Fig. 1: Product Planning Matrix "The House of Quality"
Jerrold Foutz
Rockwell International Corporation
Defense Electronics
Anaheim, California 92803-4192
Abstract - Quality Function Deployment (QFD) is used to drive user needs into a power supply design guide. These needs are met with a hypertext but not met with a printed guide. Features such as launching computer programs and semi-automatic documentation of the design, expand the concept of a design guide.
-----
The companies and products mentioned in this paper are a matter of fact and no endorsement is made or implied.
-----
W. Edwards Deming states "A company must, for its very existence, make use of the store of knowledge that exists within the company, and learn how to make use of help from the outside when it can be effective [1]." Except for knowledge in the minds of employees, most of this store of knowledge is in printed documents including design guides. Yet formal studies [2] and informal surveys indicate that engineers seek this information as a last resort rather than as a normal part of the design process. Even so, finding needed information takes up to 30% of the designer's time [3].
To address these problems, a different approach to writing and making available a design guide is taken. This approach includes the use of hypertext to deliver the guide, the use of transactional analysis discount theory to organize material, the use of Quality Function Deployment (QFD) to define user needs, and the use of information annealing for continuous improvement.
This paper describes the design of a power supply design guide using these approaches and reports on the comparative user acceptance of the resulting hypertext guide and a conventional printed version.
The following sections describe the methods used, the use of QFD to drive the user requirements into the guide, the details of the guide, the use of information annealing for continuous improvement, the use and acceptance of the guide, and conclusions.
In a conventionally written design guide, the material is presented in a sequential or linear manner that is controlled by the author of the material.
In contrast, material in a hypertext is "chunked" into small fragments of linear text containing a single idea or procedure. The location of these fragments of information are called nodes in a hypertext. The hypertext author then ties these nodes together with links that allow jumps between nodes. The presence of a link is indicated by a link marker. Multiple links in a node present the reader with many paths through the material - and reading takes on a non-linear aspect under the control of the reader, not the author.
The hypertext author can build several suggested paths through the material. One path can be a series of links that duplicates the conventional linear path of printed text. But the value of hypertext comes from other forms of linkage.
Linking by association mimics human long-term memory and was the original model used by Vannevar Bush for developing hypertext [4] [5]. This mimicking of the human mind is part of what makes finding information in a good hypertext seem so natural. The other part comes from the skill of the author in organizing and linking the hypertext.
Linking can both mask and reveal information. By creating paths for various readers, the cognitive load on the reader is reduced and their ability to find, absorb, and work with ideas is increased.
Chunking knowledge into single-idea fragments makes it easy to capture information, and linking this information in any manner conceivable to the human mind makes hypertext an ideal way to turn information into knowledge. When combined with networking, hypertext is a powerful enabler of group processes such as concurrent engineering or collecting and distributing corporate lessons learned. The latter is the heart of a design guide.
The strengths of hypertext appeared to be a good match for the requirements of a design guide (this was later verified by the QFD process) and the power supply design guide was written for hypertext -- with a linear path providing a conventional printed guide for the first version.
Having selected the hypertext approach, hypertext software was selected. Software selection criteria included
The software selected, Hyplus (tm) by MaxThink Inc., Berkeley, California, has a free run-time license, uses any word processor or text editor (that can save an ASCII file), runs fast even on the slowest PC independent of the size of the hypertext, and uses visible links (angle brackets) that ease linking and maintaining links. Also available from MaxThink are software tools for creating and maintaining large hypertexts, and a variety of compatible run-time systems. One of these, Hynet (tm), which is licensed for one-server unlimited users, is used for the network version of the guide. The ability of Hyplus and Hynet to launch other computer programs and to capture the results in hypertext are other useful features of the programs. This capability is expected to greatly enhance the utility of a design guide.
Unique to the power supply design guide is a problem-solution approach originated by the author at the Naval Command and Control and Ocean Surveillance Center (then the Naval Ocean Systems Center) in San Diego. This approach is based on transactional analysis discount theory.
Discount theory stems from research into why people don't do things or don't do them effectively [6]. The original research was directed towards treating mental illness, but it has application to normal situations -- in this case, problem solving.
