Services : Engineering Consultation

Long Life Cycle Design Requirements

Penalties of Failures

Forces of Inner Qualities

Effects of Common Design Practices

Designing for Reliabilities

Welcome to our virtual office of Engineering Research in San Jose, California, USA.

 

In light of devastating recent recall events for 8.5 million Toyota cars with unintentional acceleration, the International Symposium on Quality Electronic Design (ISQED 2010) raised an interesting wakeup call for reevaluating our engineering process for product reliability.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

During one of the ISQED panel discussion, the moderator, Tets Maniwa, opened the panel discussion with a simple question, what makes a single-use throw-away devices like the Mars Rover, lasts over 5 years when it was only designed for 90 days of use, whereas many CE consumer electronic devices designed for 5 years fail within 90 days of use?

 

My quick answer was - because that Mars Rover had to be designed and verified for 100% operation with zero tolerance for failure within 90 days of use.  Whereas, today’s CE products are designed to accept 5% failures within 30 days of purchase and 50% obsolescence within 5 years of product life cycle.

 

This small difference in the product requirement makes big difference in the final quality of a product.  These small philosophical divergence was made apparent through recent automotive industry’s colossal recalls where 8.5 million Toyotas, 950 thousand Hondas, 540 thousand Nissans, 1.3 million GMs, and etc. had serious safety concerns over resultant reliability issues.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Needless to say that there are incredible cost associated with achieving zero tolerance for failure.  There are also too many reasons why the CE products can not afford the qualities of space and industrial products.

 

Understandably, CE product engineering has accepted the challenges of developing process for balancing the market demands and product reliabilities with field upgradeable technologies.  Many of the products, such as Google remain in indefinite “Beta testing” while others like the Microsoft and Apple have automatic updates.   Even many hardware products have field upgradeable firmware that reconfigures the hardware through internet downloads.  

 

Thus the consumer electronics industry has adapted the process of “adaptive-life-cycle” products.  The fiercely competitive CE market environments had obsolete the old practices of keeping engineering margin of error.  The blindly fast cycles of consumer fads has obsolete the old practices of allowing products to age before selling.

 

So, could the traditional long life cycle products, such as automotives and home appliances be designed with current CE engineering practices?

 

The costly failures evidenced by recent recalls of automobiles would answer it—definitely NO. 

 

People will rightfully argue that these products, such as live performance gears, transportation equipments, and household white goods can not afford to fail, because people are depending on these products more these days.

 

Regretfully, more and more products are now implementing these same consumer electronic technologies where many electronics are being designed with software controlled processors.  More and more household appliances—such as ovens, ranges, cook tops, refrigerators, water heaters, and air conditionals, are being designed with “smart” features.  

 

Today’s automotives, for example, are being designed with more than 100 million lines of codes for implementing computer controlled engine managements, anti-locking break systems, diversity wheel stealing, active suspensions, plethora of heating, cooling, visual and aural compensation, and topped off with drive-by-wire user controls. 

 

Any one unexpected failures from these electronics could mean the pain and suffering, even a possible life and death situation often characterized in science fiction horror movies.

 

On the other hand, when I was working for TRW few decades ago, we designed radiation hardened electronics for military satellites.  We jokingly claimed that TRW is a software company, because we developed more lines of code than the whole codes that IBM developed for their mainframe business systems.

 

The difference, however, could be that TRW designed-in lots of fault tolerant controls to build up reliabilities, such as triple redundancy voting circuitries and protective guard bands for minimizing soft-errors due to interferences and peripheral failures.

 

In another word, we bought reliabilities through careful engineering and component selections. 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

In light of Toyota’s unintentional acceleration, I noticed that many people were voicing out the stupidity of Toyota engineering for not implementing a simple break pedal override.

 

At a first glance, the break override seemed to make good sense, but I began to wonder if that might create much more side effects.  

 

For example, when could there be a time where a simple break override will not work?  One case I thought about was during stop-and-go driving conditions of San Francisco.  These cars typically have under powered engines not sufficient for the steep hill side traffics and thus needs to be raved up to build enough torque to prevent them sliding backward.

 

Experienced drivers might have used hand breaks and accelerators to accomplish these tasks.  But, many of these cars don’t have hand breaks but a ratcheted parking breaks and computer controlled engine managements.  

 

Consequently, the central engine management controller may now have to override the break pedal inputs to rave up the engines, to avoid having the cars slide backward and stalling the engines powerless.

 

This brings to a secondary set of questions on exceptions that require situation awareness sensors.  Is the car in stop, is the car on hill side, is the car sliding backward, does the driver want the car to not slide back, etc. etc.

 

Unfortunately, all these sensors and safety override mechanisms will add more points of possible failures.  Bad electrolytic capacitors are notorious for commonly causing electronics mal functions and mechanical sensors are common causes of many software control anomalies.

 

Other environmental factors, such as salt and moisture will shortens mechanical aging and reduce operational reliabilities.  Electric, magnetic, radiation, and yes the mechanical interferences will increase intermittent soft-errors that would be difficult to replicate.

 

So, can we still use same CE technologies and engineering practices to develop long term quality products—such as medical devices?

 

The answer can not be true, but the common technologies and abundant infrastructures could be leveraged with appropriate design practices and associated architecture solutions that utilize fault tolerant redundancies and adaptive systems resiliencies.

 

The need for cost reduction and time to market competition will undoubtedly demand for cutting every possible corners, but the true artist of engineering will learn to balance these diametrically opposing demands of market and qualities.

 

Jason Kim
4/7/2010

Consulting Services

Create Silicon Complements

Services : Systems Solution

Xilicom Research Inc. , San Jose, CA 95120  [email: webmaster at xilicom.com]

Copyright, 2010. All rights reserved.
Updated:  4/7/2010

Safe Harbor:  Many of the information provide in this website are forward looking statements.  Though we do our best to keep them accurate, Murphy’s law dictates that you should double check our analysis to your satisfaction.