Single Chip Success

CSR's technology is to be found in high quality wireless stereos and in around 85% of all Bluetooth headset designs © CRS plc

CSR, the winner of the Academy’s MacRobert Award 2005, has been one of the outstanding success stories of UK engineering. The company was originally created to develop a commercial offering for a novel short-range wireless technology called Bluetooth. The company now employs hundreds of engineers worldwide and CSR's Eric Janson writes about how their success has been achieved.

From its foundation, CSR’s strength has always been its core engineering expertise. The original development of its Bluetooth technology presented a unique set of engineering challenges which, once overcome, gave the company a significant commercial and technological lead. Bluetooth provides a simple wireless connection for ‘personal connectivity’– for example, for hands-free mobile phones, PCs, MP3 player headsets and game consoles.

This lead has since been extended. CSR is now seeking to apply the same approach to a different wireless market, that of Wi-Fi. This is the communications standard initially developed to facilitate wireless communications between personal computers – in the office, in the home, or by those working on the move. CSR leads the Bluetooth wireless market both in terms of volume shipments and in the number of designs incorporated into other manufacturers’ products.

The initial challenge

By the end of the 1990s, the mobile phone market was growing rapidly. For convenience and safety many users were attracted by the idea of a wireless-connected headset. At the same time, Personal Digital Assistants (PDAs) and other mobile consumer electronic devices were becoming increasingly popular. Again, many users wanted a simple wireless means of exchanging data between these devices.

In 1998,a small group of manufacturers – IBM, Intel, Toshiba,Nokia and Ericsson – produced the initial Bluetooth specification. It defined a complete communications protocol which would operate in the unlicensed 2.4 GHz range of radio frequencies. (The name ‘Bluetooth’ derives from the 10th Century King of Denmark, Harald Bluetooth – grandfather of our own King Canute – who unified the various kingdoms of Scandinavia.)

Bluetooth was an attractive technology for mass-market products. However, in order for it to succeed, it would have to: work reliably and consistently in hundreds of millions of products; be easy to implement for the product manufacturers; be easy to use for consumers; and retail at a price that would reach a mass market. In addition, to ensure that products from different manufacturers would interoperate smoothly, the specification had to cover every aspect of the required standards and protocols.

Whilst this set of attributes makes Bluetooth a uniquely powerful and flexible communications standard, it also makes it technically challenging to implement. Any Bluetooth implementation must marry some or all of the traditionally separate technical worlds of semiconductors, computing, consumer electronics, software, radio transmission and cellular telephony. Since very few manufacturers had all the necessary technical expertise, this complexity was potentially a serious inhibitor to the widespread take-up of Bluetooth.

This, then,was the challenge that CSR set itself: to produce a low-cost implementation of a new communications standard and technology, in a sufficiently flexible and generalised way that could be incorporated into any manufacturer’s end product. And, to do this ahead of any competition.

Making bluetooth possible

In the late 1990s, engineers at Cambridge Consultants (CCL) had been addressing the challenges of integrating low power radio circuitry with conventional semiconductor designs. Seeing the opportunity for applying their skills to the market opportunities created by the Bluetooth specification,a core team of hardware and software engineers left CCL in April 1999 to found CSR. From the very start, the founders structured themselves around a single design team, encompassing all aspects of hardware, software and packaging. They also believed that attempting to re-use existing design componentry would the required design and market objectives. These factors led them to develop a superior single chip architecture from the ground up.

In many of the larger companies involved in the early days of Bluetooth development, the engineering teams charged with device design were separated into hardware and software teams, linked by design meetings and project managers. The CSR design team was small enough to be tightly integrated both in physical location and daily communication. This led not only to better design decisions, but to quicker ones.

At the time of CSR’s creation, the market for products which incorporated both radio frequency circuitry and computer circuitry was largely limited to mobile phones, using cellular phone radio technologies such as GSM (in Europe) or CDMA (in North America). In these products, the radio and computing capabilities were implemented on separate chips. The demanding performance of cellular radio was met by the use of exotic, expensive and high-powerconsuming semi-conductor technologies such as Silicon- Germanium or Silicon-on- Insulator. The less demanding computing functions (address book, calendar, memory etc) were implemented in a separate conventional low-cost, lowpower CMOS chip.

The challenge of CMOS

CMOS (complementary metal oxide semiconductor) is a process used to manufacture high-volume low-cost semiconductors for a wide range of computing and consumer electronics devices. CMOS semiconductors use both NMOS (negative polarity) and PMOS (positive polarity) circuits. As only one of the circuit types is on at any one time,CMOS chips require less power than chips that use only one type of transistor. In addition to their relatively low-cost, this makes them particularly attractive for use in battery-powered devices, such as PDAs, mobile phones and laptop computers. The CMOS process was first introduced in the 1960s and has evolved dramatically over the past four decades. Manufacturing advances have shrunk the circuits, reduced costs, increased computing power and made Moore’s Law a reality.