To solve any problem you have to be aware of it, have to believe it is relevant to you, believe that it is solvable, and believe you can solve it - in that order. Then you can get on with the solution. Discounting at any level blocks lower levels in the sequence. Problems get solved only if none of the levels are discounted. For example, a common discount heard is "We can't afford it" which is often a gross distortion of reality and effectively blocks creating excellent and affordable solutions.
Design guides are usually consulted when there is either a problem or a desire to avoid problems -- and discount theory applies. In authoring the guide this awareness, relevance, solvability, and solution approach guided the entire writing of the guide.
The application of discount theory to the guide is most explicit in the Problem/Solution displays, but it is applied throughout the guide. For example, the opening display for each reader-category quickly focuses the reader on what is relevant for them, excluding non-relevant information. Links then develop each topic further using the discount-theory model for problem solving.
Quality Function Deployment is a discipline for product planning and development in which key customer wants and needs are deployed throughout the organization. It was developed by the Japanese ship building industry (Mitsubishi) in 1972 and introduced to the United States in 1983 [7].
QFD consists in developing, and using as a communication tool, sets of linked matrices that flow customer wants down through technical requirements, part characteristics, process characteristics, and process control.
The first or planning matrix is sometimes called the House of Quality due to its shape (Fig. 1). This matrix relates customer requirements to technical requirements and identifies the degree that technical requirements bear on customer requirements. This planning matrix can be expanded to contain other information such as competitive evaluation, target values for technical requirements, and technical evaluation. Its strengths are in evaluating how well competitive technical approaches meet customer needs and in communicating complex interrelationships.
Fig. 1: Product Planning Matrix "The House of Quality"
After some use of the first hypertext version, QFD was used to analyze user needs and drive them into the guide by better use of the inherent features of hypertext. It became clear that user needs are poorly met by printed guides -- a partial explanation of their poor use. The QFD version was released in April 1992. The resulting improvement in the guide was dramatic.
Because the guide is computer based, it can instruct users in the use of a program and then launch the program, collecting the results in a file that can be linked in hypertext. The authoring features of the hypertext run-time system allow the design to be documented in an electronic notebook as the design progresses, including the results of computer programs and expert systems.
From a narrow focus, the intended users of the guide were analog circuit design engineers at the Anaheim location of Rockwell International Corporation who designed or might design power supply and other power electronics circuits. The guide was intended also for similar engineers throughout the corporation. From a broader focus, the intended users were program managers, project engineers, system engineers, printed circuit board designers, test engineers, etc. who would benefit from some knowledge of the problems, solutions, and lessons learned about power supply and power electronics applications.
In writing the original version of the guide, the author used examples of previous guides and his own background of what would be useful, including the discount-theory approach to problem solving. Care was taken to format the information so it could be used for either sequential reading or hypertext.
After being exposed to Quality Function Deployment, the author decided to use the QFD process on the next revision of the hypertext version of the guide. This resulted in a dramatic improvement of the guide, primarily by changes to the access displays.
As the first step in applying the QFD process to the guide, a list of user needs was proposed and then refined by one-on-one interviews with power supply designers. Table I summarizes the results of the QFD analysis of the user needs.
|
TABLE 1 USER NEEDS IN A DESIGN GUIDE |
|---|
| Need | Reason |
|---|---|
| Information must be useful | Or guide will not be used |
| No non-relevant information | User is already swamped with raw information |
| All relevant information | Can't afford to miss important points |
| Information in order of importance | Relevant information may still swamp user |
| Multiple levels of information | |
Awareness |
What are the problems? |
Relevance |
Are problems relevant to my design? |
Scope of solutions |
What are the solution options? |
Details sufficient for design |
How do I solve the problems? |
| Available | Or guide will not be used |
| Easy to use | Or guide will not be used |
| Up to date | To meet current needs |
| Accurate (or calibrated) | Or worse than no guide |
| Complete | To assure continued use of the guide |
Access and ease of use are fundamental to a guide, otherwise it will not be used. Beyond that, the guide is a problem avoidance or problem solving tool and the discount-theory sequence of awareness, relevance, solvability, and solution applies.
