Professional Documents
Culture Documents
Jones 1972
Jones 1972
Jones 1972
COMPARATIVE ANALYSIS
OF
CRASH SENSORS
TREVOR O. JONES
OLIVE R T. McCARTER
ROBERT N. OGLESBY
Electronic Control Systems
Engineering Staff
General Motors Corporotion
720035
Detroit, Michigan
January 10-14, 1972
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
720035
COMPARATIVE ANALYSIS
OF
CRASH SENSORS
TREVOR O. JONES
OLIVER T. McCARTER
ROBERT N. OGLESBY
ABSTRACT
Techniques and principles for evaluating and comparing various crash sensor con-
cepts are discussed. The trade-off criteria include: crash/rough-road discrimination,
target lethality, component reliability, zone of protection, environmental suscepti-
bility, and other criteria. Within the context of hardware performance characteris-
tics, currently prominent devices are described including: inertial mass, spring mass,
piezoelectric and piezoresistive accelerometers, radar, sonar, and laser predictive
sensors, and bumper mounted detectors. An example of the assessment method is
presented, comparing an inertial mass and a spring mass detector.
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
2
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
3
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
SENSOR ECONOMIC CONSIDERATIONS assigned might change with the inputs from various
design groups.
Performance - 31%
Reliability - 41%
Economic - 28%
4
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
For the purpose of describing a simple comparative But each I Wi in equation (ii) consists of three parts,
model, it is assumed that either a pre-crash or a performance (P), reliability (Rl. and economic (E). There-
post-crash sensor will be selected; not a combina- fore, equation (ii) may be further expanded to yield :
tion of both. It should be realized that several
combinations of predictive and post-crash sensors
exist. Examples of these are radar or laser with
inertial or piezoelectric sensors. If the particular
sensor under study did in fact combine both pre
and post-crash sensing ability, then the weighting
would be revised accordingly.
5
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
6
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
Deactivatable during shipment and maintenance is of resetting after a non-crash where significant
essential to avoid unwanted deployment during input has been sensed. This restoration period must
non-crash shocks encountered in these operations. be short, and there must be no significant reduc-
Rating as follows: tion of threshold during the restoration period.
7
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
Excellent Ability to test fire and rearm auto- Excellent 30 cubic inches
matically Good 60 cubic inches
Acceptable Ability to test fire and rearm Acceptable Size to match available space, but
manually without disassembly requiring complex package shape.
3.3 Economic Considerations Power for sensor operation and squib activation is
desirable to be low to cope with the unforeseen
Government cost/benefit ratio studies (6) of deviations the power source may experience during
occupant crash protection systems indicate that the crash event. It is difficult to separate the power
the economic benefits to society will be greater requirement into sensor and squib considerations.
than the economic costs, where the economic In this evaluation, providing power continuity for
benefits include fatality and injury reduction the squib will be considered a sensor function. Rat-
effects. Similar considerations from a manu- ing as follows:
facturer's point of view would call for a compari-
son of sensor concepts in terms of adequacy of
function performed versus cost to produce. That is, Excellent Power independently stored for
within the limits of adequate function, lower costs entire activation process.
should be preferred. Good Power partially stored in an
independent source for activation
Maintenance frequency must be considered in view process.
of other system maintenance requirements. Some Acceptable Power for sensing and activation
systems, particularly stored gas, may require derived solely from standard
maintenance or inspection due to possible leak battery.
rates. It is acceptable for the sensor to be inspected
at the same interval as these other components.
Ratings are as follows: A criteria for each characteristic has now been
chosen. It remains to illustrate the depth of analy-
Excellent No inspection or maintenance sis required before weighting factors can be
required for normal life of vehicle determined, and then to assign numerical values to
Good Inspection every three years these factors. Some of the more popular sensor
Acceptable No more often than other air concepts described in Appendix A will be rated to
cushion elements exercise the methodology.
8
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
4.0 DETAILED COMPARATIVE ANALYSIS an excellent reflector to a radar signal and if blown
by the wind into a radar equipped vehicle might
Substantial depth of analysis is necessary before cause false deployment.
weighting factors can be accurately applied. Analy-
ses of this nature are illustrated in the following The problems of target lethality recognition can be
paragraphs for certain of the important sensor solved to a degree by techniques which increase the
functions: system complexity. For example, radar sensors
may be made more discriminating by including a
• Crash/Rough-Road Discrimination digital computer or sophisticated analogue cir-
• Target Lethality Recognition cuitry to permit storage of dangerous and non-
• Reliability dangerous target characteristics, or "signatures,"
• Zone of Protection for comparison with the incoming signal. This tech-
• Environmental Susceptibility nology is similar in principle to that used by the
military in its radar controlled missile interception
4.1 Crash/Rough Road Discrimination systems where decoys and incoming warheads are
distinguished. If an adequate technique can be
Inertial mass, spring mass, and piezoelectric detec- found at a reasonable cost, then radar systems act-
tors mounted inside the passenger compartment ing alone may provide a solution to the crash sen-
experience a degree of horizontal force transmitted sor problem. It is anticipated that the final answer
through the mounting when exposed to the forces to this question may be several years away.
