DESIGN FOR X (DFX): KEY TO

COMPETITIVE, PROFITABLE PRODUCTS

_ _ _ _ _. - - David A. Gatenby and George Foo

George Foo is manu-

facturing and engineer-
Akeyto competitive, profitable product realization is
ing director of the AT&Tdesign for i.e., design for manufacturability, test-
X (DFX);
Little Rock Works, Lit- ability, installability, compliance, reliability, and other
tle Rock, Arkansas.
David A. Gatenby is
downstream considerations. is a philosophy and
DFX
supervisor of the practice thatensures quality products and services,
Design for X (DFX)
Technology and Imple-
reduces the time to market for a product, and minimizes
mentation group at life-cycle costs. Therefore, it is crucial to achieving custo-
AT&T Bell Laboratories mersatisfaction. This paper introduces as a strategic
DFX
in Princeton, New Jer-
sey. Mr. Foo is respon-
concept for product realization, provides a framework
sible for engineering, and examples oftechniques for supporting and
DFX,
new-product introduc- gives an overview ofthe otherpapers in thisissue ofthe
tion, and provisioning
of system solutions AT&T Technical Journal.
to the customer. He
2 joined the company in The DFX Concept
1977. Mr. Foo earned Overthe past several years, AT&T and many other companies
a B.S. in metallurgical havemadesignificant stridesin reducing both costs and productreali-
engineering from Poly- zation intervals and improving quality by reevaluating and streamlining
technic Institute of individual functions and processes. However, we are reaching a point
New York, Brooklyn; where additional cost reductions and efficiency improvements will be
an M.B.A. in finance increasingly difficult to achieve by concentrating solely on specific
from New York Univer- functions and processes.
sity; and a Ph.D. in Although wecontinue to strive for incremental improvements
materials engineering in unitprocesses, we now need to move to the next logical plateau of
from the Massachu- competitive advantage: integration and optimization ofthe entirepro-
setts Institute of Tech- duct realization process, or PRP. (See Panel 1for a listofacronyms.)
nology, Cambridge. Integrating and optimizing the PRP is especially critical for
Mr. Gatenby is responsi- largecompanies whosecompetitive advantage reliesonvertical inte-
ble for DFX technology gration. Further, the successofthese companies depends on howwell
and implementation in theyintegrate the various functional areas along the PRP andbusiness-
AT&T. He joined the operation chain. In the future, only those companies that recognize the
company in 1981. implications ofend-to-end integration-and are ableto capitalize on
Mr. Gatenby earned a it-can be competitive leaders. For example, the integration of
B.S. in computer and research and development (R&D) with production is an essential step
electrical engineering for U.S.-based companies to retainor regain technological preem-
(continued on page 13) inencein global markets.'

AT&T TECHNICAL JOURNAL. MAY/JUNE 1990

 In the sectionsthat follow, we elaborate on the Panel 1. Acronyms in This Paper
DFX environment and describe howDFX ensures quality
productsand services, reduces time to market, and ASIC application-specific integrated circuit
minimizes life-cycle costs.Then, we discussDFX trade- BIST built-in self-test
offs and the electronic packaging hierarchy. CAD computer-aided design
The DFX Environment. In a noncompetitive environ- CAE computer-aided engineering
ment, designersofelectronic systemshavefocused pri- CMOS complimentary metal-oxide semiconductor
marily on productperformance and features without EMC electromagnetic compatibility
always considering downstream needs, i.e., the needs of ESD electrostatic discharge
the processesand operations that follow design. DFA design for assembly
True, a product'sprimary operating character- DFM design for manufacturability
isticsand functionality are essential. Butthe lackofinte- DFS designfor simplicity
gration between product design and other functions DFf designfor testability
resultedin productsthat were difficult to manufacture, DFX design for X, where X standsfor manufac-
install, service, and maintain. In addition, environmental, turability, installability, reliability, safety,
safety, and reliability considerations sometimes were not serviceability, and other downstream con-
considered in productand processdesigns. The priceof siderations beyond performance and
this lack offorethought was oftenredesign, production functionality
and deployment expense, and market-entry delays. Fig- GaAs gallium arsenide 3
ure la depicts this nonintegrated, pre-DFX environment. IC integrated circuit
DFX is an integrated approach to designing lEAP Inter-Entity Assignment Program
productsand processesfor cost-effective, high-quality IM&M information movement and management
downstream operations from manufacture (including PRP productrealization process
fabrication, assembly, and test) through service and PWB printed-wiring board
maintenance. [The X in DFX standsfor manufacturability, R&D research and development
installability, reliability, safety, serviceability, and other ROM read-only memory
downstream considerations beyond performance and SRAM staticrandom-access memory
functionality. In AT&T and elsewhere.l DFX has also VLSI verylarge-scale integration
been called designing for the "ilities." The term DFX is
an extension ofthe well-known terms DFA (design for service, and maintenance-while considering reliability,
assembly), DFM (design for manufacturability), and DFf environment, and safety. The integrated DFX environ-
(design for testability) and preservesthe design for con- mentleads to higher quality, reducedtimeto market,
cept. It is flexible enough to accommodate alldown- and lowercost.
stream considerations-some ofwhich do not havean Quality. Aquality designis one that meets or
ility suffix, e.g., safety and compliance.] exceedsthe full range ofdesignobjectives requiredto
As Figure lb shows, productand process design satisfy a company's customers-i.e., performance and
are integrated from end to end in the DFX environment. features, as well as DFX considerations. Akey concept in
This integration across PRP and business-operation func- AT&T for process-quality management and improvement
tions permitsthe designofproductsand processesfor is the customer-supplier model (Figure 2).3
easier manufacture, distribution, installation, operation, In the DFX environment we showin Figure lb, the