Finally, a guide is, or should be, a living document, and continuous improvement through information annealing applies. Table II summarizes the design features of a hypertext design guide in terms of these uses.
|
TABLE 2 DESIGN FEATURES OF HYPERTEXT DESIGN GUIDE |
|---|
| Feature | Uses | Examples* |
|---|---|---|
| Menus | Access Awareness Relevance |
Opening Menu Program Management Issues Design Review Checklists |
| Lists | Access Awareness Relevance |
Design Review Checklist Booklist Problem/Solution list Directories |
| Problems/Solutions | Awareness Relevance Solvability Solution |
Thermal Runaway Electromigration Junction Paths Single Event Burnout |
| Tutorials | Awareness Relevance Solvability Solution |
Power Growth Risk Capacitor ESR in Rad-Hard |
| Algorithm | Solution | Rad Hard Power Supply Design |
| Computer Programs | Solution | Inductor Optimization |
| Expert Systems | Solution | Converter Identification Cost estimating PSE |
| Indices | Awareness Access |
Idea Index Problem Index Tutorial Index Algorithm Index |
| Glossaries | Awareness Access |
Glossary Link Lookup Command |
| Information Annealing | Continuous Improvement | Comment file |
| Network Access | Access Ease of Use |
Novell Network |
| Portability | Access Ease of Use |
Floppy Net |
| * Examples illustrate features and are not paired to uses. | ||
Fig. 1 relates user needs and design features of the hypertext guide in a "House of Quality" planning matrix.
In this application, only part of the planning matrix was used. Several features of the house, including the roof shaped correlation matrix, which compares technical requirements against each other, were excluded due to marginal utility for this application. The primary use of the planning matrix was to see how the technical features of hypertext could best be used to satisfy the needs of guide users. A competitive comparison was also made to a paper design guide. The following is a step-by-step description of applying the planning matrix. The process can be repeated to refine the results.
The customer requirements are first listed in the left column and an estimate made of the degree of importance to the customer in terms of a 2 to 10 score, and a relative weighting in percent. Scoring is: essential (10), critical (8), important (6), desirable (4), and nice (2). The most important customer requirements are found to be availability and usefulness of information (14%) followed by making the customer aware of problems (11%). Containing all relevant information and being accurate and complete follow (8%). Ease of use, being up-to-date, excluding non-relevant information, and providing solvability information are next in importance (5%). While all customer requirements are important, ranking information by priority is the least important on this list (3%). This is surprising since one motivation for starting the project was to reduce cognitive overload by using priorities to screen non-relevant information.
The potential of a hypertext and printed guide in meeting the customer requirements is assessed next. An equal sign is used where they have equal potential, a plus sign used if one has superior potential, and a minus sign used if the requirement is impossible, difficult, or not practical to provide [Post Publication Note: The blanks in the "Paper Potential" column should be a minus sign.] The superiority of a hypertext over a printed guide is made apparent by this comparison.
Next the technical features (requirements) of a hypertext are listed in the top row of the matrix and the central work of the planning matrix is performed. This is determining the relationships between the customer and technical requirements. Strong relationships are indicated by a heavy circle, value of 9, moderate relationships with a circle, value of 3, and weak relationships with a triangle, value of 1. The degree of importance to the customer and the strength of the relationship between technical and customer requirements are multiplied to give the importance weight of technical requirement in satisfying customer needs.
One initially unexpected result was the overwhelming importance of information annealing discussed further in Section V. On second thought, this should be no surprise to anyone trained in the use of feedback to improve system performance. Information annealing is the feedback loop of a guide. It was expected that the problem/solutions displays would rank high and they did, sharing a high score with tutorials. The algorithms were split to give visibility into manual, computer, and expert system algorithms. As a lumped group they score next.
A second surprise was the low scores of networking and portability, the only items that affect availability, an absolutely essential requirement.
Finally, improvements for the next edition of the guide are targeted in the Movement of Target Value row. The symbol x is used to indicate meeting a fixed target with up and down arrows indicating movements. Because of the overwhelming importance of information annealing, improving this aspect of the guide is targeted as high priority. Also targeted are: replacing manual algorithms with computer programs and expert systems, adding problem/solutions displays and tutorials, and improving the glossary -- which by adding hypertext links has some exciting possibilities.
