Application of degradation test data to advertisement of consumer electronic products

Abstract

Two kinds of reliability tests (accelerated life test and degradation test) are often performed in the process of developing a new electronic product in order to guarantee its improved performance over the lifetime. As a result of such tests, several reliability indices (parameters) can be estimated. Advertisers of a new product can utilise one of these indices as a benchmarking point against the existing products. Copies in advertisement can be typically very short and may only deliver information based on the point estimate of a parameter or even in terms of lifetime, itself, which is a random variable. In this paper, we show how the advertisement can mislead potential customers in terms of their perception in the reliability of the product. Additionally, we propose the way to identify the lifetime distribution based on the degradation test of an electronic item.

Introduction

Importance of reliability improvement at the design stage has been increasing rapidly that many interesting design concepts have been evolved recently. Life or degradation test is often performed to assure that a certain level of reliability is attained at each design review stage. As a result of the reliability test, several indices (e.g. MTTF, variance, percentile lifetime) can be obtained, which can be used as benchmarking parameters for reliability comparison. Advertisers often utilise one of these indices to attract potential customers for their products over others.

However, the copy used in the advertisement can be typically very short and may only deliver information based on the point estimate of a single reliability index instead of its confidence interval. Use of a point estimate does not reflect the variability in a sample and may not be sufficient to draw any significant conclusion for comparison. In a worse situation, advertisement is even stated in terms of a random variable instead of a parameter (e.g. longer lifetime instead of longer MTTF).

In this paper, we show how this kind of advertisement can mislead potential customers in terms of their perception in the reliability of the product. We take a case of reliability advertisement, which was claimed to be false or exaggeration by the customers' right protection committee. We show how this claim or advertisement can be reputed or supported through a proper data analysis procedure.

Electronic manufacturer ABC upgraded cathode ray tube (CRT) by replacing the cathode material from the existing one to the new and advertised that the new CRT would have three times longer lifetime than the existing one. We do not reveal the real name of the maker in consideration of confidentiality. Customers who purchased the TV equipped with the new CRT experienced some early failures of CRT and claimed that the advertisement is false.

As a researcher group, we had a chance to analyse the reliability test data that the set marker provided as evidence of their advertisement. At the stage of new product development, it is rare to observe the failure data during the normal reliability test. Therefore, instead of life test, degradation [1] test is often performed under the use condition and the lifetime can be indirectly estimated based on the degradation data. The accelerated life test (ALT) can be used as an alternative to estimate the lifetime [2], [3], [4], [5], but ALT has many problems when projecting the test result to that of use condition.

The nature of the degradation test for a CRT under use condition is as follows. Select CRT samples randomly from each lot where some lots are made by the existing material, while the others are of new material. Assume that the proxy variable for the CRT performance is Maximum Induction in Cathode abbreviated as MIK and measured in μA. The life of a CRT is defined as the time by which its performance falls down by a certain percent, say 50%, of the initial level of MIK. In order to estimate the lifetime of each CRT, one can carry on the degradation test by taking repeated measurements of MIK changes of a CRT over age under the use condition. The initial MIK is observed for red, green and blue of the electron gun of each sample, respectively, and their average is recorded, when the fresh product sample is brought to the degradation test [6], [7]. This procedure is repeated intermittently as time goes on and the changes in CRT performance taken as MIK are recorded repeatedly. When a certain amount of degradation data is accumulated, the lifetime of product i can be estimated by fitting the relationship between the MIK degradation and the age of the CRT. Estimated lifetime, in turn, can be used as a basis of lifetime data analysis. Once, distribution is identified for the new and existing product, some comparison can be made on the specific reliability indices.

In the next section, we use parametric data analysis to show how the degradation test result can be used as a basis of reputation or support of the advertisement.