Cutting corners on capacitors may save dollars today, but it risks space equipment tomorrow
by Daniel West, Lead Technical Applications Engineer, KYOCERA AVX
There’s a growing momentum among some members of the space industry community to push less-than-space-grade multilayer ceramic capacitors (MLCCs) to their limits to reduce the costs associated with entering space.
The commercialization of space and recent success stories make it easy to adopt a BOM with components rated for automotive applications and even readily available commercial-grade components to accommodate projects with extremely short design cycles like LEO constellation satellites. What makes it even easier to adopt these components is that the relative price of an automotive-grade capacitor compared to a space-grade capacitor can differ by one or two orders of magnitude.
When we consider the emerging power demands of AI processing cores and the amount of capacitance required for decoupling, the costs go up dramatically. It’s completely understandable why this NewSpace (private spaceflight) trend of pushing readily available commercial/automotive-grade components to their limit is growing and seems to be proliferating to the general space community as well, but it’s also understandable when suppliers won’t accept the liability of launching those parts into space.
This trend should prompt questions from NewSpace component engineers to high-reliability MLCC manufacturers to determine what is truly purchased from one level of reliability to another, but it can also prompt requests for an expedited root cause analysis of failed parts, followed by questions to determine what actually separates different tiers of reliability. This request for failure analysis may be rejected because commercial- and automotive-grade parts typically do not require burn-in. Since burn-in is intended to filter out parts with latent defects that can go undetected with standard electrical testing, it would be a waste of resources to test parts that accept a much higher early-life failure rate.
For those looking to use non-space-grade components for their space applications, it’s worthwhile to bring up some additional items that will foster meaningful discussions with ceramic capacitor manufacturers.
The effort to characterize the reliability of an MLCC is extremely difficult and time-intensive. Additionally, if there is a design change or equipment optimization for a given MLCC design, the tests should be run again. Much of this effort has already been undertaken by manufacturers specializing in supplying components for high-reliability applications.
High-reliability devices are typically built to a MIL-PRF-XXXXX document, a DLA/DSCC drawing, or the qualification standards of a space agency, such as NASA/ESA. There’s a lot of paperwork involved, but these documents essentially require manufacturers to comply with a list of tests to be an approved supplier. The sample sizes for these tests are typically much larger (if not 100%) than other reliability grades and have stricter pass/fail criteria than automotive- and especially commercial-grade parts. A wide range of tests can also be added, removed, or modified from these documents to accommodate custom designs, known as Source Control Drawings (SCD). Some special test examples would be radiation testing, cryogenic tests
(< 77 kelvin), tin whiskers, out-gassing, hermeticity, and harsh shock/vibration tests.
Automotive-grade reliability is the next significant reliability level below MIL/Space/Aerospace components. These MLCCs are designed to meet the requirements outlined in the AEC-Q200 document and are also intended to support an industry with extremely large volumes. As a result, MLCC designs are optimized to strike a balance between yield and quality, while remaining competitive in the market. For example, life tests for automotive-grade parts last 1,000 hours, which is the same as those for commercial-grade components. The availability would plummet if they complied with 4,000-hour life tests outlined for high-reliability grade parts, not to mention the increased costs associated with “keeping the lights on.”
Commercial-grade reliability is where the majority of the MLCC global volume resides. Billions of people hold a few hundred MLCCs in their hands every day when looking at their mobile phones. Testing on these components is limited to small sample sizes and to check for basic electrical, thermal, and environmental performance. Cost efficiency and availability are paramount over reliability; therefore, limited notifications are sent to customers when a design change occurs. Capacitance, dissipation factor, and insulation resistance are the primary electrical parameters checked when inspecting commercial-grade MLCCs. Again, burn-in is not performed, meaning a larger ratio of parts with latent defects will be shipped because this is acceptable for short-life or redundant applications.
Between commercial-, automotive-, and space-grade designs are huge jumps in requirements that are pushed to suppliers for monitoring process inputs that are impossible to see with the naked eye. The subtle differences are even difficult to discern with comprehensive DPA (destructive physical analysis) and imaging technology, unless you know what to look for.
Take, for example, the KGP (commercial-grade) and DSCC 25007 (high-reliability-grade) stacked MLCCs from Kyocera-AVX. Both use discrete MLCCs attached to a lead-frame to increase capacitance. They look very much alike and provide identical circuit functions, but the DSCC 25007 is designed to provide many years of use with confidence in adverse electrical, mechanical, and thermal conditions, whereas the lifetime of commercial-grade products relies on multiple factors at a much lower confidence level.
Commercial-grade MLCCs typically qualify for lifetime use after a 1,000-hour test at rated voltage and temperature. You can derate them and use Arrhenius to estimate an extended lifetime, but the results will vary significantly from lot to lot. You’d also have to find a supplier willing to determine accurate activation energies from HALT (Highly Accelerated Life Test) analysis for commercial-grade parts, but the likely response will be that they send you to their space-grade parts, where this request for HALT is more reasonable.
NewSpace trends, including short design cycles, cost reduction, and a higher tolerance for asset loss in non-critical missions, have revealed that there is no fully dedicated reliability grade of MLCCs for this emerging industry. Some passive electronic component manufacturers are responding to this trend with enhanced COTS (Commercial Off-The-Shelf) components. These are defined and handled differently among suppliers because there is no independent governing agency; therefore, a close conversation with them is needed to determine if they’re performing burn-in on commercial/automotive-grade devices, conducting statistical screening of standard tests, using a unique design, or simply limiting tests of high-reliability devices to meet demands. With this information, a cost-versus-risk analysis for a NewSpace project would be much more accurate.
www.kyocera-avx.com