AT&TTECHNICALJOURNAL. MAY/JUNE 1990

 ---" "-"" -----------

Process Product
design design
Process
Product design
design

------- " Process

I /
'\ design
I Environmental, safety, \
\ and ~elia~i1ity J
\ engineenng /
" -.._-----" /

* Customers are responsible

for these activities for some
products.

o Environmental, safety,

________________J
and reliability concerns I

(a) (b)

Figure 1. The design environment. (a) Before DFX principles
were used, product designers focused on product features,
functions, and performance (operational characteristics)
and were decoupled from manufacturing, installation, and
service engineers. Environmental and safety specialists and
reliability engineers were often decoupled from designers
and other engineers, as well. (b) In the DFX environment,
product and process design are integrated end to end. Pro-
ducts and processes are designed for downstream X con-
siderations (ease of manufacture, distribution, installation,
operation, service, and maintenance, each with associated
reliability, environmental, and safety concerns), as well as
for operating features and performance.

AT&TTECHNICALJOURNAL. MAY/JUNE \990

functions are linked through customer-supplier relation-
shipsand are driven by the customer's quality require-
ments, as well as by internal businessobjectives. DFX
requirements andfeedback are key elements in each of
the customer-supplier relationships in the PRP.
The fourth phase ofthe quality-systems
evolution-that is, quality by design-goes beyond the
first three phases (inspection, process control, and quality
improvement).4 In this phase,quality is designed intothe
product and intothe manufacturing process, starting
with the initial concept ofthe product. Designed-in qual-
ityis a primary characteristic ofthe DFX approach. In
terms ofthe cost ofquality (i.e., the coststo prevent,
appraise, and correct defects), the emphasis ofDFX is on
preventing defects early in the PRP rather than finding
them downstream. DFX helps to achieve manyofthe
attributes ofthe broad definitions ofquality-e.g., fitness
for use, which includes reliability, and ease ofinstallation,
service, and maintenance."
Time to Market. Through DFX, wecan reducethe
PRP time-to-market interval by minimizing the numberof
designiterations and changes, instituting parallel proc-
esses, and eliminating activities that do not addvalue.
In previous years (thepre-DFX days), there was
little dialog between design and manufacturing during
product development. A product's designers rarelycon-
sideredmanufacturing's needs in their designs. Process
designers wereforced to tailor their processesaround a
productdesign that often wasnot compatible with exist-
ing processes. In the DFX environment, that is no longer
true. For existing downstream processes, DFX is some-
timesviewed as designing intofixed processwindows.
Whenneither product nor processdesigns are fixed, or
whennewproducts are designed intonewprocesstech-
nologies, major opportunities existfor concurrent pro-
duct and processdesign. DFX techniques are key ena-
blersfor this concurrency.
Although simultaneous product and process
design (sometimes referredto as concurrent engi-
II
l._~__
Requirements Requirements
and feedback and feedback
---l
Figure 2. In thi~ customer-supplier model, customer require-
ments and feedback-as well as internal business
obJectives-drive the process as it converts supplier inputs
into outputs delivered to customers. These customer-
supplier relationships link all functions in the DFX environ- 5
ment, Figure 1b. (Reproduced from Process Quality Manage-
ment and Improvement Guidelines. 3 )
neerint) embodies inherentrisks,the benefits ofparal-
lelism can be substantial when this approach translates
intogettingproducts to marketfaster.
Cost. From80to 90percentofthe total life-cycle
cost ofa product (e.g., manufacturing, distribution,
installation, and servicing) is determined during the
designphase. But as Figure3 shows, the design-phase
efforts representonly about5 percentofthe life-cycle
costs. The downstream functions contribute a sizable
percentage ofthe totalbusiness-cost and expense base.
Thus, products that have notbeen designed for down-
streamneeds have notbeen designed to optimize our
business objectives.
Tradeoffs. For a DFX methodology to be success-
ful, the development teamthat is formed mustconsistof
designers andindividuals from the otherfunctional areas
that are part ofthe PRP.
AT&TTECHNICALJOURNAL. MAY/JUNE 1990

