Will your product survive in the field? HALT and HASS testing are not just for aerospace electronics anymore.

This column has previously considered many aspects of test and designing a product for its entire life (September 2005 and November 2005). The last link in this chain of design characteristics is the one most often overlooked; we call it Design for Reliability. DfR goes beyond the ability to produce and ship products with very high yields and robustness. It considers the long-term life of the product and takes robustness to a much higher level.

A manufacturer's worst-case scenario is a product that fails repeatedly within its warranty period. Besides the fact that costs escalate once the product leaves the factory, it is important that the product does not fail for reasons of market acceptability. Therefore some form of accelerated age testing is warranted.

HALT/HASS is gaining in popularity and relevance to manufacturers, although not enough designers and manufacturers may truly understand all the implications.

HALT (highly accelerated life testing) is "a series of controlled tests that stress a product beyond its normal working envelope with the intention of discovering latent defects that may cause issues for it in later life." The latent defect is a potential fault connected with components, materials or the manufacturing process that is not found during conventional test procedures (Table 1). HASS (highly accelerated stress screening) is "a series of controlled tests that stress all products repeatably to precipitate and identify latent defects."

The two methods may sound similar, but HALT is intended to test first-run or preproduction units far beyond their expected levels of stress to identify latent defects before production ramps. By contrast, HASS can be used to stress all products in case a latent defect occurs that did not arise during HALT. Clearly, HALT is the important regime to consider at the design stage and so we will look at it first.

[FIGURE 1 OMITTED]

HALT and HASS employ both thermal and vibration analysis together so the product can be exercised in controlled environments that it would not see regularly in normal life. Neither regime on its own will reveal all the likely latent defects, but together they can expose even a single potential defect that might cause angst for the manufacturer later.

The worst-case failure scenario, of course, is any device that is safety-critical such as an avionics controller or an instrumentation package on an oil rig. Therefore, major users of HALT/HASS include the military, avionics, automotive, medical and safety instrumentation sectors. This should not preclude others from using these techniques. All industry sectors can benefit from the good use of all DfR and DfT (design for test) protocols.

HALT/HASS tests are not a panacea and some acceptance of limitations must be kept in mind. The major issues are:

* Does not replace existing reliability or environmental regimes but does complement them.

* Cannot simulate "worst-case" operating conditions.

* Cannot provide MTTF (mean time to failure) data.

* Does not provide a temporary solution to a design fault.

Many manufacturers ignore HALT/HASS because they believe that the time, effort and costs involved are not justifiable. Other preconceived notions are:

* Prototype designs work well so other tests are not needed. (However, the issues surface later.)

* The diagnostic regimes cover everything and are very efficient. (Yet the issues they raise are solved problems and teach nothing new.)

* The worst-case tolerance groupings do not occur in practice. (If they do, it always seems to happen on a customer's premises.)

* The best design rules, DfM rules and process control regimes are always used. (But if this means doing too much extra work, the rules can be relaxed.)

* The product is evaluated under typical use conditions. (We believe anything more rigorous will introduce spurious errors.)

HALT/HASS offers a new way of considering manufacturing, one that makes added sense in companies that emphasize Lean or Six Sigma manufacturing. The advantages include:

* Apply HALT to understand a design's margins.

* Use HASS to monitor manufacturing process quality.

* Both permit the understanding of failures and root causes.

* Field returns are an opportunity for improvement.

* Both permit the understanding of the true cost of poor quality.

Typically it can cost 20,000 times more to correct something at warranty versus correcting it at design. As such, we need to find the test dimensions that reveal weaknesses or latent defects rather than test to comply with a written specification.

Failures take many forms including fatigue failure, exceeding product specification limits, corrosion, oxidation, mechanical wear and consumption of active materials (e.g., batteries). Let us consider the example of fatigue failure.

Fatigue failures represent the majority of field failures. They can often be exacerbated by transport issues or rough handling, happen incrementally, may occur late in the product's life and are often related to product design. HALT using vibration will promote fatigue failures early enough to worry designers into good practices.

Fatigue failures exhibit the classical failure mode associated with stress concentrations. A stress concentration can occur because of cracks, voids, metallurgical inclusions or design faults such as sharp features or notches.

Stress concentration figures have been created to try to show limits that indicate likely premature failure. Basically, the lower the stress concentration number, the more likely the product will fail early. A solder joint with a low concentration number will fail in perhaps as low as 1/30 of the time that a good joint would need.

Some stresses--many, in fact--are caused by cyclical thermo-mechanical excursions, and HALT testing will reveal the stress concentration factors associated with thermo-mechanical stress. Cyclic mechanical stress is effectively applied to products by thermal cycling. Random mechanical vibration will reveal stress factors caused by the intermetallic layer, and the two stress analyses together will exercise the assembly far beyond its point of normal activity. Good control over vibration characteristics and thermal cycling are essential.

Figure 1 shows that most of a product's production output performs not only to the required specification ("robustness") but that there are also random complete failures. These would not have been revealed by conventional test methods.

HALT testing sets the absolute limits of the product's resistance to stress. Now we need to establish practical stress limits for the product. HALT offers the following results leading to HASS:

* It shows weaknesses in design elements that can be eliminated before production commences.

* A final level of design robustness can be established.

* Field reliability of the product is improved.

* The starting point for HASS testing can be established.

HALT and HASS equipment is basically the same but HASS tests tend to be continuous. It is often better to use separate equipment for the tests (despite the extra cost).

Once lessons from HALT have been converted into a HASS program, the job is not over. Although the HASS program will commence with volume production, it will need to be revalidated if engineering changes are made or if failures increase; in such cases, all involved parties must be alerted. Therefore, if an EMS company is the main focus for manufacturing a particular product, that EMS firm must be included in the HASS loop and may be the "owner" of the appropriate equipment. All this hinges on meticulous recordkeeping by all parties involved and those records must be shared and used properly.

In conclusion, a reliable product is the result of careful design considerations, which should now include HALT/HASS. The more complex the technology involved, the more benefit will be derived from strong HALT/HASS policies.

Au. note: The author wishes to acknowledge the valuable input offered by Bob Page of Reliability Plus and Gregg Hobbs of Hobbs Engineering.

Peter Grundy is director of P G Engineering (Sussex) Ltd. and ITM Consulting (itmconsulting.org); [e-mail]. His column appears semimonthly.

[ILLUSTRATION OMITTED]

Table 1. Typical Issues Found by HALT

Vibration                 Thermal                 Vibration/Thermal

Adjacent PCBs shorting    Component value drift   Component fitted
                                                    incorrectly
Chafed or pinched wire    Chemical contamination  Cracked substrate
Cracked insulator         Components incorrectly  Defective fastener
                            fitted
Defective crystal         Defective harness       Defective PCB etch
                            termination
Incorrect torque          Film resistor           Defective or broken
                            metalisation            component
Insufficient solder       Hermetic seal failure   Intermittents
Loose part                Incorrect crimping      Loose hardware
Mechanical flaw           PCB opens and shorts    Poor solder joints
Particle contamination    Wrong component
Parts rubbing
Poor connection
Poorly bonded component
Poorly secured high mass
  part
Components shorting
Broken leads