At Alter Technology, we know that in the space sector, failure is not an option. Reliability is not a buzzword, but rather an engineering imperative. To guarantee that a satellite will succeed and achieve its goal we must assess and guarantee the long-term survival of the products used.
In this context, we need to define two key concepts: quality and reliability, which are related but are not the same.
- Quality focuses on the performance at a single point in time, usually the start of the process. It basically relates to the conformance of the product to the specifications agreed. It is measured in the defective parts per million (DPPM).
- Reliability, on the other hand, addresses the performance of the product over its expected lifetime. As such, it is mainly a probabilistic, statistical measure that is built into the design. There are several ways of measuring the reliability, but the main ones are FIT (failure in time) or Mean Time Between Failures (MTBF)
Quality
- Focused in the performance at one point in time (start of the process)
- Assures conformance to specifications
- Critical aspect of the whole reliability process
- Addressed by testing
- Measure in defective parts per million (DPPM)
Reliability
- Focused on the performance of the product over expected lifetime
- Based in probability /statistics
- Built into the part. Testing is a risk mitigation
- Measure in FIT or MTBF
In summary, quality is a snapshot; reliability predicts how the product will age and fail under stress
However, most space components and products are not produced in high enough volumes so that you can use statistical analysis to calculate its reliability. In order to overcome this issue use tests. Specifically, accelerated tests are used, in which the product (component, board, assembly, equipment) is subjected to higher stresses that it would see during mission, so that the test can be shorter.
One the most common tools used for these purposes is the Norris-Landzberg equation, which help us define and create thermal cycling tests to estimate the thermomechanical fatigue of electronic assemblies. It was initially developed for tin-lead solder joints. This law is derived from the Coffin-Manson thermal fatigue law. The equation governing this law is the following
Where
- f0 cycle frequency (cycles per day) during mission
- ft cycle frequency (cycles per day) during test
- ΔTt is the change in temperature during tests
- ΔTt is the change in temperature during the mission
- T0 is the maximum temperature of the cycles during the mission
- Tt is the maximum temperature of the cycles during the test
- m and n are constants that depend on the materials
- AF is the acceleration factor, which is basically a parameter that allows you change the number of total cycles between mission and test. It basically allows you to make and comparison between how many high-stress cycles on the bench equate to years of in-orbit conditions
One of the most important thing about the Norris-Landzberg equation is that it is required by ECSS in many calculations. For example, ECSS-Q-ST-70-61 and several ESA Memos indicate that the use of this equation is mandatory in order to assess if the qualification perform of PBAs covers the environment of the mission. Similar methodologies are widely referenced in MIL and NASA standards
Finally, it is worth noting that this model is not perfect at it simplifies many aspects of thermal cycling, such as dwell time, temperature ramp. It also works better with SnPb solder joints using the m and n constants.
In summary, using physics-of-failures approaches during the definition of the reliability tests needed for space applications is essential to be confident in the performance of the product during its lifetime. The Norris-Landzberg equation is a great example of this that has been used many times in ESA missions