Menus are the primary access method to the contents of the hypertext guide. Their main function is to make the reader aware of relevant information and screen non-relevant information to reduce the cognitive load on the reader.
When the hypertext is opened, Fig. 2 is displayed with the first angle bracket highlighted. The up and down arrows are used to highlight the desired link and the right arrow is used to access the highlighted topic. This display provides an opportunity for the first-time user to learn more about the hypertext through a link to an introduction. It then invites the reader to identify to which reader category they belong. By defining reader categories, the next display can be used to identify relevant information and screen non-relevant information for that reader category.
Design Guide - Power Supplies (Version 3.0F)
INSTRUCTIONS Select Topic..................Up or Down Arrow
Display Topic.................Right Arrow or Enter
Retrace Steps.................Left Arrow or Escape
Help..........................F1 Key
Return to this menu...........Ctrl-Home
Leave Knowledge Base..........Q or X or Alt-X
TOPICS
Introduction.........................................<ADG001>
Topics For
Managers...........................................<ADG014>
System Engineers...................................<ADG015>
Power Supply Designers.............................<ADGM01>
Layout and Mechanical Designers....................<ADGM04>
Test Engineers.....................................(ADGM05)
Manufacturing......................................(ADGM06)
Index................................................<ADGIDX>
Information Annealing............................<ADG000.ZZZ>
Auxiliary Menu.......................................<ADGM07>
|
| Fig. 2. Opening Display |
|---|
This screening process is illustrated by the opening display for program managers, Fig. 3. It makes managers aware that their primary concerns are requirements, schedule, cost, and risk and that NAVMAT P-4855-1A has been adopted as a corporate-wide guide. If you don't know that P-4855-1A is "Navy Power Supply Reliability, Design and Manufacturing Guidelines," a link tells you this and gives you an overview. A menu then links you to the relevant parts of the guide, screening out non-relevant information by not providing direct links. You can always get the non-relevant information by assuming another reader category or accessing information through the index, glossaries, or other means.
Program Management Issues
SCOPE Program and other management needs to be aware of
the impact of the power supply on requirements, cost,
schedule, and risk.
INTRODUCTION A good introduction into these responsibilities is the
program management section of the executive summary of
P-4855-1A (BOOK22), which as been adopted as a design
guide corporate wide. This summary also contains
information on output power density, standardization,
and design reviews of interest to program managers.
P-4855-1A Program Management Section.....................<ADG021>
CONTENTS The following sections of the design guide provide
information of interest to program and other managers.
Requirements....................................<ADG017>
Schedule........................................<ADG002>
Cost............................................<ADG003>
Risk............................................<ADG004>
P-4855-1A Executive Summary-Program Management..<ADG021>
Program Management Checklist....................<DCL014>
|
| Figure 3. Program Management Issues |
|---|
The primary use of lists in the guide are book lists, bibliographies, and design checklists. The design checklist is one of the important organizing methods used in the guide. A menu, Fig. 4, gives access to the design checklists allowing the user to go to the relevant material. A checklist is a list of questions that can be answered yes if the design is satisfactory for that item. If the answer is no or the reader is uncertain of its meaning, links give access to a Problem/Solution display addressing the question.
Power Supply Design Checklist
INTRODUCTION This power supply design checklist augments the
checklist in NAVMAT P-4855-1A <DCL002> which should
be used in addition to this checklist.
CHECKLISTS Program Management....................<DCL014>
System Engineering....................<DCL015>
Spacecraft Power Systems..............<DCL018>
Power Supply Design
Use of Design Guides................<DCL002>
General Military Specifications.....<DCL003>
Documentation.......................<DCL004>
Part Selection and Use..............<DCL005>
Magnetics...........................<DCL006>
Circuit Protection..................<DCL007>
Circuit Stability...................<DCL008>
Reliability.........................<DCL017>
Thermal.............................<DCL009>
Electromagnetic Compatibility.......<DCL010>
Radiation...........................<DCL011>
Packaging...........................<DCL012>
Figures-of-Merit....................<DCL013>
|
| Figure 4. Power Supply Design Checklists |
|---|
The heart of the guide is its problem-awareness orientation, best exemplified by the problem-solution displays. Fig. 5 is an example of the display linked to the design question "Is the heatsinking of power MOSFETs sufficient to prevent thermal run away conditions at high power levels?"
