Sensors Measure Up To Emerging Automotive Safety Standards

Currently, many of the advancements being made in automotive electronics depend directly on the development of advanced sensors and sensor systems. Sensor technology is vital to the development of many emerging automotive systems, which require not only accurate and reliable sensor elements, but also numerous electronic functions. These include signal conditioning and interface circuits, as well as the hardware and software required to process sensor data and prompt the necessary system responses.

Packaging, physical interconnect, and test needs also figure prominently in sensor system development. All elements of the system must be designed to withstand the harsh environmental conditions encountered in the automotive world while maintaining high reliability. In fact, industry veteran Joe Giachino of Visteon Corp. observes that the requirements imposed by automotive safety applications are actually tougher than those found in the military environment. At the same time, automotive components and systems must lend themselves to high-volume, low-cost production.

Despite these challenges, though, automotive applications are spawning the development of a host of sensor types for use in systems related to rider safety; engine and drive train; comfort, convenience, or security; and vehicle diagnostics or monitoring.¹ In many cases, these are MEMS devices because of their potential for high performance, integration, and low cost. Nevertheless, other types of sensors are in development too. Quite a few of these new sensors—both MEMS and other styles—are being designed to implement advanced automotive safety systems beginning to come to market now.

In planning these new systems, automakers are striving to build collision detection and avoidance systems that will protect the vehicle from harm. That's the long-term goal. But in the short term, automakers are looking to lower risks. To this end, they're starting to implement occupancy seat sensors that improve airbag deployment and angular-rate sensors for enhancing vehicle dynamic controls.

The need to deploy airbags more intelligently is spawning development of several sensing technologies. Until recently, the decision of when to fire airbags depended only on collision detection, which was accomplished primarily with accelerometers that rapidly detect the occurrence of front- and side-impact collisions. But the realization that airbags pose a risk to infants, children, small adults, and even average-size adults who aren't properly seated has created the demand for a more sophisticated airbag deployment scheme.

In developing new airbag systems, it's no longer sufficient to trigger airbag inflation solely on the basis of collision detection. What's required are "smart" airbag systems that take into account the size of the passenger and—in more advanced designs—whether the passenger is in position for a safe deployment of the airbag.

In their efforts to create smart airbag systems, the Tier One automotive suppliers designing and building them take their cues from organizations like the National Highway Traffic Safety Administration (NHTSA). The NHTSA, which determines standards for testing crash protection equipment, has established a new set of rules for how airbag systems will be tested. Whereas in the past it was sufficient to test airbags using a crash test dummy that corresponded in size to an average (fiftieth percentile) male, automakers in the near future will have to ensure that their safety systems protect a range of occupant types.

Systems will be tested with a family of crash test dummies representing a one-year-old infant, a three-year-old child, a six-year-old child, a small (fifth percentile) female, and the fiftieth-percentile male. In addition to protecting children and small adults, new airbag systems must reduce the risk of injury to passengers who are near the airbag when it deploys. The new regulations give designers different technical options, such as weight sensors and dual-stage inflators, to protect occupants. The NHTSA will begin to phase in the new rules for airbag systems in 2003.²

With these requirements in mind, automotive suppliers are actively developing a number of occupancy seat sensors based on different approaches. Techniques under development include weight, pressure pattern, infrared (IR), ultrasonic, electric field (capacitive), and video image sensing. For now, the industry is still evaluating the merits of each of these methods and hasn't come to any consensus on which will provide the optimum solution for occupancy sensing. As Roger Grace, a marketing consultant close to the automotive sensor technology scene, comments, "It's still early in the game for one strategy to be considered a standard."

Plus, given the complexity of the task and its tough reliability requirements, it's unlikely that any single methodology will emerge as the sensing solution. Craig Bezek, manager of Advanced Technologies at Motorola's Automotive and Industrial Electronics Group (AIEG), Northbrook, Ill., notes that there are drawbacks with each approach to occupancy seat sensing.

For instance, weight and pressure pattern sensing, which are already being deployed, can categorize the occupant by weight class. But these systems are limited in their ability to determine the position of the occupant. So to measure position, designers must turn to IR, ultrasonic, and electric field sensing.

Yet none of these methods is perfect. With each, tradeoffs will be made in terms of sensor mounting location, size, speed of measurement, and life cycle. Environmental factors, such as ambient light, sound, temperature, and humidity, also are concerns.

