The Science of Design

by

Kenneth A. Kuhn

October 25, 1994

revised Jan. 31, 2000

Note: This paper is an html adaptation for posting on the Internet of the original paper presented during a talk at a UAB student IEEE meeting in the Fall of 1994. As time permits I will be making a few revisions and additions to this paper.

Much more complicated than is generally credited, design is not some magical brew concocted by a witch doctor. Design is a calculated application of scientific knowledge, intelligence, judgment, and creativity. This paper describes the process of design from both scientific and philosophical perspectives. The intent is to direct the beginning designer down the right path towards becoming a true designer. But first, a foundation in the basics of design needs to be built. Although this paper is written for electronic engineers the concepts are general to any engineering.

The Two Types of Design

Re-design is the very common practice of analyzing an existing design to learn how to adapt the design to a different circumstance. This process is also known as reverse-engineering and is an excellent way for new designers to learn the design process. Note that analysis alone will not reveal the judgment of the original designer. This can only be resurrected by going through the design process and applying extensive creativity. Design begins with the difficult process of inverting the analysis to determine the design equations. The new designer then uses creativity to determine the plausible judgments the original designer must have used. Only after this process is the new designer able to make the appropriate changes to the original design to make it suitable for a different circumstance.

Original-design is a function of creativity, judgment, intelligence, and knowledge. Original-design is the creation of something out of nothing -- i.e. the first time it has ever been done. The creative skills required to do original-design come from experience in the re-design process. Designers may do original-design only a few times during a career. These occasions are career high points and may earn the designer a place in engineering history. The original-design process may be inadvertently used when the designer is unaware that the design already exists.

The Mathematics of Design

The design process can be represented by a kind of pseudo mathematics to represent the complex functions that must be performed by the designer. There are two basic operators as described below.

Analysis = Design-1. Analysis is akin to differentiation and is relatively easy. It takes the knowledge of how to analyze, the intelligence and judgment to make rational simplifications and some creativity to infer reasons for the particular implementation. The mathematics has exactly one solution which is the set of equations that specifies the performance characteristics of a given set of components. Just like in calculus where a constant term disappears when taking a derivative, the judgment in the original design disappears when doing analysis.

Design = Analysis-1 + Judgment. Design is akin to integration and mathematical inversion of the analytic equations is relatively difficult. The solution is a set of equations that specify a set of design components to meet specified performance characteristics. There are usually infinite solutions and considerable judgment is necessary to choose a best solution for a given application.

The re-design process can be described as follows:

  1. Analytic equations = design-1(existing design). The existing design is dissected to determine the analytic equations.
  2. Design equations = analysis-1(analytic equations). The analytic equations are inverted to produce the design equations.
  3. Re-design = design equations + judgment. The proper judgment applied to the design equations will complete the design.

The original-design process can be described as follows:

  1. Proposed design = creativity(given performance characteristics)
  2. Analytic equations = design-1(proposed design)
  3. The symbolic equations developed are checked to be sure they relate performance characteristics with performance requirements. If any given performance characteristics do not show up in the analysis then the proposed design can not work.
  4. Design equations = analysis-1(analytic equations)
  5. Original-design = design equations + judgment
  6. Numerical values for the performance requirements are substituted into the design equations to verify that the required components are realizable. It is very possible that negative or imaginary components may be required. If this happens then the proposed design will not work and another approach must be tried.

Notice that after the initial creativity the original-design and re-design processes are very similar. Note also that original-design is an iterative process whereas the re-design process is not.

The Four Components of Design

Judgment and creativity are two very important components of design. These two components can only be performed by human designers -- it is doubtful that computers will ever take over these functions. Design algorithms are incomplete without human judgment. The original-design process can not even get started without human creativity. Knowledge and intelligence are components of judgment and creativity. Computers are very good at storing knowledge and to a lesser extent implementing human guided intelligence. These four components are discussed below beginning with basic knowledge and concluding with creativity -- the ultimate in thought process.

Knowledge is simply the accumulation of facts. Knowledge in of itself can do nothing. A book, computer, or human with only knowledge can not make any decisions or accomplish any function. Knowledge is an entity that is easily transferable from one user to another.

Intelligence is the ability to use knowledge. Thinking is applied intelligence. Skill is also the application of intelligence. Through intelligence, one infers complex relationships among pieces of knowledge. These relationships form the basis of an action to perform for a particular situation. Intelligence can not be learned -- it only can be refined by a complex function of experiences both good and bad. Therefore, intelligence is not a transferable entity.

Judgment is a higher form of intelligence. Judgment is the ability to make good decisions when the knowledge base is incomplete or inaccurate. Under this situation, basic intelligence alone can not achieve a conclusion. Through judgment, one infers pieces of knowledge that are not explicitly known. Intelligence is then used to test this pseudo knowledge for validity. Like intelligence, judgment can only can be refined, not transferred. If you can learn from your (or other's) mistakes then good and bad judgments together enhance your judgment ability.

Creativity is a complex process using knowledge, intelligence, and judgment to achieve a particular objective. Where as intelligence and judgment can modify an existing something into something new, creativity is the ability to transform nothing into something. Through creativity, one infers both a knowledge and intelligence base and then uses judgment to check validity. There are many different combinations of knowledge, intelligence, and judgment to achieve a particular objective. Creativity does not lend itself to any hard rules or procedures. Creativity can not be managed -- the greater the attempt to manage it, the less of it will exist.

General Design Procedure

While different designs will require different procedures, the following general procedure is intended to focus on what is important at each step. Exceptions and additions are made using the judgment of the designer.

