Brett Miller

Writer’s comment: I chose the topic of this paper out of convenience; at the time, I had just spent a summer researching an application for accelerometers. As it turned out, what was truly convenient was taking ENL 104E. The writing tools we learned in the class-how to think about audience, purpose, and professionalism, to name a few-have proved to be absolutely invaluable. This assignment, in particular, helped me to understand the difference between writing a basic informal lab report and writing a professional project report. I think that the sooner one can make that transition, the better. Thanks to my instructor, Victor Squitieri, for helping me to make the transition and for the exceptional teaching that inspires prized writing.
—Brett Miller



MEMORANDUM

I. OVERVIEW

This report presents the results of an investigation conducted by Miller Consulting for Vehicle Control Systems, Inc. into the relative strengths and weaknesses of various commercially available MEMS (Micro-Electromechanical Systems) accelerometers for use in a future airbag deployment system. In addition, we will make a recommendation as to the accelerometer most suited for the task.

Accelerometer suitability was evaluated in terms of several key performance characteristics: shock survivability, power consumption and zero-g bias level vs. temperature. Shock survivability was evaluated by determining the maximum powered acceleration that an accelerometer was able to withstand before damage occurred. Power consumption was evaluated under normal operating conditions. The shift in zero-g bias level vs. temperature was determined in the range of temperatures likely to be experienced during normal operation. In addition to these tests, we also took into consideration the relative costs of the accelerometers.

Four accelerometers were tested for the purposes of this investigation: Accelerometer A, manufactured by Active Semiconductor; Accelerometer B, manufactured by Global Dynamics; Accelerometer C, manufactured by Inertial Systems; and Accelerometer D, manufactured by Acme Sensing.

Conclusions and Recommendations
The following are the results of the tests for each performance characteristic. Accelerometer C withstood the highest levels of acceleration. Accelerometer A consumed the least amount of power under normal operating conditions. Accelerometer C demonstrated the least amount of drift in zero-g bias level over the tested temperature range. Finally, in manufacturing quantities, Accelerometer D has the lowest cost, followed closely by Accelerometer C, with Accelerometer A and B a distant third and fourth, respectively.

Based upon the results of this investigation, we recommend Accelerometer C for use in the airbag deployment system because of its high performance in the shock and temperature drift tests. Although Accelerometer A performed marginally better in the power consumption tests, we feel that this characteristic is least significant in terms of its design impact. Also, although Accelerometer D was least expensive, the cost of Accelerometer C was still reasonable, especially in comparison with Accelerometers A and B.

II. DISCUSSION

In this project we have sought to determine the most suitable accelerometer for use as the automobile collision-detection sensor in an airbag deployment system currently under development by Vehicle Control Systems, Inc. In this system, the accelerometer plays a key role in overall functionality. During a collision, vehicle deceleration occurs, which is detected by the accelerometer as a deviation in voltage from the nominal zero-g (zero acceleration) bias level. If this deviation is of sufficient magnitude, it will electrically trigger the subsequent ignition of a gas generant that rapidly expands to inflate the airbags of the automobile (Han).

Methods
Given the importance of both properly detecting a collision and not deploying an airbag prematurely, the choice of the accelerometer cannot be made arbitrarily. In order to determine the accelerometer most suited for use in this system, we have tested several key performance characteristics of four commercially available accelerometers.

The test for shock survivability is important in that it establishes how well a given accelerometer will perform under the high-magnitude acceleration conditions of automobile collisions. For this test, we subjected a set of five accelerometers from each respective manufacturer to ten collisions of increasing magnitude, simulating the deceleration conditions that would be experienced during crashes at speeds ranging from 5 mph to 95 mph, in increments of 10 mph. Our objective was to determine the success rate within each set of accelerometers for each respective collision test.

The test for power consumption, while not of great significance to the performance of the accelerometer during collisions, is still a concern expressed by design engineers of Vehicle Control Systems, Inc. We tested accelerometers from each manufacturer under two operating conditions: powered operation, with no acceleration; and powered operation, with a 10 Hz, 2.0 m/s/s amplitude sinusoidal vibration. The average power consumption for a five-minute period was determined in each case.