High
t
~
o does not affect the assembly's function. Alternatively, the
Low 1-0::::::::::..-
Time-
Conceptual Production Operation
design and
evaluation
Full-scale
development
Distribution and,
installation
Service and
maintenance
6 circuit wasachieved. The gate countwashalved, com-
Figure 3. Design and Iife-cycle costs. The earliest stages of
product realization have the greatest effect on Iif~cycle
costs, yet they represent the smallest proportion of life-
cycle expenditures.
The team needs to understand howa design's
attributes affect allthe Xs, so that one DFX concern is not
inadvertently optimized at the expense ofanother. For
example, suppose a development teamwants to combine
or integrate parts to improve assembly operations and
reliability. The teamalsoneeds to consider:
- Manufacturability from a parts-fabrication viewpoint.
- Repairability and serviceability.
- Material-logistics overhead for introducing additional,
unique parts. (Material logistics is the planning, man-
aging, and movement ofmaterials neededto manu-
facture a product and deliver it to customers.)
- The effects on purchasing volumes ofexisting parts.
These parts-reduction DFX considerations are
AT&TTECHNICAL JOURNAL • MAY/JUNE 1990
important for mechanical piece-parts andhousings," as
well as for silicon integrated circuits (ICs) and printed-
wiring boards (PWBs).
Similarly, the teammustconsciously evaluate
tradeoffs between DFX and performance or functionality
considerations. For example, an assembly that hasfewer
parts is usually mucheasierto manufacture. However,
one criterion for eliminating a partis that its elimination
assembly's function could be reengineered, so thatthe
part could be eliminated or combined with anotherpart.
Similarly, reducing the numberofgates on an IC
_ could leadto a lower cost chip. Butmaintaining the
requiredfunctions, while reducing the gate count, canbe
a challenge. For example, whencircuits that generate
dual-tone, multifrequency signals for dialing werecom-
binedwith those that produce the telephone's ring, a
cost-effective, single-chip implementation ofa telephone
paredto separate dialing and ringing circuits."
These typesofDFX relationships and tradeoffs,
their inherentinterdisciplinary scope, and rapid changes
in many technologies allmakeDFX complex. Product
development teams mustworktogetherto understand
the breadth ofdesignobjectives, including DFX consider-
ations, and makethe appropriate tradeoffs to ensure
profitability as well as customer satisfaction.
Electronic Packaging Hierarchy. DFX considerations
are important for alllevels ofthe electronic packaging
hierarchy. For example, design fortestability (DFf) tech-
niques'' are applied to devices, PWB assemblies (also
known as circuit packs), and unitsand systems.
Designing for electrostatic discharge (ESD) pro-
tection" andfor electromagnetic compatibility (EMC) 11
also requiresDFX for every packaging level. For example,
we must design for ESD protection at the:
- Device level-e.g., apply CMOS (complementary metal-
oxide semiconductor) protection-circuit design rules.
- Board level-e.g., selectdevices with the highest pos-
sible ESD thresholds.
- Unitand system levels-e.g., provide groundingand
circuit-isolation arrangements.
Similarly, designingfor EMC is necessary at the:
- Device level-e.g., minimize IC emissions.
- Boardlevel-e.g., apply filtering and grounding tech-
niques in the layout.
- Unitand system levels-e.g., use shielded enclosures
and cabling.
Evendesign for assembly (DFA), a well-known
DFX approach for mechanical assemblies, is importantfor
device-packaging and circuit-pack design.
DFX must be integralto product realization to
achieve high quality, rapid product introduction, and
profitability. This integration must be driveninto educa-
tionalprograms,product realization processes and infor-
mation systems, and organization and reward systems.
The next sectiondescribes DFX-related tech-
niques in each ofthese areas.
Supporting DFX in Product Realization
DFX considerations should be built into every
design early in the product realization process. In this
section, we describe the mechanisms for supportingDFX
in the PRP:
- The technicalcore: knowledge base, development
processes, and information systems
- Education and training
- Managerial considerations, such as organization and
reward systems.
Communication and teamwork, which underlie allthese
support mechanisms, are also briefly discussed.
Knowledge Base. DFX knowledge is the under-
standingof how a design's attributes affect downstream
operations, and how the characteristicsof a downstream
