Two players offer Bluetooth silicon

Two players offer Bluetooth silicon
CAMBRIDGE, England — Cambridge Silicon Radio Ltd. (CSR) has demonstrated a single-chip CMOS Bluetooth transceiver link, and Philsar Semiconductor (Ottawa) has separately announced plans to develop a single-chip radio transceiver that it will use as the basis for an entire line of products built around the Bluetooth specification for short-distance wireless networking.

"Bluetooth is the biggest market opportunity we have identified in a long time," said Joe Rouse, marketing manager at Philsar. "By 2005, we expect the total market for Bluetooth-enabled devices to reach $3 billion."

Last month, both Silicon Wave (San Diego) and Atmel Corp. (San Jose, Calif.) announced integrated Bluetooth solutions, and analysts have become more bullish about Bluetooth shipments.

Cambridge Silicon Radio (CSR), which spun out of Cambridge Consultants Ltd. earlier this year, has demonstrated a 2.4-GHz Bluetooth radio link between single-chip Bluetooth transceivers built in 0.35-micron CMOS. This week CSR taped out its BlueCore01, which includes the RF transceiver, baseband circuitry and a proprietary 16-bit microcontroller, the XAP2, in a single chip. The company has also announced details of its BlueCore road map, which shows BlueCore02 and BlueCore03 debuting late in 2000 with a move to 0.25-micron CMOS and protocol firmware included in on-chip memory.

The Bluetooth specification was created by a consortium that includes Ericsson, IBM, Intel, Nokia and Toshiba, to establish a short-range wireless connection for mobile phones, desktop and notebook computers at data rates of up to 1 Mbit/second. More than 500 companies have joined the Bluetooth Special Interest Group.

"Ours will be the first truly single-chip solution," said Glenn Collinson, marketing director at CSR. "Other companies are talking about single-chip solutions but theirs are really only the radio section. If you look in detail you find the baseband logic and microcontroller are external so those are really two- or three-chip solutions at best. Other companies are using SOI silicon-on-insulator BiCMOS or even GaAs for the radio section. If you tried to bring the logic onto such processes, at 0.5-micron or 0.6-micron, you'd end up with an impossibly large die."

Collinson said CSR expects to have silicon of the 3-V, 0.35-micron BlueCore01 back from its European foundry in January and in the hands of leading customers in February. He said BlueCore01 chips will hit high volume in the middle of 2000.

Collinson declined to name the foundry. "Because we are using a CMOS process we have tremendous portability of our design. Time-to-market is key and that was one reason we are fabbing in Europe for our first single-chip design, but we will definitely go with a second source next year in the Far East," he said. Collinson added that CSR already has a number of customers for BlueCore01 but declined to name any. "We have a number of major OEMs in the GSM handset business, RF modules and wireless headset markets. We're also in the PC and PDA arena."

CSR's superheterodyne RF design is a near-zero intermediate frequency version. Collinson said it requires few external components. "The design avoids the disadvantage of a direct down-conversion receiver — principally large variable dc offset. We only require 12 or 13 surface-mount capacitors and two or three surface-mount resistors; no inductors and there's no external SAW filter. We also have a fully integrated VCO voltage-controlled oscillator."

Memory requirements

BlueCore01 does require a separate memory to contain the Bluetooth protocol firmware and includes a host interface to a separate microprocessor. Collinson said the memory would be integrated in BlueCore02, due to follow about six months after BlueCore01. He said BlueCore01 and 02 have a host interface because the Bluetooth specification requires it, but that the XAP2 microcontroller is more than adequate to run the protocol. "It's very low power, being only 8,000 gates of logic, but more than sufficient to handle the data rate of 1 Mbit/s at the RF interface and the payload data rate of 723 Mbits/s at the host interface."

With a shrink to 0.25-micron CMOS, BlueCore02 will include embedded memory. There are likely to be one-time-programmable EPROM and mask-ROM versions, Collinson said, because embedded flash will either not be available from foundries or not be appropriately priced.

BlueCore03 will dispense with the host interface and is intended for standalone embedded applications that do not need processing power beyond that provided by the XAP2. "Bluetooth can go into anything and many of these things will not require great complexity but will require power and cost efficiency," said Collinson.

The XAP microcontroller was developed by Alistair Morfey at Cambridge Consultants Ltd.

Collinson believes BlueCore02 will hit a $5 price point, providing low total cost for OEMs using it. "$5 for the RF section is reachable today. We know that because a DECT digital European cordless telephone front end is $5 or $6. But a complete Bluetooth interface is further away. With two or three chips a Bluetooth bill-of-materials is about $15 today. We will get to $5 cost for the BlueCore02 chip, but with BlueCore02 you only need a handful of passives and a 10-cent antenna, so the chip cost is effectively the bill of materials."

Ready for adoption

Meanwhile, Philsar has big plans for Bluetooth and its single-chip radio transceiver chip. Within a year, Philsar plans to roll out several more products using the same architecture, and it expects this ramp to occur at the same time as a large-scale consumer adoption of Bluetooth-based applications.

The PH2401 transceiver operates in the 2.4-GHz ISM frequency band, and runs at 1.8 V. Later versions of the chip will use even less power, the company said. The first product is expected to sample next quarter, and it will debut at prices below $5 in volume.