A Reliability Assessment of Automotive Electronics 741219
Electronic reliability, as related to the automotive system, is discussed in four parts: the dual requirements of high reliability and of low unit price imposed on the semiconductor industry by the automotive system; how semiconductor device reliability has been and is being raised to levels of reliability commensurate with the needs of the automotive system; some suggestions as to how the semiconductor industry and the automobile industry might work together most effectively; and the overall situation as the semiconductor industry sees it.
The automobile system represents a stringent environmental requirement for semiconductor devices. Reliability-wise, the requirements are not too different from those of military- and space-use situations. The automobile system imposes on the semiconductor industry two diverging requirements. One is a rigorous level of device reliability and the other is a demandingly competitive price figure.
Reliability data is shown for discrete devices (transistors, power devices) and for integrated circuits. These data show that semiconductor devices have adequate reliability at present for use in automobile systems.
Field performance data are given for two specific applications of Texas Instruments electronics to the automobile system - the skid-control module and the seatbelt-interlock module. From April 1970 to January 1973, Texas Instruments supplied skid-control modules to a segment of the automobile industry. On a data base of 229,000 units, only 14 returns are known to have been made for electronic defects. This is a reliability performance of 99.9925%.
Texas Instruments has been supplying seatbelt-interlock modules for use in certain automobiles for about a year. Customer's reliability goal for the first year's production was 99.00%. During ten months' use in the field, the reliability of this module was 99.95%.
The triple characteristics of semiconductor failures with time are: early failures, random failures, and wearout failures. Overall reliability can be increased by use of test screens to remove early failures. Wearout failures have been seen only in testing, not in field use.
Two points should be made with regard to semiconductor reliability. One is the steady increase in the reliability of semiconductor devices over the years. The second is that step-function gains in reliability will be made when process technologies, now restricted to manufacture of ultrahigh reliability devices, can be made sufficiently economical to be used in the manufacture of semiconductor devices for the automobile industry.
Good communication is imperative if the semiconductor industry is to translate quantitatively and correctly the reliability needs of the automotive system into those semiconductor screening tests used to achieve definite levels of device reliability. This will require a high degree of technical interchange between the two industries. The semiconductor industry must understand fully the automotive industry's cost and use situations while the automotive industry needs to comprehend in detail the cost implications of the various levels of semiconductor device reliability. This latter task cannot be done effectively in an arm's-length relationship but by teams from each industry working together to gain a thorough understanding of each other's requirements. Experience has shown that working together is necessary because the best-intentioned and best-written semiconductor-device specifications cannot alone assure that the subsequent device will perform satisfactorily in the automotive system.
There is no doubt that the semiconductor industry has achieved or can achieve the reliability level required by the automobile system. The challenge is to produce this reliability within the cost constraints of the auto industry. These cost constraints have two aspects: initial cost, the price paid for a given semiconductor device; and replacement costs.
It is in this area of cost-reliability trade-off that close and effective dialogue between technical representatives of both industries will pay off for the automotive industry. The automobile-systems engineer who can knowledgeably relate his system-use environment to the various levels of severity to which semiconductor devices are tested will be able to make an intelligent trade-off between initial device cost and subsequent reliability problems when the car is in a customer's hands. The cost of system repairs in the field can quickly negate apparent savings made in the procurement of devices whose reliability levels are below those required by the system. Equally intolerable, but even more frustrating, is to have paid a premium price for a semiconductor device only to have it show unsatisfactory reliability in the automobile because of inadequate communication and understanding as to the way in which the device was to be used.
Together, the automotive and semiconductor industries have the elements needed to attain increased reliability of the automotive system through wider use of electronics. The proper melding of these elements is critical. It will not occur automatically. It must be done by meaningful dialogues and information exchanges between the technical people of both industries. The net result must be a clear mutual understanding of each other's capabilities and boundary conditions.
These technical interactions will work to enhance the reliability of the products of both the automotive and semiconductor industries. An engagement with the auto industry across a broad spectrum of semiconductor device technology, we are confident, will result in further improvements in the present reliability of the automotive system.