process constrain a design's attributes. For example, the
characteristicsof the fabrication process for ICs or PWBs
determine the minimum clearancesbetween intercon-
nectionson an IC or PWB. Similarly, the capabilities ofan
injection-molding or sheet-metal-forming process deter-
mine the bend radiifor features on a mechanical part.
One usually acquires DFX knowledge through
experimentswith test vehicles (e.g., IC and PWB inter-
connection features), failure-mode and effects analyses,
or experienceswith designs and downstream operations
(e.g., servicing or maintaining electronic products). The
systematic acquisition of DFX knowledge is a first step
toward implementing DFX in productand process designs.
Once understood, DFX knowledge can be com-
municated in severalforms:
- Verbal-e.g., in dialogs with DFX experts or in multi-
functional development teams
- Written-e.g., in designguidelines, rules, standards,
or practices
- Digitally encoded-e.g., in computerized databasesof
components and design rules.
The mode ofcommunicating DFX knowledge influences
its effectiveness. For example, rule-based toolscan be a
must for some applications. However, DFM experts have
observedthat some product-design techniquesdepend 7
too much on rules and that these rules are no substitute
for experienced people.l/
Throughout the PRP, clear,continual communi-
cationof DFX knowledge is essential.
Development Process. Here, we discuss methods
for applying DFX knowledge in development processes.
A carefully formulated and documented PRP
description or methodology is needed to achieve produc-
tivity and quality improvements, especially for complex
electronic products, such as computersand network-
systems equipment.
Process definitions include DFX-related tasks at
key pointsin the development process.Among these
tasks are:
- Establishing DFX rules, guidelines, checklists, and
targets
- Team design or redesign activities
- Team inspections, evaluations, and reviews ofa design
relative to DFX considerations
- TrackingDFX performance (e.g., the conformance to
rules, and the results in target X processes).
AT&TTECHNICALJOURNAL. MAY/JUNE 1990
 Proceduresfor executingthese tasks are also important.
For example, disciplined processes for design reviews'!
and inspections'! are considered best practices that
reduce design defects.
Teamworkis essential to achieve DFX. Selecting
the right team for each activity is crucialto an effective
process. Integratinginformation systems into the pro-
cess is also important. For example, on-line design sys-
tems at severallocations can be networked, so that
designers and manufacturers at those sites simultane-
ouslyviewdesigns and jointly improve DFX in real time.
Information Systems: CAE/CAD. Here, we discuss
computer-aided engineeringand computer-aided design
(CAE/CAD) support ofDFX.
In a CAE/CAD environment, DFX can be sup-
ported in manyways. The CAE/CAD tools can:
- Manage DFX knowledge on line-e.g., in component
or DFX-rule databases."
8 - Build in DFX considerations automatically-e.g., incor-
poratetestability into circuitsand physical layouts.v'!
- Audit designs against DFX rules-e.g., design-rule
checkingof PWBs and of VLSl (verylarge-scale integra-
tion) chips.I0.15,16
- Model or predict DFX performance or behavior-e.g.,
for manufacturing yields of chips'? and for board- and
system-level reliability.l"
- Enableon-lineDFX reviews-e.g., via networked
CAE/CAD systems used for simultaneously reviewing
and improving DFX of mechanical and PWB designs and
documentation.
CAE/CAD support of DFX is closely coupled with
the development methodologies and PRP definitions dis-
cussed in the previous section. Where DFX criteria are
more subjective or qualitative (e.g., designingfor foresee-
able misuse to enhance product safetyand preventliabil-
ityl9) , development processes and checklistprocedures
are the dominant means of DFX support.
Simulation and modeling techniques (e.g., for
circuitand physical designs) are increasingly important
to ensure proper design margins and robustness of
AT&TTECHNICAL JOURNAL • MAY/JUNE 1990
designs-e.g., when combined withrobust-design
approaches.P'-! Today'semergingdesign-automation
systems are incorporating more DFX considerations into
their on-line knowledge bases and algorithms. Examples
includeexpert systemsfor PWB producibility.F synthesis
toolsthat incorporate built-in self-test (BIST) ,23 and
optimization systems that use robust circuitdesign'" to
improve the yieldsofsemiconductor manufacturing. DFX
support is one advantage of in-house developed CAE/CAD
systems.
Education and Training. Courses and workshops on