  1. Clearly identify what is to be done without any concern for how it will be done. Identify all available or needed inputs and required outputs. Also, determine what the user interface is to be. Note that at this stage it is inappropriate to think about implementation details such as block diagrams, circuits, software or other algorithms -- doing so will only corrupt your thinking. Only think about the entity as a whole.
  2. Determine the mathematical relationships that will produce the required outputs from the available inputs. This step involves knowledge of the appropriate techniques, the intelligence to use them wisely, and a little judgment where exact solutions are not possible. Again, do not clutter your mind with dreams of the implementation details.
  3. Formulate a block diagram for implementing the required mathematics. This is the process of partitioning and sequencing the mathematics into manageable pieces in preparation for hardware or software implementation. Do not let favorite implementation methods corrupt the partitioning.
  4. Analyze each block and decide what methods of implementation are appropriate. This is a preliminary step to detail design. Do not waste time thinking about the details of implementation. Only think about general valid methods.
  5. Extensively review the design process thus far and decide if any changes would improve things.
  6. Now you are ready for the detailed implementation of each block. This is the fun step that engineers like to do. Note that it is the last step in the design process. In this step the engineer can use all abilities and skills to produce the most appropriate implementation.

Learning to Design

The first step in learning to design is to analyze the designs of others. You must be proficient in analysis. Then you must be proficient in the more difficult task of inverting the analytic equations to produce design equations. Then you must possess sufficient knowledge, intelligence, judgment, and creativity to establish the basis for choosing the best solution of the design equations to a particular circumstance. After you are proficient at these tasks, you are ready to perform original design.

There are some common misconceptions about design. Design is not copying a circuit out of some electronics made simple book, a magazine article, or an application note. Any non-engineer such as a hobbyist, technician, or engineering-technologist can do that. Design is also not just the result of some algorithm or computer program. Computer programs do not have intelligence much less judgment. Only a competent designer has that. Computers, though, can be a tremendous aid to a qualified user.

Design techniques can be shown and practiced but it takes much more than knowledge of a few techniques to be an true designer. Becoming a true designer is a long process that involves considerable individual study of many things, observation and study of the techniques of an experienced designer (even better if you are fortunate enough to be an apprentice to one), and many experiences both good and bad. Any short cut will result in your becoming another pseudo designer. The problem with pseudo designers is that they do not know enough about what they are doing to realize their ignorance. Thinking they already know it all, they have difficulty learning something new. The beginning designer must be able to distinguish pseudo designers from true designers and learn the right lessons from the true designers and not be corrupted by the pseudo designers. The following humorous illustrations should help point out the difference.

Pseudo designer: Skip the design process and go straight to implementation. Implementation is more fun (and easier) than the design process. Implementation is what you learn in school.
True designer: Is horrified to skip the design process; knows that time saved in skipping the design process will spent many times over in the debug phase where it is too late to do it right.

Pseudo designer: The "More is better" syndrome. The pseudo designer uses what is assumed to be massive overkill to make up for lack of analysis and ability. The pseudo designer does not comprehend that there could be negative consequences of overdoing something.
True designer: Calculates what is really needed and strives to implement just that; Uses a calculated margin of safety instead of assumed overkill.

Pseudo designer: Either does not realize that a mistake has been made or learns the wrong lesson from a mistake.
True designer: Makes mistakes but recognizes and learns from them.

Pseudo designer: The "Engineer by fear" syndrome. The pseudo designer will include things in pseudo-designs that have no relevance other than to appease various gods of the unknown whose wrath the pseudo designer is trying to avoid. The pseudo designer is trying to mask inability by mimicking things that others have been observed doing, but without understanding.
True designer: Can explain why every component is in a circuit or system and can explain why each component is a particular value. If the true designer is concerned that the circuit or system needs protection from something, then the extent of that something is determined and then designs in calculated protection. The true designer is horrified to leave something to chance. Otherwise the true designer does not waste time including extra junk in a design.

Pseudo designer: The "Gut feeling supersedes any analysis" syndrome. The pseudo designer is confident that gut feeling is the optimum solution to a problem. It is clear then, that the pseudo designer's brains are in the intestines.
True designer: Has gut feelings but knows how to verify or discard them with the appropriate analysis.

Pseudo designer: Assume that a favorite implementation will achieve the required performance.
True designer: Has favorite implementations but is open-minded to look at appropriate alternatives.

Pseudo designer: Any complexity that the pseudo designer either does not understand or know how to implement is trivial and can be ignored.
True designer: Seeks to gain knowledge in any area where a deficiency is felt. Also seeks the advice and leadership of someone who is proficient.

Pseudo designer: The "I can hide my ignorance by doubting you" syndrome. The pseudo designer does not want own ignorance to be found out and will use doubt to infer knowledge as much or more than the true designer.
True designer: Does not doubt someone unless there is a scientific basis for doubt. Can also identify what would be generally required to clear the doubt.

Pseudo designer: The "Not invented by me" syndrome. If it is not done exactly the way I would have done it, then it is inferior.
True designer: May like own way better but has no objections if another way works sufficiently well. Recognizes that there are generally many ways to implement something. The true designer does not worry about achieving the ultimate optimum in cases where optimums are very broad.

Pseudo designer: The "instant expert syndrome". The pseudo designer has just heard about something new and is already an expert on the subject.
True designer: Is eager to learn new things but would not make any judgments without consulting an experienced designer on the subject.

Pseudo designer: The pseudo designer thinks of own abilities as having little room to improve and is very happy being unaware of the shortcomings of pseudo designs. Any malfunctions are blamed on the magical nature of science.
True designer: The true designer is aware that own knowledge, intelligence, and judgment are less than perfect and continuously strives to improve. Is concerned that shortcomings may lead to malfunctions later.