Power MOSFET Thermal Runaway
Problem The on-resistance of a power MOSFET increases with
temperature. Along with the resistance increase, the
power dissipated for a constant on-current increases,
the junction temperature increases, and the resistance
further increases until the device is in thermal
equilibrium with the heat removal system. If the heat
removal system is inadequate, thermal run away occurs.
Relevance The existence of the problem can be found by
Test graphically or numerically solving the mutually
dependent equations for power dissipation and junction
temperature. If no solution exists, thermal runaway
exists.
Solvability The normal solution is to improve the heat removal
system by lowering the thermal resistance from junction
to ambient.
Solution Siliconix Application Note AN83-10 <AB1S -SEVE84A>
Details
|
| Figure 5. Problem/Solution Display - Thermal Runaway |
|---|
First, a short description makes the reader aware of the problem. Second, a test is proposed that lets the user determine if the problem is relevant to his design. Third, the general solvability of the problem is discussed. Fourth, details are provided that allow the reader to solve the problem. Until more problems and solutions are entered into the guide, the solution details are a pointer to a reference that contains the solution approach. Later, the solution can be incorporated into the guide directly, including expert systems or computer programs that will expedite the solution and its documentation.
Tutorials: Tutorials differ from Problem/Solution displays primarily in length. The same awareness, relevance, solvability, and solution outline is followed.
Algorithms: An algorithm is a cook-book approach to solving a problem. When practical, written algorithms are replaced by computer programs or expert systems activated by hypertext links in the guide.
Indices: Ideas, not words are indexed. Links from the index can take the reader directly to the relevant material. The indices are poorly developed at this time.
Glossaries: At present the glossary mirrors a conventional glossary except that it can be activated by highlighting the word of interest in the text or by typing in the word at the command line. Hypertext glossaries have much more potential as described in [9].
Information Annealing: Information annealing is described in Section V.
Portability: The last version of the guide could be compressed to a single high density disk and expanded on any compatible computer. This was the main method of distribution. The drawback was the space taken when installed on a hard drive.
Network Access: Recently the guide has been installed on a network and expanded beyond what can be compressed on a single high density disk.
The guide continues to evolve and improve as users interact with it through a process called information annealing [8].
Information annealing consists of creating a feedback loop between the hypertext users and authors and using the feedback process to continually improve the hypertext. In practice, whenever a user sees an error, or a way to improve the hypertext, the reader leaves a comment in some media for the author of the file. The author then incorporates the information in the hypertext or leaves a record of why-not along with the original comment. In this way group knowledge is used to continuously improve the hypertext and controversial viewpoints are not lost.
The feedback loop can be implemented several ways, such as personal contact or telephone contact -- direct or by a telephone answering machine. On a local area network (LAN), electronic mail or a bulletin-board comment file can be used.
The first information annealing of the guide took place when the paper version was circulated to approximately 40 designers whose primary or secondary job responsibility consisted of power electronics design. Several reviewers provided extensive comments, with minor comments from others.
Hypertext information annealing began when people started calling about something they did not understand in the hypertext version. The question was answered and the guide annealed.
However, most information annealing has been the result of people asking questions of the author and his office companion independent of the guide. If the answer is in the guide, they are referred to it (to get them used to using the guide). If not in the guide, a summary of the question and answers is added to an information annealing file in the guide. This provides immediate information in an unorganized form and provides a to-do list for updating the guide.
The guide has recently been installed on a LAN but information annealing procedures have not yet been implemented. They are expected to be implemented through the use of a comment file or by electronic mail. The results of the QFD analysis indicate creating an effective information annealing environment is one of the best ways to spend effort in improving the guide. Good results are expected based on other projects using a comment file on a network. The comment file usually appears empty because comments are incorporated into the original files within hours of receipt.
Neil Larson, president of MaxThink, Inc., is the definer and major advocate of hypertext information annealing and his writings on the subject are recommended reading [8]. The concept goes well beyond what has been discussed here.