According to Bezek, electric field and IR methods tend to be the fastest. When it comes to coverage area, though, electric field and ultrasonic sensing cover a wider area than IR. It's possible to increase coverage with multiple beams, but that adds cost.

In addition, there are limitations associated with different operating conditions. For IR, sensitivity varies with the color of the object being detected. Therefore, environmental factors like the occupant's clothing could affect sensor performance. On the other hand, ultrasonic sensing could be susceptible to high-frequency noise sources in the environment. Because electric field sensing can be affected by the presence of metal objects or water, wet clothing could interfere with sensor readings.

Imaging systems that exploit stereo triangulation might ultimately provide the most accurate assessment of occupant size and location, but for now cost is still an issue with these systems. The cost of this approach should come down in time, though. Crash testing poses additional challenges. Sensor performance using test dummies may vary from that of live passengers. As Bezek notes, crash dummies might reflect sound differently, they don't give off body heat, and they possibly contain steel bars. Consequently, NHTSA permits the use of real people in static testing of occupant protection systems.

Given the evolving safety standards, we can expect to see a migration to the more sophisticated sensing schemes. But because of the tradeoffs inherent in different methods, Bezek says that "most companies are approaching it with a combination of sensors."

Echoing this sentiment, Henriette Eles, advanced safety systems manager at Visteon Corp., notes that with regard to occupant safety, "No one technology can really stand alone." Visteon is currently investigating a variety of approaches to occupancy seat sensing, including radar systems. One possibility is that the company could extend the technology being developed for radar-based collision detection and collision avoidance in order to perform occupant detection within the vehicle.

Some of the first occupancy seat sensors going into production now and in the next few years are based on weight and pattern sensors. For example, Delphi Automotive Systems of Kokomo, Ind., introduced its Passenger Occupancy Detection System B into the market earlier this year. System B is strictly a weight measurement system that can be used to classify the occupant into the child/child seat category or the adult category.

For weight sensing, System B relies on a fluid-filled bladder that mounts beneath the seat cushion. Within the bladder exists a single pressure sensing element. According to Walter Kosiak of Delphi Automotive Systems, weight sensing is "the next step in the evolution of restraint systems."

The company plans to extend occupancy sensing technology beyond occupant classification to determine occupant position for optimum deployment of front and side airbags. Ultimately, though, the company wishes to implement even more advanced systems. Kosiak suggested that Delphi would like to employ a personalized occupant detection scheme that's tailored to individual occupants and expands the airbag safety system to account for specific crash conditions.

In the short term, though, the company plans to introduce another Passenger Occupancy Detection design known as System A. This technology will accomplish occupancy detection through pattern recognition. In System A, an array of Flexpoint bend sensors will create a pattern that will be correlated with known patterns for objects like car seats and occupants of different sizes. The company hopes to launch this technology in 2002.

Something similar is being planned by another vendor, Siemens Automotive Corp. of Auburn Hills, Mich. Dave Ladd, manager of public relations at Siemens Automotive, claims that the company expects to see its Occupant Classification System (OCS) implemented in 2001 model vehicles from Ford. This system includes a sensor mat, which is integrated within a foam seat cushion and contains an array of 100 resistive pressure-point sensors. These sensors create a pixelated image or "footprint" for the passenger or object on the mat.

The footprint can be used to determine whether a passenger is actually present and if so, the type of passenger (Fig. 1). By evaluating the footprint, OCS can determine whether a baby seat is present or if a small, medium, or large passenger is seated. Siemens plans to follow OCS with a more advanced occupancy sensor system that's based on weight measurement.

The weight classification system (WCS), which is slated for production in two years, embeds strain gauges into the corners of the vehicle seat. The four sensors are fabricated from micromachined steel. Such a system determines the passenger's weight and center of gravity, which not only allows the WCS to conclude the occupant's weight category, but also proximity. Additionally, there are plans to enhance WCS by adding proximity sensors.

Beyond these systems, Siemens Automotive is looking to three-dimensional camera technology based on CCD or CMOS image sensing. Ladd notes that such image sensing techniques are attractive not only for their ability to perform occupant sensing—both size and position—but also for their ability to carry out other functions. They might be used to detect driver drowsiness or vehicle theft. Ladd says that image sensing technology is probably still about five years away from production.

At least one vendor is pursuing a system based on electric field sensing. Although details about the design are limited, Motorola AIEG indicates that the company has developed a prototype that uses electric field sensing combined with another form of sensing. The prototype can sense both the size and position of the occupant, and its electric field sensors have greater sensitivity than standard capacitive sensors. Plus, the system relies on special algorithms to process sensor data.

