Engineers who study data sheets carefully when they are evaluating an active or passive component for possible design-in, know that a good data sheet is a treasure trove of information beginning with the obvious facts. The best data sheets include supplemental information illuminating how best to design with the product, hopefully including important subtleties. One of the many sections at the front of these data sheets is a table labeled “absolute maximum ratings” (AMR) and it is easily overlooked or put on the “I’ll look at it later” list. After all, what’s to worry about if you are just using the part in a plain-vanilla mode?
Consider paying more attention to that information — subtleties can sometimes be found here. AMR numbers cover electrical overstress (EOS) and its subset of electrostatic discharge (ESD), temperature extremes, and other factors, as well as their interaction, Figure 1.
Especially for power, graphs showing the safe operating area (SOA) — which bounds the acceptable operating conditions before bad things happen — are vital, Figure 2.
AMR brings significant issues, questions, and considerations especially for power devices. That’s why I was pleased to see a two-part article on it in the always-interesting publication In Compliance (see References). It’s worth reading even if you don’t think AMR is a big deal in your design, as the articles raise many important considerations when looking at (or ignoring) AMR data. The article even discusses industry guidelines and standards that are used by some vendors when establishing AMR numbers. That was a real eye-opener, as they say.
What are these bad things? They actually can cover a range of problems including transient or permanent degradation of some performance specifications, stresses that don’t affect performance now but will diminish longer-term reliability, or even immediate failure.
At the same time, some designers consider AMR numbers to be general suggestions or guidelines rather than absolutes, thinking maybe it’s OK to exceed them just a little bit, or for just a tiny amount of time. After all, what’s the difference between hitting an AMR number of +175⁰C versus going up to +176⁰C? Are the consequences of exceeding AMR somewhat gradual or perhaps linear? Is there an inflection point or threshold at which the part goes from merely degraded specifications to “failure?” Or do you take a hard line and say that if the selected part may exceed its AMR even for a moment when in nominal or anticipated operating modes, it will be necessary to take steps including adding additional cooling, imposing voltage/current limiting, or using supplemental transient protection?
Although designers of products for what seem to be routine applications may give only a cursory glance to the AMR section — especially if the product is being used in a benign setting — the AMR numbers and SOA are critical for power devices. There’s still an issue here: how seriously should designers take these AMR numbers? Some are cautious and regard them as absolutes as their name indicates, while others regard them as analogous to guidelines such as highway speed limits — meaning they can be exceeded, if you are careful and know what you are doing, analogous to using the passing lane on the highway for just a little while. But they are not guidelines, they are firm limits — and exceeding them is risky, foolish, and asking for trouble.
As with many less-glamorous topics, AMR doesn’t get a lot of highlighting in the engineering curriculum or acknowledgement in mid-range design cases; it’s more of an issue for sensitive analog front ends at one end of the power spectrum and power-device users at the other. Still, companies doing specialized or mil/aero work generally have component engineers whose job is to assess the suitability of individual components for the project, qualify vendors and the supply chain, ask the difficult questions about the corner-case conditions the components will endure, and determine operation is within bounds.
Do you study the absolute maximum ratings section of data sheets? When do you do that — early on, or as you approach the preliminary design review? Do you assume that since you are using a component only around its nominal page 1 specifications, you really don’t have to look at them too carefully? Have you ever been “burned” by ignoring them, either in the prototype stages or with products in the field?
In Compliance, “What’s the Trouble with AMRs?”
Stevan Hunter, Ashok Alagappan, DfR Solutions, 2019 Design for Reliability Conference, “Electrical Overstress (EOS): An Introduction to Electrically Induced Physical Damage”
Brett Barr, Texas Instruments, “Understanding MOSFET data sheets, Part 2 – Safe operating area (SOA) graph”