the technical aspects of DFX knowledge, processes, and
CAE/CAD systems are importantfor enhancingthe DFX
skillsofthe development team. Day-to-day teamwork
withpersonnelfrom other functions is also an important
part ofthe DFX education and trainingprocess.
Rotations and internships [acrossfunctions in
the DFX environment (Figure 1b)1are excellent ways to
accelerate DFX knowledge acquisition and appreciation.
For example, the Inter-Entity Assignment Program
(lEAP) at AT&T BellLaboratories (a division ofAT&T)
enables BellLaboratories designers to spend 6 to 12
months in manufacturing organizations. Designerswho
have participated in the lEAP havelearnedvaluable DFM
lessons that they have shared withtheir development
teams. lEAP alsogives manufacturing engineers the
opportunity to spend 6 to 12 months in design organiza-
tions at BellLaboratories. Internshipsin other functional
areas, such as marketing, are also offered.
Personnel at alllevels need to understand DFX
concepts and the relationship ofthese conceptsto pro-
ductivity and quality improvement. And, they need to
understand that DFX requires skillsand discipline that
may differ from those embedded in past, traditional cor-
porate cultures.Top-management supportofDFX is essen-
tialbecause DFX activities typically require increased
efforts earlier in the PRP, withpaybacks to be reaped
downstream. Middle managers need to understand the
reasons for changingtheir operations and allocating peo-
plefor DFX-related activities. Personnelat the operational
levels need an appreciation for DFX and its benefits to
help motivate them to implement DFX in their work.
Universities have started to pave the way for DFX
by emphasizing designand the interaction ofmultiple
disciplines in engineering curricula and research. Exam-
plesinclude graduate-level DFM courses'" for mechanical
design, and research in rc designfor enhanced yields
and designcentering. 26•27 Afundamental need at alllev-
els ofour educational systemis to emphasize cooperative
or team learning, notjust individual learning. Team atti-
tudes, methods, and communication skills need to be
nurtured to laya solid foundation for DFX practitioners.
Organization and RewardSystems. Successin high-
technology markets cannotbe achieved through tech-
nical solutions alone. People-the human resources of
a business-have many nontechnical needs and dimen-
sionsthat must be recognized and managed for DFX suc-
cess. Some productivity expertscontendthat the major
problems ofour work are more sociological than techno-
Iogical/" This is especially relevant for the multifunc-
tional teams requiredfor DFX.
Organizational approaches and reward systems
are people-oriented DFX supportmechanisms that have
tremendous leverage. Buthistorically, they have
received short shriftin high-technology companies.
For effective DFX, organizing the multifunctional
teamfor development is critical. If weget the right peo-
plefrom the different PRP functions and "grow" them into
jelled teams,wehelp enablethe collaboration required
for DFX implementation. Rewarding DFX teamwork, to
ensure that the team's desirable behavior continues, can
involve the application ofpsychological principles. For
example, we shoulduse nonfinancial recognition, avoid
interfering withintrinsic motivation, and reward teams
and the individuals in them-but tip the balance toward
the team-"
Many successful companies haveachieved DFX
through functionally integrated, organizational struc-
tures.-" For example, one approach to ensuringcoopera-
tionbetween productdesigners and manufacturing
engineersis to combine them intoa single organization
that reportsto a common manager. Thus, manufactura-
bility becomes a common, high-priority goalthat
designersand manufacturers workonjointly from
design concept through production. The Japanese have
been particularly successful at the teamapproach. Their
manufacturers are involved at productconception, and
their designersare involved through the start ofproduc-
tion. Variations on this approach have proved successful
in many industries.F-" Atthe macro level, AT&T's
recent restructuring intofocused business unitsis help-
ing us to achieve some ofthe end-to-end PRP integration
that improves DFX.
DFX accomplishments need to be singled out and
rewarded. For example, in determining teamand indivi-
dualperformance, DFX should haveequalor higher
weightthan fire fighting (Le., solving an immediate prob-
lem) , thus creatinga balance between DFX and short-
term needs. 9
DFX innovations must be highly valued, just as
functional innovations havebeen in the past. These
