Signality Solutions

Failure Modes Effects Analysis

For safety critical applications, you need to know what happens when your circuit fails.

Using modelling and simulation as an integral part of Failure Modes Effects Analysis (FMEA) can help to significantly improve the safety and reliability of a product right from the start of the design

Simulated circuits can be tested to - and even beyond - destruction. With no physical components to be damaged in the process, detailed FMEA can be carried out from the earliest stages of a design. Safety and reliability improvements can be made as the design is being developed instead of waiting until first prototypes are available or trying to introduce them to products already in production. This saves design and test time as well as the cost of units damaged or destroyed during testing.

FMEA in simulated circuits removes the reliance on the original designer's expertise and in most cases the guesswork, in discovering exactly what the failure effects are. This can be an important consideration if the product is already in production or access to the original design information is restricted for whatever reason.

Testing through simulation can also be very much faster than physical testing because multiple failure modes for each component can be introduced and simulated in seconds.

At the same time simulation based FMEA can be more revealing than physical testing because some failure modes may be very time consuming - or simply impossible - to introduce through physical fault insertion tests. Faults can be introduced into simulated circuits in ways that more closely replicate how they would occur in real applications. For example, it may be hard to introduce a slowly drifting parameter change or a shorted component without so severely compromising some other aspect of the circuit's performance and behaviour that it calls into question the validity of the outcome of such a test. Physical fault insertion testing may only be possible by turning off the power, introducing the fault and then turning the power back on again. This may give a very different result from that obtained were the same fault to occur in a unit in the field with an uninterrupted power supply.

Through detailed understanding of the circuit to be analysed often obtained as part of an independent design review, it is possible to develop modelling strategies that can significantly reduce simulation times. It then becomes possible to simulate, in just a few seconds, multiple failure modes of a component - even in highly complex switching circuits - over real time periods of tens or even hundreds of seconds.

One example of this was for a circuit powered by a switched mode supply, transferring dc power and control signals across an isolation barrier using precision synchronous chopper/rectifiers whilst also providing isolated power through a second switched mode supply. Before introducing any fault, it was essential that the circuit operation had settled into a steady state, which in this instance was about 10 seconds after being switched on.

With the client's initial modelling of just one of the chopper stages taking over two hours to simulate 50ms of real time, detailed analysis of the circuit function led to the development of a hierarchical approach to modelling that simulated real times of over one hundred seconds in roughly one second per failure mode. With the analysis of the results of each simulation run taking only a few minutes each, reducing the simulation time by a factor of over 10⁶ meant that every failure mode of all (roughly) 200 components in the circuit could be analysed in about 17 hours.

In this particular example, the analysis was carried out on a product already in production with only minimal input needed from the original design engineer. This not only saved their time but also several hours of insertion testing on the bench plus the cost of scrapping several units usually damaged in the process.

Links to simulation files and results can be pasted into database and analysis tools such as Item Toolkit to help create concise reports with clear and understandable content.

In another example, a design flaw was exposed in an existing product which could have been avoided if this part of the circuit had been simulated at the design stage.

Applying simulation in this way can provide insight into - and remove much of the uncertainty from - failure effects. It can also provide detailed information that can be used in conjunction with reliability tools based on standards such as IEC 62380 to significantly improve the reliability of a product.

For instance, in the event of a short circuit failure of the high side switching transistor in a step down power supply, you can see if a MOSFET in series with the output and controlled by an over-voltage detection comparator, really does turn off fast enough to protect the downstream circuit from an over-voltage spike that even momentarily exceeds the absolute maximum VCC ratings of the components used.

The simulation of a short circuit failure of the output of one of a pair of current sharing power supplies OR-ed together through low forward drop Schottky diodes may show that the leakage current through the suddenly reverse biased diode is so high that the power dissipation will cause it to go into thermal runaway and short the output of the remaining supply.

If you don't just need to know how your circuit works but what it does when it fails then just contact us.