The power supply design guide started as part of a general upgrade of design guides in Rockwell International Corporation Autonetics Strategic Systems Division. The power supply design guide was originally released in a printed form in January, 1991. At the start of the project, an informal survey indicated that most engineers made little use of existing guides. There was also an increasing use of personal computers and plans to tie these computers together into a network. In an attempt to increase the use of the guide and to make it compatible with the increasing use of personal computers and networks, the power supply design guide was concurrently released in hypertext form on a diskette for any IBM compatible personal computer.
As expected, little or no use has been made of the printed copy of the design guide. An informal survey by the author failed to find anyone who had actually used the printed power supply design guide or other specialty design guides written at the same time.
The results have been better for the hypertext, primarily because of its novelty. Since going on the LAN there have been many comments on the hypertext design guides on the network, many from others who want to make similar hypertext guides.
The capability of the guide to launch computer programs and expert systems and collect the results in a design notebook have been implemented and tested but have not been made available to LAN users pending a software upgrade to allow more available memory when shelling from the hypertext.
The real test of the hypertext guide will come shortly when all designers get access to the network containing the design guides. This presents both an opportunity and a danger. The opportunity is to increase the use of design guides by a paradigm shift, reaping the benefits mentioned by Deming in the quote at the beginning of the paper. The danger lies in the incompleteness of the guide. If the users do not find what they are looking for the status quo of seldom used guides will continue. The best prescription to prevent this appears to be exploiting information annealing to gain as widespread an ownership of the guide as possible. Another prescription to increase the use of the guide is to increase the number of useful computer design algorithms launched directly from the guide.
Quality Function Deployment analysis shows a hypertext is clearly superior to a conventionally printed design guide in meeting user needs. The hypertext provides tailored reading paths for various reader-categories that focus on relevant information and reduce cognitive overload by screening non-relevant information. Access is provided through a variety of paths, thereby increasing the probability of finding needed information. Information is presented in a problem-solution approach based on psychological research on why people do not do things effectively. Network access, combined with information annealing, provides the means for rapid and continuous improvement of the guide. The ability to launch computer programs and expert systems and electronically capture the results in a design notebook greatly expands the concept of a design guide.
[1] Deming, W. Edwards, "Out of the Crisis," Massachusetts Institute of Technology Center for Advanced Engineering Study, Cambridge, Mass., 1986, p. 466.
[2] Raitt, Dr. D. I., "The information needs of scientists and engineers in aerospace," in AGARD Proc 385, "The Value of Information as an Integral Part of Aerospace and Defense R&D Programmes," Cheltenham, UK, 4-5 September 1985, pp. 3-1 to 3-5.
[3] Cole, Bernard C., "Uncorking documentation bottlenecks - Electronics firms turn to on-line hypertext for relief," Electronics Engineering Times, June 1, 1992, p. 43.
[4] Bush, Vannevar, "As we may think," Alantic Monthly, July 1945, pp. 101-108.
[5] Berk, Emily and Joseph Devlin, Editors, "Hypertext/Hypermedia Handbook," McGraw-Hill, 1991, p. 4.
[6] Schiff, Jacqui Lee, "Cathexis Reader, Transactional Analysis Treatment of Psychosis," Harper & Row, New York, 1975, pp. 14-18.
[7] Hauser, John, R., and Don Clausing, "The house of quality," Harvard Business Review, May-June 1988, pp. 63-73.
[8] Larson, Neil, "HyGlos(tm) - Hypertext Knowledge Construction System,", 1992, MaxThink, Berkeley, CA, Chapter 9, "What is Information Annealing?"
[9] Ibid., Chapter 7, "What are glossary and thesaurus systems?"
Copyright Notice. The following copyright notices must be displayed on the initial screen displaying IEEE-copyrighted material electronically:
"© 1993 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE."
"This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright. In most cases, these works may not be reposted without the explicit permission of the copyright holder."
For additional copyright information see http://www.ieee.org/about/documentation/copyright/policies.htm.
Original: Foutz, J., Design of a Hypertext Power Supply Design Guide, IEEE Applied Power Electronics Conference Record, San Diego, CA., March 7-11, 1993. Posting to Web without content change April 14, 2003.
Author: Jerrold Foutz, foutz@smpstech.com