values should be reflected in reward systems, because
both DFX and functional innovations translate into
profits-the bottom line.
Communication. Acommon thread throughthe
DFX supportmechanisms is communication. Effective
communication is essential for teamwork, for developing
and usingthe DFX knowledge base, andfor making DFX
processes work. In a DFX environment, a premium is
placed on interpersonal communication skills. Informa-
tionmovement and management (IM&M) technology is
also valuable for enhancing communications on DFX.
AT&T successfully applies its voice, data, and
video IM&M technology to overcome many ofthe prob-
lems ofdoing DFM whenboth design and manufacturing
are geographically separated. For example, liberal use of
IM&M technology helped an AT&T R&D organization in
New Jersey and a manufacturing organization in Colo-
rado to achieve DFM at a distance.'!
DFX technology, trials, tribulations, and triumphs
AT&TTECHNICAL JOURNAL. MAY/JUNE 1990
 Table I. Profile of Downstream Impact for DFX Papers
Downstream activity
DFXpaper Manufacture Distribution
Design for simplicity? ..j ..j
Design for multinational EMCl1 ..j ..j
Computer-aided design for ..j
quality'"
Design for material logisticsv ..j
Design for ESD protection 10 ..j ..j
Automated BIST in ASICs9 ..j
NOTE: Each downstream activity also has associated reliability, environmental, and safety considerations. Also see
Figure lb, the DFX environment.
should be shared throughouta corporation. AT&T, for
example, holdsan annual DFX conference where employ-
10 ees share experiences on implementing DFX.
Industry-wide conferences on DFX (e.g., on DFM)
havealsobeen valuable sources ofinformation on DFX
experiences and techniques. One added benefitofthese
external activities is that they enableus to take the pulse
ofprogress in DFX. Overthe past five years, the scope
and intensity of DFX-focused activities in industries and
universities haveincreased dramatically.
Summary and Future Directions
This paperhas given a broadview ofDFX-
designing for downstream considerations from manufac-
ture through service and maintenance. We havediscus-
sed knowledge bases, development processes, CAE/CAD
information systems, education, and managerial
approaches for supporting DFX in productrealization.
For U.S.-based electronics companies to become
more competitive, their level of DFX commitment in pro-
duct realization and PRP supporttechnologies needs to
increase. DFX specialists (e.g., in designing for EMC, ESD,
safety and liability prevention, reliability, and maintain-
AT&T TECHNICAL JOURNAL • MAY/JUNE 1990
Service and
Installation Operation maintenance
..j ..j ..j
..j ..j ..j
..j ..j
..j ..j ..j
..j ..j
ability) and personnel from all PRPfunctions are becom-
ing better integrated intomainstream PRP activities. They
are alsobecoming moreinfluential in the development of
PRP supporttechnologies likeCAE/CAD.
Although muchprogress has been made, DFX
knowledge bases and educational programs mustbe
extended, DFX toolsand processesmust be improved,
and DFX technologies and experiences must be shared
and leveraged. Those whofocus on prevention in the
PRP, which is synonymous with DFX and quality by
design, will successfully compete-and profit.
Some DFX Experiences at AT&T
This issue's papers provide a sampling ofrecent
and ongoing DFX workin AT&T. Their authorsare a
blend of representatives from productdevelopment
teamswhohaveimplemented DFX, and technologists
whoworkwith the teams to develop and deploy DFX
methods and tools.
The paperscontain a blendofpragmatic,
experience-based information, as well as technical princi-
ples and results. Therefore, the issue should appeal to a
broad audience: productmanagement, marketing,
design, manufacturing, installation, and service people
who are directly involved withbringinga productto the
marketplace; and R&D peoplewhowork on the underly-
ing technologies ofproducts, productivity, and quality.
The papers coverthese DFX disciplines:
- Design forassembly andsimplicity- Watson, Theis, and
Janek describe? a successful DFX method-design for
simplicity (DFS)-that AT&T has applied to mechani-
cal equipment. This equipmentrepresents the
mechanical and electromechanical portions of our
electronic products, such as computersand telecom-
munications networking systems.
- Design for electromagnetic compatibility (EMC)-
Heirman, Morris, and Crosby address" the develop-
ment ofproducts that satisfy regulatory compliance
requirementsin a multinational environment. They
describe a successful design processfor EMC compli-
ance,testing considerations, and quality metricsfor
