The RECOM Three Year Warranty – A Guide to the Essentials

Part of the answer is past experience with similar product families that use the same type of construction and share common components. We know for instance, that the RMA (Return Merchandise Authorisation) rate for some series is very low indeed. For example, the R1S 1W isolated SMD DC/DC series has less than one RMA per million pieces sold. So when RECOM released a similar 1W product (R1SX) that shared the same topology and used the same QC procedures, although it has an improved automated manufacturing process, the demonstrated reliability of the original 1W design could be used to justify the same warranty period for the new lower-profile 1W series.

Another way of predicting reliability is to use the calculated MTBF (Mean Time Between Failures). This is a statistical analysis based on adding up all of the reliabilities of the individual components that make up a converter and then applying different accelerating factors that could influence the overall reliability such as Environmental Stress (temperature extremes, shock and vibration, high humidity, etc.), Functional Stress (interaction of components, poor tolerances, on/off switching cycles, etc.) and Maturity (well-known and tested design or a new approach). One example of such an approach is the US Army’s Military Handbook – Reliability Prediction of Electronic Equipment, better known as MIL-HDBK-217F, which is essentially a long list of component types with their historically demonstrated reliability figures. The calculated result can be a million hours or higher.

An MTBF of a million hours does not mean that the converter will survive 114 years! MTBF is a statistical measure of how reliable a product is during its lifetime which is set by wear-out mechanisms. The way to interpret an MTBF figure is to work backwards from the desired failure rate. If it acceptable for 1% of the converters to fail after three years of constant use, then the required MTBF level is:


Taking the R1SX product as an example, it has an MTBF of over 21 million hours at room temperature, so the statistical failure rate of a single converter would be 0.125% over 3 years of continuous operation, or an annual failure rate of one converter out of a population of around 2400 pieces.

Does this mean that if the MTBF figure is lower than three million hours, that every year around 1% of the converters will fail? Not so in practice. Most converters are not 100% loaded during normal operation. If the static load is 50%-70% of the maximum rated load, then the stress is less, MTBF increased and fewer parts will fail per year. If the converter is only powered up during the working day, then that alone will reduce operating hours and fewer parts will fail per year. On the other hand, if the converter is used in a “24/7 always on” condition, then the internal temperatures will be stable and the electrical stress caused by switching on and off the converter is eliminated, also increasing reliability. The calculated MTBF hours represents worst case, constant full stress operation. Therefore, RECOM also offers its standard three-year warranty on product series that have less than 1 million hours calculated MTBF figures, based on additional testing and evaluation (see next chapter).

We have said that the three-year warranty can be justified by past experience based on our decades of manufacturing experience and a knowledge of how the converter will be used. However, how can we be sure about a completely new product, where we do not have any equivalent production Failure Mode And Effect (FMEA) data to fall back on, no market feedback of how the product will be used, and when we employ a novel production process? There are four techniques that can be used – DVT, PVT, HASS and HALT-testing.

Design Verification Testing (DVT) checks the datasheet operating limits of each and every component used in the design against the actual operating conditions. If a component such as a resistor or diode has a maximum operating temperature of 125°C given by the manufacturer, then the design rule is not to exceed 115-120°C in worst case operation (for example, full load at the maximum ambient temperature) to give a certain margin of safety. Capacitor lifetime is known to be dependent on the voltage across the part and the ripple current through it – ideally these should not exceed 70% of the maximum rating. Electrolytic capacitors are often considered the weak link in the chain of reliability, but by derating the operating conditions (effectively ‘over-designing’ the component), calculated lifetimes of more than 20 years can be achieved along with high reliability during that lifetime. Similar design rules apply to the operating conditions of active components and to the core temperature of the transformer under full load. DVT ensures that under extremes of temperature, voltage and current, margins of safety are maintained by design.

Production Verification Testing (PVT) goes one step further and takes samples of production batches to check the actual performance against the DVT design limits. Recom typically tests a minimum of 50pcs of each new converter part to give reasonable degree of confidence in the accuracy of the measurements, then uses the data to generate average and maximum deviation figures for efficiency, regulation and output accuracy tolerances for the datasheet. This pre-qualification testing is repeated for different input/output combinations to give a realistic analysis of the product series as a whole. This involves many thousands of different test conditions and the fully automated test set-up designed and built by RECOMS’s Test and Evaluation Laboratory greatly accelerates this process by, for example, cycling through input voltage and load conditions, then collating the data and generating the required performance graphs automatically.

In addition to electrical performance testing, climate chambers are also used to perform Highly Accelerated Stress Screening (HASS). This involves long-term tests at extreme hot and cold temperatures, temperature cycling or a combination of high heat and high humidity under various operating stresses such as maximum or minimum input voltage, full load, power cycling, etc. These tests accelerate the aging effects on the converters and allow us to simulate the result of many years of continuous operation in 1000 hours of testing.


RECOM also has in an in-house HALT (Highly Accelerated Life-Time) test chamber. This high performance HALT climate chamber subjects the DUT to extremes of temperature and abrupt temperature changes (liquid nitrogen can be used to rapidly cool down the DUTs inside the oven), while the parts can simultaneously be shaken on a random vibration table. HALT is a process to find any design weakness that could reduce the lifetime or reliability of a product by testing to failure using a combination of stress conditions and by exceeding the datasheet specifications to destruction (unlike HASS which stays within the operating limits of the DUTs).


Acceleration Stress factors include vibration, thermal stepping and cycling, high humidity, over-voltage/over-current and whatever other stresses that can be devised that could l expose weaknesses in the product. Combined stress environments such as vibrational stress under temperature cycling may also be necessary to compress the time-to-failure. Studies have shown that around 70% of failures can be detected with a mechanical stress (all-axis vibration) in combination with temperature stress, based on the relationship:

Where D is accumulated fatigue damage, n is number of stress cycles, and σ^α is the mechanical stress with an exponential factor depending on the severity of the vibration. To determine the general robustness of our products against mechanical shock and vibration, we also have two shaker tables (one in Vienna, the other in Recom sister-company PCS in Italy) that can be programmed to deliver sinusoidal vibration profiles with programmable displacement and g²/Hz accelerations for products weighing up to 130kg.


The use of mechanical stress testing can not only expose weaknesses in a completely new design, but in combination with HASS and PVT results, can be used to give credibility to a warranty period of three years for any existing product.

RECOM prides itself on designing efficient, cost-effective and reliable AC/DC and DC/DC power supplies, whether they be off-board or board-mounted. While all manufacturers say that they supply reliable products, RECOM backs up this claim with an across-the-board minimum 3-year warranty on all of our products. We use extensive verification testing, production QC and accelerated stress testing and screening to ensure high reliability and long operational lifetimes and to provide datasheet information that is accurate and based on documented test protocols.