generated on a rough road. The effect can be
counteracted in two ways: Sensors utilizing predictive detectors can also have
their target lethality recognition ability improved
• threshold escalation by the addition of an auxiliary detector of the post
• integration time escalation crash type . Unfortunately, whichever post crash
detector is selected, that detector often imposes
By raising the threshold level of a sensor, the lower the same limitations as it had when operating alone
rough-road g-levels are ignored. However, raising which can negate some of the advantages of the
the g's threshold makes it more likely that low anticipatory detector. For example, a sensor con-
speed crashes will also be masked. By lengthening sisting of a (pre-crash) radar detector and a (post
the integration time, one is saying that a crash will crash) deceleration detector would produce a
last longer than a bump. However, the longer the sensor whose reaction time was limited by the
integration time the longer will be the total sensing deceleration detector, and thereby eliminate much
time. Also, a long integration time imposes a of the advantage of the radar. The radar would
requirement on the sensor that its restoration time then serve a secondary role to the decelerator
be short, otherwise a series of short bumps on a detector to help prevent problems such as rough
rough road might look the same as a crash to the road inadvertence and provide more precise veloc-
sensor. ity of impact threshold data. Similarly, if a bumper
mounted detector is combined with a radar detec-
Another approach to the problem of eliminating tor to form a sensor, the speed of actuation will be
rough road effects without sacrificing actuation that of the bumper mounted detector.
time is the predictive sensor. Examples include:
4.3 Reliability
• Radar
• Sonar In addition to the design constraints imposed by
• Laser the need for crash/rough road discrimination and
• Infrared target lethality recognition, a detailed comparative
analysis must consider component reliability, zone
The special determinations (5) that predictive crash of protection, environmental susceptibility, and
sensors must make include: other factors. Among the most important of these
is reliability. The Reference 7 paper states that
• Distance to Object reliability is primary among the considerations for
• Relative Velocity public acceptability of passive restraint systems:
• Strength of Obstacle or Degree of
Hazard "The ultimate success or failure of
• Height of Obstacle a dynamic passive restraint system
.. . will probably be determined by
4.2 Target Lethality Recognition considerations relating to reli-
ability."
Measurement of the degree of hazard (5) of a
potential obstacle is a serious anticipatory sensor Failure to operate when needed and failure by
problem. For example, a thin sheet of metal makes inadvertent operation can be caused by a number
9
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
of factors, one of which is component failure. Very frontal. Until protection can be provided by
low levels of permissible failure rate are required: deployable passive systems for side collisions and
Let us assume that the production of vehicles is 10 roll overs , there is no need to sense in other regions
million per year, and that 3,000 hours driving time than frontal. The frontal requirement (3) is
will be averaged on each vehicle during its life time, demonstrated by driving into a standard SAE
or 30 billion hours for each year's production. If barrier with a 30° wedge (Figure 1). In addition
we were to seek, for example, less than one in- there are two special cases that affect sensor
advertent operation in each year's production, then performance:
the failure rate goal for the total restraint system
would be two orders of magnitude better than • Obstacles below bumper height that can
what is currently the best failure rate assigned for a stop the car's forward motion
single electronic terminal, or in terms of mechani- • Higher obstacles where the major con-
cal equipment, of a simple shim. tact does not directly restrain the major
masses of the vehicle
From a component failure point-of-view, inadver-
tence can be reduced to an acceptable level for Sensor performance under these special conditions
those sensors where redundancy and self diagnos- must be included in the appraisal.
tics are appropriate. Non-operation in a crash due
to component failure can be reduced to an accept- 4.5 Environmental Susceptibility
able level by a series-parallel or voting configura-
tion without greatly affecting the inadvertence There are a number of environmental conditions
rate.I Of course, any kind of redundancy increases that an automotive crash sensor might be exposed
the Icost for the total sensor beyond what it would to, depending on the mounting location and pro-
be for a single device sensor. tective hardware, and depending on the geographi-
cal area of use. Some of the environmental condi-
4.4 Zone of Protection tions are shown in Table V.
Current requirements (3) for frontal impacts
require protection against impacts ± 30° from
10
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
bridge circuit within the piezoresistive accelero- useful for comparative purposes. Examples of the
meter. Magnetic and RF fields have negligible derivation of complete figures of merit follow for
effects on piezoresistive strain gage elements. How- the inertial mass sensor and the spring mass sensor.
ever, adequate isolation must be provided against A description of current sensors under develop-
ground loops and stray signal pickup. Piezoelectric ment of the depth needed to provide a baseline for
accelerometers must be designed to protect against the establishment of numerical weighting factors is
acoustics by obtaining a resonant frequency several presented in Appendix A.
times higher than the highest acoustic frequency
expected. Otherwise, low level vibrations in the 5.1 Inertial Mass Sensor and Spring Mass Sensor
presence of high level sound might cause actuation.