the EMC-compliance design process.
- Design formanufacture (DFM)-A paper by Liu,
Nazaret, and Beale discusses'" a computer-aided,
robust-design approach that has been successfully
applied to integratedcircuits. They describetheir
sample-approximate-optimize method, givean optimi-
zation example using a gallium-arsenide (GaAs) IC
circuit, and compare their method to other robust-
design approaches (e.g., the Taguchimethod).21,24,32
A paper by Foo et al. hlghlights-' design-
documentation considerations for the planning,
management, and control ofmaterials. Designing for
material logistics is an important dimension ofDFM
because the cost ofmaterials dominates the totalcost
ofmanufacturing a product.
- Design for reliability and maintainability-Welsher
et al. report'? on designing for ESD protection in
telecommunications products. They review ESD
modelsand effects, describefactory and field ESD-
controlprocedures, explain designed-in protection for
alllevels ofthe electronic packaginghierarchy, and
discussindustrystandards and future trends.
- Design for testability (DFT)-Aadsen, Scholz, and
Zorian describe? efforts on BIST for regular structures
[e.g., read-only memory (ROM) and staticrandom-
access memory (SRAM) 1, embeddedin application-
specific IC (ASIC) devices. They provide somehistori-
calbackground on BIST; describefault models, test
algorithms, and accessfor BIST (e.g., interfaces for
boundary scan); present some specific applications;
and share their views on future directions for BIST.
Table I showsa profile ofthe downstream
impact ofthe DFX techniques described in this issue's
papers. Techniquesin allthe papershave a strong
impact on manufacturability. Distribution material logis-
tics are affected by DFS, EMC design, and ESD protection.
DFS,EMC design, and ESD protection also affect installa-
tionproceduressuch as assembly; and BIST can stream-
linetests and diagnostic proceduresduringinstallation,
as well as duringservice and maintenance. Techniques 11
in four ofthe six papers (DFS, EMC design, CAD for qual-
ity, and ESD protection) influence both operational relia-
bility or robustness and the ease and frequency of ser-
viceand maintenance.
Although this issue's papers concentrate on
hardware DFX, most ofthe DFX principles and techniques
they describehave analogies in software. There are
somefundamental differences between hardware and
software (e.g., withrespect to manufacturing variation
and wear-out mechanisms), but both must be designed
forX to realize competitive products.
References
1. R. B.Reich, 'The QuietPathto Technological Preeminence,"
Scientific American, Vol. 261, No.4, October 1989, pp.41-47.
2. B.Huthwaite, "Manufacturing Competitiveness and Quality by
Design," Proceedings ofthe 4th International Conference onProduct
Design forManufacture andAssembly, Newport, Rhode Island, June
5 to 6, 1989, Institute for Competitive Design, Rochester, Min-
nesota, 1989, pp. 17-25.
3. R. Ackerman, R. Coleman, E. Leger, andJ. MacDorman, Process
AT&TTECHNICALJOURNAL. MAY/JUNE 1990
 Quality Management andImprovement Guidelines, Issue 1.1,Select
CodeNo. 500-49, AT&T Customer Information Center, Indianapo-
lis,Indiana, 1988.
4. E. Fuchs, "Quality: TheoryAnd Practice," AT&TTechnical journal,
Vol. 65, No.2, March/April 1986, pp.4-8.
5. C. 1. Pettijohn, "Achieving Quality in the Development Process,"
AT&TTechnical journal, Vol. 65, No.2, March/April 1986,
pp.85-91.
6. R. I. Winner, J. P. Pennell, H. E. Bertrand, and M. M. G.Slusar-
czuk, The Role ofConcurrent Engineering in Weapons System
Acquisition, IDA Report No. R-338, Institute for Defense Analysis,
1801 N. Beauregard Street,Alexandria, Virginia 22311, December
1988.
7. R. G.Watson, E. M.Theis,and R. S.Janek,"Mechanical Equipment
Design for Simplicity," AT&TTechnical journal, Vol. 69, No.3,
May/June 1990, pp.14-27.
8. P. Wallich, "PhoneChip: SharedCircuits Save Space," IEEE
Spectrum-Special Report: On Good Design, Vol. 24,No.5, May
1987, p. 54.
9. D. R. Aadsen, M.N. Scholz, and Y. Zorian, "Automated BISTfor
Regular StructuresEmbedded inASIC Devices," AT&TTechnical
journal, Vol. 69, No.3, May/June1990, pp.97-109.
12 10. T. 1. Welsher, T. J. Blondin, G.T. Dangelmayer, and Y. Smooha,
"Design for Electrostatic-Discharge Protection inTelecommunica-
tionsProducts," AT&TTechnicaljournal, Vol. 69, No.3, May/June
1990, pp.77-96.
11. D. N.Heirman, R. C.Morris, and S. M. Crosby, "Design for Multi-
national EMC Compliance," AT&TTechnicaljournal, Vol. 69,
No.3, May/June 1990, pp.28-45.
12. D. E.Whitney, "Manufacturing by Design," Harvard Business
Review, Vol. 66, No.4,July-August 1988, pp.83-91.
