by Richard Dixon and Jérémie Bouchaud, Wicht Technologie Consulting
The market for MEMS inertial sensors (accelerometers and gyroscopes) is set to grow from $835 million in 2004 to over $1360 million in 2009 — a CAGR of 10%. Currently, the main applications are in the automotive industry. These markets are well established and growth rates range from a stagnant 1% for airbag acceleration sensors up to 8% for gyroscopes used in ESP units and GPS navigation assistance.
Much more exciting for MEMS inertial sensors is the market opportunity for mobile applications and consumer electronics (see table). Over the next few years, we predict annual growth rates exceeding 30% for accelerometers. Mobile phones in particular will provide multi-axis accelerometers with interesting opportunities in menu navigation, gaming, image rotation, pedo-meters, GPS navigation and the like. Gyroscopes are largely servicing markets for image stabilization and HDD protection in camcorders.
In contrast to the automotive sector, consumer applications feature relaxed specifications. Failure rates for automotive electronic control units (ECU) that house airbag accelerometers must be less than 50 ppm, and down to a few ppm for ASICs. Car manufacturers deploy reliable, high performance accelerometers that are relatively expensive (up to $5 to measure lateral acceleration in ESP units, for example). Mobile phones manufacturers on the other hand tolerate failure rates of 5000 ppm. Active control of the car motion is critical to safety whereas failure in consumer applications result more in an inconvenience. Sensor manufacturers can therefore sacrifice some accuracy and reliability to lower the price.
Tri-axis accelerometers for mobile applications
At least 10 companies are working on tri-axis MEMS accelerometers, some of which were initially developed for the automotive industry. Three axes sensing is needed to fully describe three-dimensional motion, although certain applications benefit from 2-axis devices if one axis (like the horizon) is assumed fixed or for a simple 2D (tilt) game in a cell phone.
With few exceptions, accelerometers work by measuring the motion of a proof mass versus a fixed frame or reference. The main sensing approaches are capacitive (e.g. Bosch, Freescale, Kionix, oki Electric, STMicroelectronics and Analog Devices), piezoresistive (e.g. Hitachi Metals, Matsushita, Fujitsu, and Hokuriku) or the less common thermal accelerometers from sole proponent MEMSIC (with 2-axis and 3-axis in development). Each has its advantages but price is the bottom line.
Requirements for cell phones
Mobile phone integrators want the following attributes from accelerometers:
- Small packages, e.g. 3.5 × 3.5 mm² and 1 mm thick or less
- Sensor and signal conditioning asic in the same package
- Digital output
- Power consumption < 3 mW (1 mA @ 3V). Many devices feature a microwatt-level sleep mode to reduce power consumption further
- Compatibility with lead-free soldering processes
- Availability in volume for $1.35 or less in next 2–3 years.
These conditions have already been met. The smallest accelerometer to date is made by Hitachi Metals, a 3-axis device in a 12 mm³ (3.4 × 3.7 × 0.92) package at least 50% smaller than its nearest competitor.
The lowest cost 3-axis sensors come from Analog Devices or Freescale ($2 in volume). MEMSIC provides the lowest cost solution for 2-axis sensors (now under $1 and $0.50 expected in 2007). Meanwhile, the power consumption prize goes to Analog Devices’ ADXL330 with power consumption well under 1 mW.
Gyroscopes improve picture quality
Today, gyroscopes are used in image stabilization, where two axes of motion sensing are needed to remove vibration and shake. While other approaches are also used to damp vibration, gyros are a superior solution.
Gyroscopes have been a feature of the top range of Panasonic camcorders for some years. They are now also employed in some Pentax and Panasonic digital still cameras costing as little as $250. Interestingly, MEMS gyroscopes are also used for HDD free-fall protection in recent camcorders from JVC and Toshiba (these camcorders retail for over $1,000).
Most commercial gyros are vibratory and use the transfer of energy between two vibration modes caused by Coriolis force. They are challenging to manufacture (hence their higher cost compared to accelerometers), being essentially two very high performing MEMS devices in a single device, usually a self tuned resonator for drive and micro-g for the sensing. Compared to automotive applications, packages in consumer electronics are about 1/10th the size (targeting 25 mm³). The resolution requirements are lower, i.e. 5–10 °/s compared to 0.1–1 °/s in cars.
