PIN process. A copper/tin bonded flip-chip approach that minimizes parasitic inductances at these millimeter-wave frequencies has been introduced in the last few years for the chip set.
“You place your flipped MMICs onto the substrate, do a reflow cycle and then everything is connected,” explains Behet. “You don’t need to bond a large number of wires that are a potential source for performance degradation at these frequencies.” The flip-chip approach creates interconnects with very low losses. Use of this method does necessitate that TriQuint’s customers have access to flip-chip assembly capabilities – either in-house or at sub-contractors, but Behet says that manufacturers have readily accepted the innovative new process because of the advantages of fewer bond wires and a significant reduction of insertion loss for flip-chip components.
A typical lead customer will produce around 50,000 radar units in the first year with the potential for a few hundred thousand in years after. Automotive radar chip sets command prices of $20-60 depending on volume and radar architecture, so TriQuint’s lead customer could generate revenues of several million dollars per year for the company. This should provide a healthy income for the next few years, but further down the line success could be hampered by the rise of SiGe. “GaAs has a track record for 77 GHz auto radar chip sets but I see SiGe manufacturers looking actively at the market,” says Behet.
Infineon Technologies is leading this charge, and last December it announced that its chip will go into Bosch’s third generation of longrange- radar. This system, which has an improved range of up to 250 m, can be used to maintain a constant distance with the vehicle in front at high speeds, warn of collisions, support predictive brake assistance and eventually deliver automatic emergency breaking.
Wolfgang Lehbrink, Infineon’s marketing manager responsible for the product marketing of radar components, claims that the company’s fully automotive qualified SiGe-technology is deployed in the best long-range radar in the world. “We have ramped up at the beginning of this year, and we’re in volume production.”
Lehbrink’s claim for superior performance may raise a few eyebrows, because GaAs has the edge over SiGe in many RF products, and electronic devices made from GaAs are known to have low-noise figures and good switching performance. Lehbrink, however, believes that it doesn’t make sense to make component-by-component comparisons, because GaAs and SiGe are two significantly different technologies. “You can’t compare on a feature by feature level. What counts is the system performance. Infineon’s highly integrated front-end chip improves this by minimizing the number of RF transitions as a result of integration of multiple blocks.”
The system cost will also play a role in the success of this technology. However, it doesn’t necessarily follow that the winning semiconductor technology will be the cheaper one.
Anwar, for example, believes that other factors will make a bigger difference than the cost of the technology, because the tens of dollars spent producing GaAs chipsets is very small in comparison to the price tag of automotive radar systems, which sell for several thousand dollars. Incidently, the cost of producing an equivalent technology in SiGe is higher, according to Anwar, thanks to the low volumes of the market and the higher development costs of SiGe. The mask sets for this technology can cost several thousand dollars, although it is unlikely that Infineon has absorbed the full price for its SiGe development, because a significant proportion of its effort has been expended in projects sponsored by the German Federal Ministry of Education and Research.
Lehbrink’s response to the question of chip costs is markedly different. “What really matters is the system cost. That’s where Infineon’s SiGe transceiver, with its high integration, higher yield and significantly better reliability, gives the cost advantage to the sensor company.” According to him, SiGe allows the migration of tasks onto the chip that were previously performed discretely. This ultimately drives down the total cost of the system. “And obviously mounting one or two chips on a board [with SiGe], is a much more cost effective way, than seven or nine chips, as it used to be in GaAs.”
If Infineon is to wrestle a significant share of the automotive radar market from the manufacturers of GaAs chips, then it also has to convince potential customers that SiGe is the better product. GaAs, after all, is the incumbent technology, and many of its users may be reluctant to move away from a proven device technology. However, Infineon appears to have this issue in hand. It was a partner in an automotive radar project called KOKON that ran from 2004 to 2007, and this enabled the company to forge strong relationships with systems manufacturers Bosch and Continental and carmakers BMW and Daimler. “If you want to be successful in high-frequency radar, you have to have good system knowhow linked to close cooperation with your customers,” says Lehbrink. “Sensor development and production at 77 GHz is very different from taking off-the-shelf parts and putting them on a printed circuit board. Offering technology, design, test and qualification from one partner with long-term experience in high frequency technology is crucial to success in automotive radar applications.”
Long-term success in any market also demands the continual development of better products. Infineon is well aware of this, and has just embarked on a three-year project called “Radar on chips for cars”, which is partially funded by the German government. Collaborators in the project include BMW, Continental, Daimler and Bosch.
One of the goals of the project is lower cost automotive radar operating in the 77-81 GHz band. Turning to this far wider bandwidth enables an increase in spatial resolution. “If you look at the more bandwidth-hungry midrange and short-range applications for dense traffic, it is clear that you need more bandwidth and that will be provided by this frequency band,” explains Lehbrink.
Infineon is clearly mounting a strong charge on the automotive market, and SiGe looks well poised for success. Anwar certainly thinks so. He predicts that the market for SiGe chips in automotive radar will rise from $3 million in 2008 to nearly $100 million in 2013, when the industry will have reached a tipping point. SiGe will then start to enjoy some significant uptick, mainly due to the activities at Infineon.
Lehbrink agrees with this view, and believes that virtually every automotive radar manufacturer will go to the effort of switching from GaAs to SiGe, which will be utterly dominant by 2015. If Anwar’s predictions are correct, 10.6 million automotive radar units will be deployed in cars in 2013. “On the one hand that’s very, very strong growth. But on the other hand, if you consider that there are going to be 100-120 million new cars on the road in 2013, then that’s still less than 10 percent of the market.”
However, it is possible that the market could take off even more strongly if studies showed a significant decline in road deaths through the introduction of automotive radar, because this could herald the introduction of legislation that required this technology to be fitted to all new vehicles. This chain of events has already happened with tire pressuring monitoring, but this technology is far cheaper.
Even if governments don’t intervene, it seems that GaAs is destined to enjoy some good years of increasing sales to the automotive radar manufacturers. But there is that nagging thought that it could have been so much better if penetration had occurred far quicker, and it appears that SiGe will be the longterm winner. So manufacturers of GaAs automotive radar chipsets must capitalize on the next few years, and then fight to maintain strong relationships with their customers in an effort to slow down declining revenues from this sector.