13. AT&T BellLaboratories, Design Reviews Reference Guide, Contract
No. N0060o-88-C~2307, Reliability, Maintainability, and Quality
Assurance (RM&QA), Office ofthe Assistance Secretary ofthe
Navy (Shipbuilding and Logistics), Washington, D.C., to be pub-
lished1990.
14. P. J Fowler, "In-Process Inspections ofWorkproducts at AT&T,"
AT&TTechnicaljournal, Vol. 65,No.2, March/April 1986,
pp.102-112.
15. W. A. Burling, B.J. BartelsLax, 1. A. O'Neill, andT. P. Pennino,
"ProductDesign and Introduction Support Systems," AT&TTech-
nicaljournal, Vol. 66, No.5, September/October 1987, pp.21-38.
16. C. R. Bonapace and c.-Y. Lo, "An O(n log m) Algorithm forVLSI
Design Rule Checking," Proceedings ofthe 26th Design Automation
Conference, LasVegas, Nevada, June 25to 29,1989, Institute of
Electrical and Electronics Engineers, NewYork, 1989, pp.503-507.
17. C. M. Drumand D. 1. Gerlach, "Yield Model with Critical Geome-
try Analysis forYield Projection from Test Siteson a Wafer Basis
AT&TTECHNICALJOURNAL. MAY/JUNE 1990
with Confidence Limits," Proceedings ofthe 1989 IEEE Interna-
tional Workshop onDefect andFault Tolerance in VLSI Systems,
Tampa, Florida, October 23to 24, 1989, Institute ofElectrical and
Electronics Engineers, NewYork, 1989, pp.197-208.
18. B. F. Gundaker, D. E. Martinich, and M.J.Tortorella, "Quality
Technology in ProductRealization Systems," AT&TTechnical jour-
nal, Vol. 66, No.5, September/October 1987, pp.5-20.
19. J F.ThorpeandW.H. Middendorf, What Every Engineer Should
Know About Product Liability, DekkerPublishing, New York, 1979.
20. 1. Liu, W. A. Nazaret, and R. G.Beale, "Computer-Aided Design for
Quality (CADQ)," AT&TTechnical journal, Vol. 69, No.3,
May/June1990, pp.46-60.
21. M.S. Phadke, "Design Optimization CaseStudies," AT&TTech-
nicaljournal, Vol. 65, No.2, March/April 1986, pp.51-84.
22. R. Hawiszczak, "Integrating Producibility Toolsintoa CAE Design
Environment," Proceedings ofthe 3rdInternational Conference on
Product Design forManufacture andAssembly, Newport, Rhode
Island, June 6 to 8, 1988, TroyConferences, Rochester, Minnesota,
1988, pp.79-84.
23. C.E. Stroud, "An Automated BIST Approach for General Sequen-
tialLogic Synthesis," Proceedings ofthe 25th Design Automation
Conference, Anaheim, California, June 12to 151988, Institute of
Electrical and Electronics Engineers, New York, NY, 1988, pp.3-8.
24. C.C.McAndrew, S. R. Nassif, K. Singhal, andV. Visvanathan, 'The
CENTER Design Optimization System," AT&TTechnical journal,
Vol. 68, No.3, May/June1989, pp.77-92.
25. J. Freckleton, "Graduate Level Training in DFM forthe Practicing
Engineer," Proceedings ofthe 2ndInternational Conference onPro-
duct Design forManufacture andAssembly, Newport, Rhode Island,
April 6 to 8, 1987, TroyConferences, Rochester, Minnesota, 1987,
pp.141-145.
26. S.W.Director, "Designing for Enhanced Yields," SRC Workshop on
Design For Manufacturing, SRC Report No. P89002, Semiconductor
Research Corporation, Tampa, Florida, January 9, 1989, paper
II.2.1.
27. A. J Strojwas, "Design For Manufacturability andYield," Proceed-
ings ofthe 26th Design Automation Conference, LasVegas, Nevada,
June 25to 29,1989, Institute ofElectrical and Electronics
Engineers, New York, 1989, pp.454-459.
28. T. DeMarco andT. Lister, Peopleuare: Productive Projects and
Teams, DorsetHousePublishing Co., New York, 1987.
29. J C.Mower and D.Wilemon, "Using Psychology to Reward
Teamwork-Four Important Principles," Proceedings ofthe 1987
IEEE Conference onManagement andTechnology, Atlanta, Georgia,
October27to 30,1987, Institute ofElectrical and Electronics
Engineers, New York, 1987, pp.294-299.
30. J W.Deanand G. I. Susman, "Organizing for Manufacturable
Design," Harvard Business Review, Vol. 67, No.1,January-February
 1989, pp.28-36.
31. J. W.Wesner,"DFM at a Distance," Proceedings ofthe 1990Inter-
national Computers in Engineering Conference, Boston, Massachu-
setts, August5 to 9, 1990. American Society of Mechanical
Engineers, Chicago, Illinois, to be published 1990.
32. R. N. Kackar and A. C.Shoemaker, "Robust Design: A Cost-
Effective Methodfor Improving Manufacturing Processes," AT&T
Technical Journal, Vol. 65, No.2, March/April 1986, pp. 39-50.
33. G.Foo,]. P. Clancy, L. E. Kinney, and C. R. Lindemulder, "Design
for Material Logistics," AT&TTechnical Journal, Vol. 69, No.3,
May/june 1990, pp. 61-76.

Biographies (continued)
from Trinity College in Hartford, Connecticut, and an M.S. in
computer engineering from Carnegie-Mellon University in
Pittsburgh, Pennsylvania.

(Manuscript received December 21, 1989)

13

AT&TlECHNICALJOURNAL. MAY/JUNE 1990