Having accurate information about the occupant is just one step on the road to smarter airbag systems. Another piece of the puzzle is knowledge of crash severity. Today, the decision on whether to fire airbags is based on a simple threshold of collision force, measured by front- and side-impact accelerometers or pressure sensors.

Intelligent Crash Evaluation In the future, however, the system may evaluate crash severity more intelligently by taking into account the angle of impact and how quickly the crush zone is crumpling. Not only will this information affect airbag deployment (including variations such as curtain, knee, and foot bags), but also seatbelt pretensioners, which can be used to bring the occupant into a safer crash position. These systems might also be called on to respond dynamically, deploying safety elements multiple times if necessary.

Occupancy seat sensors and collision detection accelerometers will initially communicate with the electronic control unit over a dedicated control bus. A more advanced safety bus, however, might expand to interface with drive-by-wire and collision avoidance systems. (For more on this topic go to the web site www.elecdesign.com/magazine/2000/sep0500/specreports/2SR1.shtml).

Angular-rate sensors, sometimes referred to as gyroscopes or gyros, can be used to perform three types of inertial measurements—yaw, roll, and pitch. Sensors designed to measure the first two motions are now finding application in vehicle safety systems. Rollover sensors are being designed to alert the airbag system in the event of a rollover. The goal is to keep passengers inside the vehicle and protect them from impact by deploying top and side airbags as well as seatbelt pretensioners and other safety equipment. The rollover sensor is necessary because the accelerometers used to detect vehicle impact during a collision won't detect the roll of the car.

On the other hand, yaw-rate sensors are necessary to run vehicle dynamic controls. Such controls represent an extension of anti-lock braking. With a yaw-rate sensor in place, vehicle control systems can better detect spinouts and skidding and take corrective action. Some of the present development of yaw-rate sensors also is geared toward applying them in emerging applications, like navigation and intelligent cruise control.

Although their requirements for sensitivity and measurement range differ, both rollover and yaw-rate sensors that are intended for automotive applications typically exploit the Coriolis effect to measure rotational velocity. The Coriolis effect is based on the principle that when an object moves within a rotating frame of reference, there's an inertial force called the Coriolis force that produces an apparent deflection of that object.

Various approaches are used to build angular-rate sensors based on this principle. One style available from Systron-Donner Inertial Division, BEI Technologies, Concord, Calif., which has been in production for about 10 years, consists of a double-ended tuning fork micromachined from monocrystalline piezoelectric quartz. This technology was originally developed for aerospace and defense before being applied to automotive applications.

The company's gyro contains two similar, mechanically coupled quartz tuning forks—a pair of drive tines and a pair of pickup tines. An oscillator within the sensors' associated electronics drives the drive tines to vibrate at a precise amplitude. These tines are driven by an external oscillator to produce vibrations of a precise amplitude. As they vibrate, the drive tines move together and apart. Due to the Coriolis effect, vibration of the drive tines makes them sensitive to angular rates about an axis parallel to the tines.

These angular rates produce an oscillating torque on the drive tines that's sensed by the pickup tines, which begin to vibrate up and down. In other words, their vibration is in a plane perpendicular to the fork and the vibration of the drive tines. The amplitude of the vibration on the pickup tines is proportional to the angular rate. Through piezoelectric action, the vibration of the pickup tines converts to an ac output that's amplified by the sensor's pickup amplifier and then demodulated. The resulting dc output varies with the measured angular rate of rotation.

The quartz tuning fork design is the basis for a family of products that target different automotive applications, which are now used to implement various angular-rate sensing functions. Presently, when multiple angular rate sensors and accelerometers are needed, they are applied in a decentralized sensing scheme. Systron-Donner, though, is developing a complete inertial package that will combine acceleration and angular-rate sensors in one unit. This will provide centralized measurement in stability control, intelligent cruise control, and rollover detection, as well as incident monitoring systems.

Rather than using separate sensors for all of these functions, a centralized accelerometer-gyro sensor unit will do the required sensing and communicate its readings with the various control systems over a communications bus. According to Brad Sage, director of business development at Systron-Donner Inertial Division, this product will be in production "in the not too distant future."

