Technicians at the European Space Agency (ESA) spent months trying to work around the problem. But in early August last year, the satellite operator decided it was time to call it a day on the Copernicus Sentinel-1B mission and wait for its replacement to launch.
The cause? The most likely culprit is a combination of a multilayer ceramic capacitor (MLCC) and the way it was soldered onto the PCB of a critical power unit. There is a chance its partner, the Sentinel-1A, has the same latent flaw though that satellite has now surpassed its planned mission lifetime.
Though PCB flexing is a common cause of damaging cracks in MLCCs, the cause of the Sentinel-1B problem looks as though it may have been caused by the thermal shock of the manual soldering technique used when this particular device was used to replace a component already marked as potentially problematic. Sentinel-1C, due to launch in a matter of months, was made using different procedures.
The stresses of space are unusual but the kinds of tests and analysis this industry performs to avoid in-flight problems seem likely to expand into other areas as the electronic circuits they use – and the passives used in all of them – become more pivotal to their operation.
At the Passive Components Networking Symposium in 2021, Würth Elektronik CTO Alexander Gerfer pointed to changes in automotive design putting a much greater focus on the way in which passive components need to be tested and analysed not just before they are put into the bill of materials but also how evaluated after assembly.
The electric vehicle (EV) presents a major and possibly unexpected challenge. Driving is already stressful on electronic components close to the engine and driveshaft in conventional vehicles with internal combustion or hybrid engines.
With EVs, experts expect reliability to be less of an issue overall because the drivetrain is a lot simpler mechanically. However, for the electronic circuits controlling everything, long-term stress will likely increase. Petrol-driven cars spend much of their time parked and dormant. Autonomous driving and sharing schemes point to vehicles spending many more hours of their lifetime on the road, servicing different users. And, with an EV, the power systems are active over a longer period because they will be recharged when parked.
Above: The Copernicus Sentinel-1B mission had to be abandoned after a problem with how an MLCC was soldered onto a PCB
Design changes
There are other changes within the designs, such as increases to voltage in the power electronics to improve efficiency.
“More high voltage testing is definitely being done today as EV systems operate at 400V to 800V and even ‘low voltage’ automotive systems are moving from 12V to 48V,” says Philip Lessner, executive vice president and CTO of Yageo Group. Another change lies in the increasing use of wide-bandgap semiconductors, which increase some circuit stresses. “We are being asked to characterise components under high ripple and high transient current conditions.”
This may have other ramifications for reliability, says Ron Demcko, senior fellow at Kyocera AVX. “The higher-voltage bus trend, as it transitions from 12V to 48V, may change the power distribution topology and layout and could possibly limit some SMT device sizes and traditional layout schemes.”
The question posed in 2021 was whether existing testing schemes could keep up with the changes in EV. As they have turned to using more commercial off-the-shelf (COTS) components, many of which have been tested according to the automotive AEC-Q200, space electronics companies often supplement these tests with their own analyses.
Analysis can be difficult and time-consuming, and not always successful. Physical sectioning is a common technique for investigating the causes of internal failures.
Thales Alenia Space reported at the 10th Workshop on Electronics Materials and Processes for Space in 2019 how its team had assessed the contribution different materials and construction contributed to the likelihood of cracking occurring. It was more prevalent in some dielectrics and the work also identified important differences between manufacturers once external effects such as soldering practice were considered.
Because physical sectioning can cause cracks itself, which make analysis more difficult and destroys the components, some have taken the step of bringing in X-ray equipment.
This risks damaging parts as well through radiation absorption but could potentially be used in manufacturing as a screening step or at least gauge how design changes can help bolster reliability. Several years ago, ABB used X-rays in a project to identify how orientation and position affects the likelihood of cracking and gauge the technology’s usefulness for in-line testing.
Bolstering testing
In practice, bolstering the battery of tests performed for standards such as AEC-Q200 looks most likely to provide the way forward.
“Design requirements and stress-based tests, such as AEC-Q200, have significantly contributed to the advancement of a whole class of passive electronics that exhibits consistent levels of acceptable reliability,” Demcko says. “The expansion of these requirements is likely to be an effective first step towards further component reliability improvements.”
Lessner says the focus of testing is changing. “In the past, testing was mostly done to meet certain limits at a certain time. Now there is much greater interest in predictive life modelling for the life of the application.”
Such tests might involve replicating the equivalent of hundreds of thousands of hours under a variety of temperature, voltage and humidity conditions in order to develop a model that the customer can use to predict component lifetime under their specific use conditions.
Lessner points to the K-LEM model subsidiary Kemet developed for film capacitors as an example.
The Yageo group also uses physics-based models, using data derived from experiments, to help predict where hotspots might form, as well as other properties that can affect performances. Such experiments and simulations will help drive improvements in reliability.
“New components, material systems and designs will likely be key to improving the reliability of the electronic content within EVs. These efforts are vital since electric content is increasing so rapidly. In fact, reliability improvements are required just to be on par with the overall system reliability levels of less complex systems with dramatically fewer passive components,” Demcko says.
“Luckily, existing material systems that are qualified to AEC-Q200 are already optimised from a device physics and reliability perspective. So, the first sets of new components will likely be based on new configurations of these existing materials.”
Though construction analysis and simulation will help pinpoint problem areas, testing will stay the main weapon in determining lifetime reliability for most. Demcko sees an expansion of test regimes to cover new areas, seeing an increase in “the requirements for existing tests as the first step”. That may mean requiring a larger number of temperature cycles in a test to gauge how well components will fare in a typical EV application.
Tests could be extended to more stringent electrostatic discharge levels, Demcko notes, among others. “As the desire for higher capacitance density in MLCCs grows, it’s worth considering requiring DC bias testing to help designers understand capacitance stability as a function of applied signal, which is important since ripple voltage requirements will tighten with reduced rail voltages.”
The key to the future of reliability will be in aligning this increasing battery of tests to the requirements of the target application, as Lessner notes, “There is also much more application-specific testing being requested by the customer. But, the ‘bread and butter’ tests, as defined for example in the AEC-Q200 standards, are still the ones most often requested.”