A gyro is a larger device than an accelerometer, and can require as much as 10 times more power due to the higher drive requirements (5V).
There are many different sensing approaches and materials employed to make MEMS gyroscopess. Silicon micromachined capacitive gyros are produced by Analog Devices and Kionix. Piezo-ceramic devices are available from NEC Tokin and Murata, thin-film resonators deposited with PZT from Matsushita, quartz from Seiko Epson and Microcomponents (part of Swatch). These are all single axis solutions.
Currently, Murata manufactures a piezoceramic single-axis solution that costs $7–8. Sony and Invensense are examples of companies with two-axis solutions estimated to cost under $10.
Today, the smallest packaged device is 21 mm³ — a single axis quartz piezoelectric gyroscope from Seiko Epson. At 6 mW, it is also the lowest power solution, in part due to a lower drive voltage compared to 5V drive capacitive gyroscopes.
The cell phone remains the dangling carrot for gyroscopes, but they are still too expensive for this application. Cell phone companies tell us the acceptable price is somewhere around $3–5 for 2-axis sensing for an application like GPS navigation.
Currently it is not clear which technology or solution will win. While Sony is manufacturing a quartz 2-axis gyro, Murata is developing silicon capacitive micromachined sensing elements (in parallel to its ceramic gyros) to meet the consumer price targets. We believe the route for low cost 2-axis gyros will likely be a monolithic silicon capacitive solution. This is an important consideration to reduce package size and for integration of the ASIC.
Inertial measurement units arrive to consumer markets
There is a trend in consumer markets to integrate accelerometers and gyroscopes into inertial measurement unit (IMU) modules. This follows the military muster and recent automotive trends. In automotive, modules are partitioned to provide measurement in four degrees-of-freedom (DOF) to be distributed between navigation and vehicle dynamics (e.g. vehicles from Volvo and Fiat).
IMUs are targeting navigation applications in PDAs and mobile phones. AKM already supplies an electronic compass, albeit with a 3-axis magnetometer and mixed-signal processor in phones from NEC, Sanyo, SonyEricsson and LG. One application that has gained traction in Japan is an electronic compass that allows the calculation of the position of the constellations. AKM is working on compasses that include a 3-axis accelerometer, due in 2007. The package size is 4.8 × 4.8 × 1.3 mm³. A leading Asian supplier is also developing a navigation module, first using a 3-axis accelerometer and magnetometers, and subsequently a 2-axis gyro.
The first IMUs with MEMS sensors are estimated to cost $20, e.g. for high-end PDAs ($800). In 2-3 years time we believe $6-8 is the goal, whereby a gyro will need to cost under $5 for 2-axes. Eventually, prices for mobile phone IMUs will not exceed $6.
An important requirement is the signal processing algorithm — which is non-trivial for five or six degrees-of-freedom. In contrast to the automotive industry, consumer integrators prefer to plug and play.
Three ways to deal with tight margins in consumer markets
Finally, let’s look at how to make money in the consumer sensor market, where margins are tight.
1. Prices can be pressured downwards and better margins generated when sensors are developed to be transparent to CMOS processes. An example is MEMSIC who uses fabs to produce its low cost dual axis accelerometers in high volume.
2. The margin for an IMU module is higher than for sensors alone.
3. There are also opportunities for companies providing algorithms for IMUs. These are required to harness the power of a 5- to 6-axis functionality by turning the complex signal signatures into commands and for signal recalibration in real time.
Accelerometers have already met the requirements in terms of size, power and cost to open the markets for mobile consumer electronic. However, we believe it is will be a considerable challenge to reach the price target of $1 and below set by cell phone integrators for 3-axis devices.
Gyroscopes remain challenging to manufacture. There is great interest in a 2-axis gyro costing under $5. Such a price would expand the applications of gyroscopes from image stabilization in digital still cameras and camcorders to cell phone cameras, navigation imus and even further into protection of hard disc drive.