Sage notes that the centralized approach to measuring angular rate and acceleration has already taken hold in the aircraft industry and is now moving to the automotive industry. But if there's some reluctance on the part of automakers to go with a centralized approach to measurement, it reflects concerns over the reliability of such safety systems. Another issue is the lack of a common standard for the communications bus that would be required in a centralized measurement scheme. The car industry has yet to reach a consensus on one communications standard. "There are still three or four bus architectures," notes Sage. So for now, sensor suppliers must prepare to support multiple bus concepts. Ultimately, though, a centralized sensor system is desired because it will reduce the costs associated with using multiple sensors and complex wiring schemes.

Lower cost also is the motivation for automotive suppliers to develop gyros based on silicon micromachining or MEMS technology instead of older mechanical technologies. Jim Doscher, the director of consumer and industrial business units at Analog Devices' Micromachined Product Division, Norwood, Mass., says, "Presently, yaw-rate sensors cost up to 20 times the price of accelerometers."

Not only will the price of a MEMS-based yaw-rate sensor be lower than non-MEMS solutions, it will be possible to integrate multiple sensors—accelerometers and gyros—on the same silicon. Still, obtaining these benefits isn't easy. "A gyroscope is one or two orders of magnitude more difficult to make than an accelerometer," Doscher explains.

Essentially, this is because with a MEMS gyro you are taking an accelerometer and imparting a motion to it at a right angle to its normal plane of measurement. Doscher notes that one of the challenges in building a MEMS yaw-rate sensor is minimizing its sensitivity to vibration. Analog Devices, known for its MEMS accelerometers in airbag applications, is one of those companies with a MEMS gyroscope in development.

Micromachined Angular-Rate Sensors Another company is SensoNor, the Norwegian maker of MEMS devices. It's developing a family of angular-rate sensors fabricated in a micromachined bulk silicon process. The sensors use a "butterfly" structure that has a gyroscopic scale factor comparable to that of tuning-fork gyros. But unlike the tuning-fork-style sensor, this MEMS device operates with single-sided electrostatic excitation and capacitive detection.

The butterfly gyro contains two masses connected by an asymmetric beam. Using the piezoelectric effect, the masses are driven to vibrate. Because the beam is asymmetrical, the vertical forces associated with the vibration cause the beam to bend both vertically and horizontally. The masses are shaped so that when they vibrate, their velocity vectors are primarily horizontal. As a result, the vertical Coriolis forces associated with angular motion can be detected capacitively. The masses are forced to vibrate in antiphase because this approach minimizes offsets and lowers sensitivity to external vibrations.

The first application of this structure will be in a rollover sensor. According to Hans Petersen, manager for business development at SensoNor USA, it's due for release later this year, and it should be in production sometime next year. This sensor will have a measurement range of ±250°/s, which is sufficient for the relatively slow speed of a vehicle rollover. The company, however, plans to extend this design by increasing the part's sensitivity and accuracy. The rollover sensor will be followed by an angular-rate sensor for vehicle dynamic controls and then a version with even greater performance (a measurement range of 400° to 500°/s) for vehicle navigation.

The latter sensor will be required to pinpoint vehicle location in GPS-equipped cars. This is a necessity for navigation and to meet future FCC cell-phone requirements. GPS would be sufficient by itself, if the GPS receiver didn't lose lock with its satellites. Unfortunately, this occurs frequently in urban environments. But a yaw-rate sensor used in combination with a wheel speed sensor provides a basis for calculating a car's exact location based on the GPS's last known good reading.

One vendor already has a MEMS yaw-rate sensor in production. Two years ago, Robert Bosch GmbH, Reutlingen, Germany, started manufacturing a sensor that uses a combination of bulk and surface micromachining.

The yaw-rate sensor contains two accelerometers located on top of oscillating seismic masses. In order to make the masses oscillate, they are excited with an ac current while in the presence of a constant magnetic field. The accelerometers are configured to measure linear acceleration along an axis orthogonal to the oscillation of the seismic masses. But because of the Coriolis force, the accelerometers also are sensitive to rotation around a third orthogonal axis. Subtracting the output of one accelerometer from that of the other eliminates the signal corresponding to linear acceleration and doubles the signal due to the Coriolis force (Fig. 2).

The output of the sensor is synchronously demodulated to produce an output proportional to yaw rate. Because the oscillator and accelerometer are physically separated—unlike in other designs—the performance of each can be optimized. Another distinct feature of this design is that oscillator and accelerometer deflections are in the plane of the chip, allowing them to be more precisely defined by planar technology than if the deflections occurred perpendicular to the chip.