The CSR team believed that the Bluetooth specification could be met by a design that integrated the radio and computing capabilities on a single CMOS chip. It had hitherto been regarded as impossible to produce a reliable implementation of the required radio circuitry using CMOS technology. In addition, the radio and computing circuitry would interfere with each other unless they were effectively shielded. CSR’s unique insight was to realise that the apparent design disadvantages of using a single CMOS chip could be compensated by using additional control circuitry in the same chip. This circuitry allowed a relatively simple CMOS radio design to be kept within the tolerance limits of the Bluetooth specification.

In addition, the partitioning of functional requirements between dedicated hardware (eg for time-critical tasks) and flexible firmware (ie software running on the chip) lowered power consumption, whilst allowing CSR to keep up with evolving specifications. As a consequence of these organisational, design and technology decisions, CSR achieved its objective of producing a flexible family of Bluetooth chips, known as BlueCore™, which has become the implementation of choice for the majority of Bluetoothenabled devices.

More than a chip design

It takes more than good design to make a successful product, let alone a successful company. CSR is a ‘fabless’ semi-conductor company: it contracts out the manufacturing of its chips to a semiconductor fabrication facility, or ‘fab’. However, CSR retains its own in-house experts in both process technology and packaging techniques. This gives CSR a clear understanding of CMOS manufacturing processes and the ability to liaise constructively with its manufacturing and packaging partners. This has led to the early identification of manufacturing problems and significant improvements in the yield, ie the number of working chips on each silicon wafer.

CSR’s investment in software was critical to gaining initial market share. By adapting their firmware to the particular requirements of the endproduct, the optimal balance of performance, reliability and flexibility could be achieved for each manufacturer’s product which incorporated a BlueCore™chip. This approach also meant that the end-product manufacturer was completely insulated from the complexities of both the Bluetooth standard and the radio implementation. This enabled these manufacturers to focus on their own addedvalue features.

To take this ease of integration even further, CSR grew a large applications support team to help its customers build BlueCore™ into their products. Not only did this team help win business and get products to market, it also quickly built up a huge experience with the practicalities of Bluetooth design. This resulted in the development of robust reference designs. Feedback into the design group, in turn, improved the next generation of products.

CSR managed to build on its initial success by diversifying the range of Bluetooth products on offer. For the first generation there was only one Bluetooth chip, used by all applications. Today there are specific solutions aimed at phones, headsets, PCs and automotive markets, amongst others. Each includes dedicated chips with additional capabilities required by the particular market (eg microphone and speaker connections for headset chips), as well as special software and firmware.

Despite the commercial challenges caused by the collapse of the ‘tech bubble’ just after the company was launched, Bluetooth-enabled devices eventually captured the consumer imagination. Bluetooth headsets are increasingly common and the usage of Bluetooth in mobile phones is expected to rise to between 35% and 40% in 2006.

Future challenges for CSR

CSR has grown significantly since its early days, but its engineers continue to push the boundaries of wireless technology. UniFi1 is the company’s first foray into a single-chip implementation of all the variants of the Wi-Fi standard. The company recently declared its support for the WiMedia Alliance UWB (UltraWideBand) standard which will be incorporated into Bluetooth. The suggested UWB frequency ranges, up to 9 GHz, will provide new engineering challenges for the CMOS single-chip BlueCore™ model.

As mobile phones and other portable electronic devices become more complex and multifunctional, the issues of physical size and power consumption become increasingly important. CSR has been instrumental in expanding the remit of Bluetooth from a simple cable replacement technology to a fundamental part of our modern mobile lifestyle. For example, CSR engineers have developed a Bluetooth variant which supports wireless streaming of stereo music to Bluetooth headphones, while maintaining audio quality, and allowing simultaneous hands-free control of mobile phone calls from the headphones. CSR has also produced a variant of BlueCore™ that incorporates an FM radio on the same chip.

While wireless technology will never cease to create new challenges for engineers, CSR has proven that ‘impossible’ is no limitation to progress.

Biography – Eric Janson

Eric Janson is Senior Vice President, CRS Worldwide Sales.On joining CSR in May, 2000, he established CSR's US sales and applications engineering operations, headquartered in Dallas,Texas. Prior to joining CSR, he was with Lucent Technologies Microelectronics for 12 years in Munich, Germany and in Reading and Allentown, Pennsylvania.