Let us assume that the inertial mass sensor concept
Piezoelectric accelerometers are less susceptible to consists of a simple single level sensor with two
temperature and can usually be counted on to give detectors in series (for component inadvertence
satisfactory performance from -65°F to + 230°F. protection) and a g-level setting of about 18 g's
Those piezoelectric materials that exhibit pyro- (for rough-road inadvertence protection). The
electric effects in which an electrical output occurs schematic for this sensor is shown in Figure 2. Let
as a function of temperature change would follow us further assume that the necessary minimum of
the slow variation in ambient temperature, and as a automatic diagnostics are associated with this
result, the signals generated would be generally sensor. The inertial mass sensor is described in
below the low frequency cutoff and cause no Appendix A. The spring mass sensor concept is
problem. considered to be a restrained spring mass with a
firing pin as described in Appendix A.
Predictive sensors are sensitive to ordinary environ-
ments. For example, a radar sensor may have sub- Tables VI, VII, and VIII present characteristics and
stantial signal attenuation in rain, or if water or assessments. The particular assessments follow the
mud has condensed on the radar lens. Care must be guidelines of the above discussion utilizing engi-
exercised when designing radar systems to avoid neering judgement. It should again be noted that
susceptibility to electrical transients caused by the the values assigned are arbitrary and would change
automobile itself, to high intensity signals radiated with the inputs of various design groups.
from radio and television transmitting facilities,
and to other cars similarly equipped with radar.
Other predicitive sensors are similar to radar in
terms of environmental susceptibility.
6.0 SUMMARY
The continuing concern of General Motors to pro-
vide automotive accident avoidance and crash
protection involve the total vehicle environment
and design. Within the inclusive crash protection
category, and specifically for passive restraints, a
sensing device is generally required that rapidly
signals that a crash has occured, or is imment.
ACKNOWLEDGEMENT
" I ncluded for purposes of this example as "1 .0" without
substantiating criteria.
The authors wish to acknowledge the support and
Note that the total figure of merit of 72% for the dedication of their colleagues at General Motors
inertial mass sensor is greater than the figure of Corporation for their help in evaluating and
merit of 51 % for the spring mass sensor. However, developing crash sensor systems. Special thanks are
the assessment potential for each sensor may due to members of the Engineering Departments of
change, and this example of methodology should Delco Electronics Division, Fisher Body Division,
not be interpreted as necessarily excluding spring and R.A. Dork of Engineering Staff, General
mass sensors from further consideration. Motors Corporation.
12
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
13
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
Sonar
14
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
New Text
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
An acoustic sensor may have its target signal It is obvious from these three parameters, that an
reduced to a non-critical level by a 10-15 mph acoustical system cannot assess target mass and
wind (Fig. 13), or by certain humidity conditions hardness. For example, a medium sized dog might
(Fig. 14), and is very susceptible to noise. The "look" about the same as a tree, cardboard nearly
three parameters that characterize targets to an the same as concrete.
acoustical system are:
Laser
• Size
• Reflectivity (or absorptivity) Essentially, a laser sensor utilizes coherent light but
• Roughness is functionally much like a radar sensor. One of the
16
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
Combination Sensor
17
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
18
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
19
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
20
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
21
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
22
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
23
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
24
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
25
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
26
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
Trigger Time vs. Weight of Seismic Mass These runs were made for a device with the follow-
ing parameters.
In another set of runs, the weight of the seismic
mass was varied and the time to trigger (for .lS75 An Sg, SO msec half sine acceleration input was
inch travel) noted for an Sg, SO msec half sine wave inserted and trigger times for .125, .250, and .50
input with spring preloads of .25 and .3 lbs. These inch travels were noted for preloads from .0 to 6g
results are shown in Figure lS. As the mass is (the mass traveled less than .25 inch in the 6g
doubled (from .1 to .2 lb), a fairly significant case). It is desirable to keep the preload as low as
decrease in trigger time of from 6 to S msec is possible to reduce trigger time. However, since the
noted. However, as the mass is further increased, friction level is fairly high for this particular sensor,
the reduction in trigger time becomes fairly insigni- this represents the limitation on reducing the
ficant. preload, since the spring must overcome the fric-
tion in returning the mass to zero for non-firing
Effect of Spring Preload on Trigger Time cases. The results are plotted in Figure 19.
27
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
REFERENCES
1. W. D. Nelson, "Lap-Shoulder Restraint Effectiveness in the United States,"
Society of Automotive Engineers Report No. 710077, January 1971.
2. R. A. Rogers, GM-ADAP, "An Efficient System for the Collection, Storage, and
Retrieval of Accident Information," August 1969.
28
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018
Downloaded from SAE International by University of Bath , Sunday, September 23, 2018