A good HASS or a Bad HASS?

Picture of Adam Bahret
Adam Bahret

HASS doesn’t break stuff like HALT. HASS is like a good physical examination from a doctor: It gives you the thumbs up or thumbs down on health based on a quick and thorough examination. But a good thorough examination is possible because of research, education, and practice.

HASS is HALT’s little cousin. With it, you can take what was learned in HALT and very quickly derive a set of stress tests that can enable you to determine whether a product has a quality defect with great accuracy. Alternatives to HASS are “burn in” and environmental stress screening (ESS). Let’s compare these three approaches:

  • Burn in involves simply turning on the product. If it runs at nominal conditions, you are good to go!
  • ESS takes known operational stresses and runs the product through a protocol of higher and lower stresses. None of the stresses go beyond the highest standard use case profile.
  • HASS is a test protocol that uses stresses that go beyond the operational limits—but less than destruct limits—for a set period of time, thereby exposing defects that would cause issue during full product life.

With HASS, getting such high performance also means you have the power to make a mess just as quickly. You are in the danger zone, so surgical precision is required. A little too much stress or a little too long a test time and you are sending damaged product out the door. Not enough stress and you are sending more defective product out the door than you would with a simple ESS. The difference between a good and bad screen can be very small increments of applied stress, but if you do it right you can keep almost all bad product in-house and speed up production with a very fast quality test.

So let’s say you are new to HASS. What should the results look like? Would you see a bunch of failures from the test? Should everything pass? How do you know whether you have a good HASS profile?

Like Goldilocks sampling the bears’ porridge, you have to find a protocol that is “just right,” one that is not too strong but also not too weak.

Not too strong?: To determine whether your protocol is too strong, you need to execute a “proof of screen.” This is a set number of cycles on which you run the HASS protocol for a test population. It takes far longer than the normal HASS screening. The proof is that, if there are no failures, the screen does not take too much life from the product before it is sold. This is important because all the products that have been screened with HASS in production end up in the customer’s hands.

Not too weak?: Determining whether the protocol is too weak is a little more difficult, and, because of that, it is a frequently skipped step. To confirm that the HASS screen is strong enough to weed out the weak products requires constructing some known “semi-duds.” These can’t be flat-out “duds.” They need to be products that appear to be ok but are actually too weak to complete their full life. So, it’s a tight tolerance on “defectiveness.” No outsider can tell you how to make your “semi-duds.” You have to understand your technology, the common quality defects, the root cause factors, and stresses to which they are exposed. Then your engineering team needs to create a product with exactly this level of defect.  It takes some investment.

Here is an analogy of what you are trying to do.  I forewarn you that I went off the rails a bit with this analogy, but it worked… so here it is.

Let’s suppose you are at a track & field race. You are going to watch a full 50-lap race (13 miles) of long-distance runners. All the fans in the stands are picking their favorite runner and betting some big cash. These are your customers. As the organizer, you guarantee that every runner will finish the race. No one will end up betting on a quitter; i.e., this is your warranty.

Here are your options for how you let the customers place their bets:

  1. You have all bets placed before the runners go out on the field. This is the equivalent of manufacturing a product and just sending it to the customer.
  2. You can have the runners go out on the field, warm up, and then let everyone place their bets. At least everyone can see that the runner they selected showed up and knows how to run. This is “burn-in” screening.
  3. You can have the runners go out, warm up, and then do four casual laps, and then let the bets be placed. This is ESS.
  4. You can have them go out on the field, have them warm up, then do a timed two-lap “all-out” top-speed sprint and set a cutoff time for who is allowed to compete. This is HASS.

But now you have to answer the Goldilocks questions:

  • Is the two-lap “all-out” sprint too much strain before a long-distance run? Are you setting up otherwise healthy runners to sustain an injury during the race?
  • Or is this sprint too little? Are runners who are unable to finish the real race able pass the sprint test?

This is how you can answer those two questions.

Before the next race, do the following:

  • Get some other racers, by borrowing some college track team or something.
  • Take five of the racers and put them on the track. You then get five hyenas and put them on the track. You let the hyenas chase the runners so that they are going top speed around the track until the runners drop of exhaustion. Measure how far they went before dropping: This is HALT.  Tip!: You can’t use just one hyena because when the first runner drops the hyena stops…he’s happy. You need all five runners to drop from exhaustion.
  • Do some calculations to derive a HASS sprint distance from the HALT results. You then take another group of 10 runners and have them run that HASS distance (protocol) several  times in a row to prove that it has enough margin to not be deadly to regular runners. This is a “proof of screen.”
  • Now comes the tricky part. Is the HASS profile strong enough such that anyone who can do it is surely strong enough to finish a whole race? You don’t want your customers placing bets on quitters. The best way is maybe to take 10 of the runners and create some slight injury (making them “semi-duds”) but you have to make sure you don’t create a disabling condition, in which case they are just out.

This analogy just went full Quentin Tarantino, but stick with me: We’re 90% of the way through! I did warn you at the beginning that this goes off the rails but I guarantee you’ll remember how to validate a new HASS profile if you follow this through to conclusion. 

  • You need to create a few different types of injuries and to different degrees. Have each of the “semi-duds” do some equivalent of the full long-distance run. Look for the one who still can start strong but doesn’t make it through the full normal race distance. Whatever you did to him/her is your “semi-dud” recipe.
  • Your two-lap sprint screen is strong enough if your homemade “semi-duds” drop from the screen and everyone else who finishes the two-lap screen can complete the long-distance run.

So there it is! You have created an impressively fast and precise manufacturing defect screening profile for runners—and perhaps a solid case for a few lawsuits and jail time. But, at least, the hyenas are happy!

-Adam

Share this post