This process was already developed before the concept of robustness validation were released. Nevertheless, this approach also realizes some of the principles of robustness validation. Within the MOST community this approach is well proven, and mandatory not only for the German car manufacturers but for others as well.
After having some experience with different products also coming from different manufacturers, the concept of benchmarking was introduced. The goal of this method is to compare products with functionality with the best performance. The definition of best performance is based on the requirements for the automotive application. The challenge of this approach is that all relevant aspects are taken into account.
Principle of recommendation process
The recommendation process can be divided into several steps:
- Valid specification: based on the real operating conditions and common description of requirements.
- Assessment of project: review of project flow, technical implementation, planned verification steps, quality strategy and assessment of the manufacturing facility regarding new technologies used.
- Review of measured data: the measurements are mainly done by the manufacturer.
- Judgment-based on the results of the developed investigation tool.
This paper describes the method to review the data and make an objective judgment of the tested devices. For this goal the test sequences are separated into different sections:
- Characterization respectively electrical distribution: Measurement of all important parameters over the whole range of conditions. These are normally the operating temperatures, supply voltages, and for the receiver the specified optical input power range.
- Life time tests: Examples for this section are high temperature operating life (HTOL), temperature humidity bias (THB), and temperature cycling (TC).
- Characterization after life test: Same procedure as before, performed with only a few devices that were stressed by the complete required duration of life test.
- Short time tests: Examples for this section are mechanical shock (MS), solderability (SD), and physical dimension (PD).
In each section for each single test row some statistical values are calculated to describe the behaviour of each particular parameter. An example for this calculation is given in figure 1. The criteria in this calculation are the number of failures, capability index, the distance to the specification limits, and the distribution of the measurements.
For a complete description it is also necessary to consider the behaviour over the total set of different conditions. This is true especially for the characterization and the life time test sequences. For this purpose the applicable automotive specification describes a method for calculating a design index, which is similar to the capability.
Figure 2 shows this approach for a test sequence that has several read-outs from start to 1000 hours.
The statistical characteristics of the different distributions are the relevant criterion for this judgment. For this design index a minimum value of 1.0 is required. This is quite similar to the capability index where a minimum of 1.66 is required.
Extension of the method to benchmark approach
The calculated values of the investigation are set in relation to the specification limits. Together with the automotive requirements, perfect performance is defined and set to 100 percent. With this definition it is possible to describe the real performance of the product as a percentage.
For better understanding of the benchmark the values from each single calculation were merged together to come to one single value for the judgment. This procedure is similar to the guideline for auditing of quality systems at automotive suppliers. In this approach a classification from A to C was defined to characterize the performance of the product:
- Classification A: >90 % (status: green)
- Classification AB: 80 … 90 % (status: green; improvement possible)
- Classification B: 70 … 80 % (status: yellow; improvement highly recommended) Classification BC: 60 … 70 % (status: yellow; improvement required)
- Classification C: <60 % (status: red)
The total counts of measurements are mentioned as additional information to increase the confidence of the judgment.