HALT! Watch out for that weakest link. [transcript]

HALT! And I don’t mean “stop”, I mean Highly Accelerated Life Test which is different from ALT, Accelerated Life Testing. In fact, HALT really isn’t a life test at all. (Why the confusing moniker?) I will tell you what it is, what you could use it for and how it could benefit your design. Right after this brief introduction.

Hello and welcome to Quality During Design, the place to use quality thinking to create products, others love from us. My name is Dianna. I’m a senior level quality professional and engineer with over 20 years of experience in manufacturing, design, and quality. Listen in, and then join the conversation at qualityduringdesign.com.

There are many types and styles of reliability testing. Sometimes there are a lot of acronyms and they don’t really line up with each other, either. But these different tests can be pulled together to serve different purposes for product development or for a particular project. They are employed at different times of the product’s development lifecycle. So, the reliability program that you followed or conducted for one project is going to look a little different than what you did for another project. It really depends on where you’re starting out in your project, what information you know going into it, what it is you’re designing, and honestly, what industry you’re working for. I want to talk about one particular type of testing today which is HALT. Highly Accelerated Life Test is what the acronym means. I want to describe what it is and what it isn’t and when you might want to use it.

HALT is all about discovering weaknesses: what will fail? HALT uses the weakest link mentality. If you have a system, you’re going to apply stresses beyond what it would normally see in the environment. And it’s meant to be an iterative test program where you are testing, analyzing the results, determining the root cause, fixing the design, and then testing it, again. If this sounds a little bit familiar, that’s because HALT’s roots are really from step-stress and over-stress testing.

The failure that you’re looking for when you’re performing a HALT is catastrophic failure (it just fails to operate); degrading performance of whatever function your design is trying to perform; Increasing errors in the functionality of the design; or some other soft failure. It doesn’t always have to be just an all-out break. HALT also does not have any acceptance criteria. In HALT, you are pummeling your product with stress is beyond what it would ever see in a normal use environment and the sole purpose is to find the weakest link of the components in your system. Whatever breaks first, fix it, change it, and then rebuild your system with a new component. And test it again (it may be tested harder with higher temperatures, higher humidity). Find out what’s going to fail next and then that next component that fails: fix it, change it, modify it. And, then with your new system, really test it, again. Run it through the wringer again. Kirk and Paschkewitz put it nicely, I think. To quote them, they say, “HALT has no predetermined pass-fail criteria nor [stress] requirement to achieve. It’s an open-ended weakness discovery and strength improvement opportunity process” (134). There is a lot of words in there, and that statement encompasses a lot of ideas, but again:

  • there’s no acceptance criteria,
  • there’s no requirements that need to be achieved,
  • it’s open-ended until you decide it’s done,
  • you’re discovering weaknesses, you’re finding out where you can improve your design to make it stronger,
  • and, you’re looking for opportunities to do that through a iterative process.

That’s what HALT is.

When do you want to perform a HALT in your design development program? It should really be performed at the prototype stage on a system, not individual components. Start at the lowest functional assembly that you have. Test it, beat it up, find its weakness, make it stronger, and then add it to your larger system. And then, test the larger system together, and beat it up and see what’s the weakest and fix that. Doing HALT early with prototypes allows you to remove the weakest link of your design before you even get to any production phases. It is much easier to choose another component or pick a different layout of a circuit board early in the prototype phase rather than when you’ve got a lot of things set into the production phase and in your design specifications. ASQ is quoted as saying, “Highly Accelerated Life Testing should be a standard component of the design and development phase if there is an expectation that the delivered product will perform at zero or very few failures” (Benbow 206). And this makes sense. If you have something that is a one-use, disposable product, you don’t need to do HALT testing. If you have something critical (say, it’s part of an airplane) that’s expected to never fail, then HALT testing is a great idea for that kind of design.

Our friends, Kirk and Paschkewitz, also have some other things to say about HALT: “HALT is used to find limits and to determine if a product is robust and the design is at or near the fundamental limit of technology. [What it’s not is] it’s not simulation of worst-case stress conditions, or what the average life cycle environmental program may be in the future” (125). It’s not really a life test at all. From the results of a HALT test, we cannot calculate reliability performance of a system, there is no maximum or minimum reliability calculations at certain performance limits, there is no mean time to failure or mean time between failures that we can calculate, there’s no annual failure rate, no life duration HALT doesn’t provide any of that because it’s not ALT: accelerated life test.

ALT is testing a product at an accelerated rate but within the environmental conditions expected. ALT is what speeds up test time by altering the conditions of test. Have you ever seen the commercials where a chair manufacturer has a mechanical system that’s just pounding on it? Maybe dropping 100-pound bags on it and then punching the back, dropping the 100-pound bag onto the seat again…that’s an example of an Accelerated Life Test.

So, back to HALT. HALT is a test goal shift from reliability modeling calculation to eliminate the weakest link, especially for electromechanical systems. A reliability model for electromechanical systems is sometimes too difficult to create. There is no standard fatigue model that can be applied because failure is dependent upon so many variables of a design. That’s why you use HALT in those systems to really beat up your product and remove the weakest link.

To perform HALT, you have to approach it like a program with management support and involving several engineering functions. And the engineering functions are determining stresses, making and designing special tools and fixtures or modifying test equipment, evaluating the root causes when things do fail, and reworking the design. And maybe even procurement and logistics departments are involved. To get started with the whole program, you need somebody that specialized in Reliability Engineering. Contact your local Reliability Engineer. To get a start on your own or to learn more about it, I’d recommend looking up any of the texts that I’m going to cite in the podcast blog.

One last note about HALT testing is that it can be associated with Reliability Growth Testing. Depending on what industry you’re working in (and sometimes if it’s a regulated industry or if you’re a contractor for the military), you may need to produce some plots of Reliability Growth Testing. It really depends if your internal or your customer metrics require it. Reliability Growth Testing answers: is the product’s reliability getting better? Reliability Growth Rate is a measure of your design efforts. “A higher growth rate shows at the effort to eliminate design weaknesses has been given top priority. A lower growth rate indicates that reliability improvement actions are taken to eliminate only the most obvious design flaws” (Benbow 208).

Just to conclude, HALT is not a life test, but it can be an early phase of a reliability test program to weed out the weakest links of your designs, especially if your design is not expected to fail in the field. It takes care to implement by a cross functional team, but it is believed to guide designers to successfully improve reliability of their designs, and that it has direct impact on the long-term reliability of designs in the field. Add it as a potential tool for you as a designer to use to make your best, most functional designs.

This has been a production of Deeney Enterprises. Thanks for listening!

Cited directly in this podcast:

Gray, Kirk A. and John J. Paschkewitz. Next Generation HALT and HASS: Robust Design of Electronics and Systems. John Wiley & Sons, Ltd, 2016. doi: 10.1002/9781118700228.ch5. 

Benbow, Donald W. and Hugh W. Broome. The Certified Reliability Engineer Handbook, Second Edition. ASQ Quality Press, 2013.