Measuring Dynamic Stability This sensor (part number MM1) is combined with a MEMS accelerometer to measure vehicle dynamic stability within a system known as the electronic stability program (ESP). The yaw-rate sensor detects rotation around the car's vertical axis, while the accelerometer determines linear acceleration perpendicular to the driving direction.

The ESP has been implemented in Mercedes and other European models, where it supplements the anti-lock braking system and helps prevent skidding. Wheel speed, steering angle, and high-pressure braking sensors are required to implement ESP as well. The MEMS yaw-rate sensor designed for ESP replaced an earlier Bosch gyroscope based on a more expensive, complex mechanical system that required extensive tuning.

According to Jiri Marek, head of automotive sensor development at Bosch, the company is planning to extend its MEMS technology further to produce angular-rate sensors for rollover and yaw sensors for GPS navigation. Full-scale measurement ranges for these devices are 250°/s and 100°/s, respectively. These devices are slated to begin production by early next year.

Delphi Automotive Systems is producing a MEMS angular-rate sensor, too. The device, currently in low-volume production, targets navigation and vehicle dynamic controls. Team leader for yaw sensors at Delphi Automotive Systems, Doug Sparks, claims that the company's yaw-rate sensor is unique in its integration of CMOS circuitry for signal conditioning (buffers and amplifiers) with the micromachined sensor. A second ASIC holds the control loops used to program the sensor for factors like sensitivity and offset control over temperature.

Whereas other angular-rate sensors tend to rely strictly on comb structures, Delphi's device combines comb and ring structures to obtain the performance advantages of each approach (Fig. 3). The comb structure provides greater capacitive area, leading to greater signal output. But, the ring is less sensitive to vibrations.

Operation of the sensor is comparable to a resonating wine glass. The ring is driven into vibration in the plane of the chip so that nodal points are formed. When the chip is subjected to angular motion, the vibrations at the nodal points increase in amplitude. These vibrations are detected capacitively by the comb structures. A signal is fed back to the ring to maintain a standing wave pattern with nodes at the defined locations. Because it's proportional to the angular rate being measured, this error signal serves as the output of the sensor.

Furthermore, the company claims the sensor offers a packaging advantage. A wafer-to-wafer bonding process seals the sensor in a chip-scale vacuum package. In addition to allowing for single-axis sensing in a chip-scale package, this type of wafer-level packaging also gives the company the option to do multiaxis sensing in a DIP or SIP device. That capability has obvious benefits when the design calls for a complete inertial measurement unit.

Optimum packaging for angular-rate sensors will remain a concern for MEMS developers well into the future. As sensor makers perfect the art of producing reliable, high-performance angular-rate sensors for automotive applications, there will be an increasing incentive to achieve high levels of integration for these components. The resulting gains in functionality and economy will foster a proliferation of advanced safety systems into the automotive mainstream, while pushing out the performance envelope for safety systems on the cutting edge.

References:

1. "Application Opportunities And Successful Commercialization Of MEMS/MST In The Automotive Market," Roger H. Grace, president of Roger Grace Associates; (415) 436-9101; see web site at www.rgrace.com.
2. "U.S. Transportation Secretary Slater Announces Advanced Air Bag Regulation That Improve Benefits And Reduce Risks," May 5, 2000. See NHTSA's web site at www.nhtsa.dot.gov/nhtsa/announce/press/pressdisplay.dbm?year=2000&filename=pr18-00.html. For more details on Federal Motor Vehicle Safety Standards for Occupant Crash Protection, see www.nhtsa.dot.gov/airbag/AAPFR/reg/.

Companies That Appear In This Report
Analog Devices Inc. Rick Sheehy (781) 461-3505 [email protected] www.analog.com/IMEMS Delphi AutomotiveSystems Milton Beach (765) 451-0655 [email protected] www.delphiauto.com Motorola Automotive and Industrial Electronics Group Kathy Farrar (847) 217-3053 www.motorola.com/ies/AIEG/ Robert Bosch GmbH Matthias Illing +49 (0) 7121 35-1558 [email protected]	SensoNor USA Hans Petersen (415) 986-6059, ext. 225 [email protected] Siemens AutomotiveCorp. Dave Ladd (248) 209-5822 [email protected] www.siemensauto.com Systron-Donner Inertial Div., BEI Technologies Brad Sage (925) 671-6403 [email protected] www.systron.com Visteon Corp. Carly Lamprecht (313) 755-0879 [email protected]