Finally, the test for shift in the zero-g bias level over the likely temperature range of operation is important because we must ensure that the final system will work regardless of the temperature of the accelerometer, which can be affected both by the external environment of the automobile as well as fluctuations in temperature due to automobile operation. For this test we monitored the zero-g bias level (measured with respect to system ground) while slowly varying the temperature in a controlled environment from -50 to 100∞C, with a supply voltage of 5V and a dynamic range of +/- 2.5V.

Results
During each successive deceleration test, the success rate of the accelerometers for each manufacturer was determined. These data are presented in Figure 1.
Figure 1
Success Rates for Shock Survivability Test

Note that data from the tests between 5 mph and 45 mph have been omitted because the accelerometers from each manufacturer demonstrated a 100% success rate. Only between 55 mph and 95 mph did trends appear in the data.

Table 1 presents the data for the power consumption tests:

Table 1
Power Consumption Test Results

Unit	Quiescent Supply Current (static)	Quiescent Supply Current (dynamic)
Accelerometer A	1.1 mA	1.5 mA
Accelerometer B	2.9 mA	3.2 mA
Accelerometer C	2.3 mA	2.5 mA
Accelerometer D	4.6 mA	5.9 mA

In our tests for zero-g bias level drift, we recorded the maximum deviations, both positive and negative, over the tested temperature range. The results of these tests are presented in Table 2.

Table 2
Zero-g Bias Level Temperature Drift Test Results

Unit	Maximum Negative Deviation	Maximum Positive Deviation
Accelerometer A	-95.3 mV	72.2 mV
Accelerometer B	-438.9 mV	474.1 mV
Accelerometer C	-53.1 mV	46.5 mV
Accelerometer D	-197.0 mV	188.5 mV

In manufacturing quantities, Accelerometer A has a unit cost of $42.10, Accelerometer B is $75.15, Accelerometer C is $18.94, and Accelerometer D is $15.33.

Analysis
The results of our shock survivability tests clearly differentiate the accelerometers. As shown in Figure 1, Accelerometer A consistently failed at a higher rate than the other tested accelerometers. Indeed, in the final 95 mph collision test, it demonstrated a 100% failure rate. These results suggest that it should not be considered for use in the airbag deployment system since the unit might fail before properly triggering deployment in the event of a collision. On the other hand, Accelerometer C consistently demonstrated a 100% success rate in all tests except the last. It is clearly better suited to withstanding the accelerations associated with an automobile collision.

The rates at which the accelerometers consume power, both under static and dynamic acceleration conditions, are very similar and likely would not have a significant impact on system design. As shown in Table 1, in all cases maximum current draw is below 6mA, which is negligible in comparison with other electrical system loads.

In Table 2, we can see that in certain cases accelerometer bias levels drifted by nearly 500 mV. Considering that the sensitivities of the accelerometers are on the order of 25 mV/g, this drift is significant and could potentially result in unnecessary or unsuccessful deployment of airbags. We saw nearly order-of-magnitude better performance from Accelerometer C over Accelerometer B. In addition, Accelerometer C performed significantly better than both A and D.

Conclusions and Recommendations
Based on our test results, we find that Accelerometer C demonstrates the greatest ability to withstand the acceleration conditions of an automobile collision. This accelerometer is also a good choice in terms of its low sensitivity to temperature changes that might be experienced under normal operation. While it was not the best performing accelerometer in power consumption tests, it was a close second, and we believe this characteristic is not significant for two reasons. First, every accelerometer performed reasonably well in this test. Second, in terms of suitability for the task of detecting a collision, power consumption is not of the greatest concern. Finally, Accelerometer C is among the lowest-cost accelerometers available, though not the lowest. It is our recommendation that Accelerometer C be used as the sensor in the airbag deployment system because of its superior overall performance in key tests.

III. SOURCE

Han, Cheol-Hyun. “MEMS accelerometers for air bag deployment for automobile.” March 2, 1998. http://www-ee.eng.hawaii.edu/~cheol/class/ee626/accelerometer.html (October 17, 2001).