Structured Electronics Design

Structured Electronics Design#

In sections Selected topics from systems engineering and Electronic information processing , we have summarized topics from systems engineering and information processing. Together with statistical signal processing, control theory and network theory, these disciplines are the basis for Structured Electronics Design.

The idea behind Structured Electronics Design is the creation of a design language. In the next section, we will give an outline of such a design language.

Outline of the design language#

Like any language, the design language must have a set of words and rules

, and we want to create an unlimited amount of stories using only relatively small sets of words and rules. The stories we want to write with the design language are electronic information processing systems.

The words of the design language are the so-called basic functions. We state:

Definition

The functional behavior of any electronic information processing system can be obtained from a combination of basic functions.

The materialization of a function in an object requires the implementation of one or more physical operating principles in some technology. Therefore, physical operating principles are part of the grammar of the design language.

Due to the physical, technological, and economic limitations, the performance of these objects may be insufficient or the cost to achieve their desired performance may be too high. The application of techniques and methods from network theory, control theory, and statistical signal processing can improve the performance-cost ratio. Hence, these so-called error reduction techniques are also part of the grammar of the design language. Application of error reduction techniques changes the functionality or optimizes the implementation of one or more operating principles.

We now have the design language with its words and its grammar:

Definition

With relatively small sets of basic functions, physical principles, and error reduction techniques, we can create an unlimited amount of information processing systems.

Fig. 7 illustrates the design of a physical composition from a functional decomposition.

At this moment, the above may sound rather abstract, and the concepts introduced require further explanation. In section Basic functions and basic objects, we will introduce a set of basic functions and discuss their materialization in basic objects. Materialization or implementation is the application of one or more physical operating principles in available technologies. In electronic systems, the implementation of functions is based upon operating mechanisms of electronic devices.

The reader is assumed to be familiar with elementary physical principles from electrostatics and electrodynamics. The operation of semiconductor devices is summarized in Chapter Active Devices.

In section Performance, costs, and FOM we will summarize important performance aspects and costs factors of electronic information processing systems and give examples of Figures Of Merits.

In section Error-reduction techniques, we will give some examples of error reduction techniques. In this book, we will apply:

  1. Balancing techniques

  2. Negative feedback techniques

  3. Frequency compensation techniques

In the concluding section Differences with traditional analog design we will discuss the differences between Structured Electronics Design and the traditional approach to the design of electronics.

Basic functions and basic objects#

Top-down definition of basic functions#

One way of defining basic functions is to consider elementary mathematical operations required for information processing. Information processing deals with performing operations on time-varying physical quantities.

Below, is a list with basic information processing and reference functions for the functional decomposition of information processing systems:

  1. Addition of signals

  2. Subtraction of signals

  3. Multiplication of signals

  4. Integration of signals

  5. Differentiation of signals

  6. Selection of signals

    Mathematical selection is simply comparing two variables. The physical nature of the variables is not of interest. From the viewpoint of information processing, it is useful to distinguish comparison in different physical domains:

    1. Selecting of signal levels

    2. Selecting of signal frequencies

    3. Selecting of signal time intervals

    4. Selection of signal locations

    Selecting also requires a reference, which brings up the group of basic references:

  7. References

    1. Level reference

    2. Frequency reference

    3. Time reference

    4. Location reference

  8. Shifting of signals

    Mathematically shifting of a signal is adding a constant to it. From the viewpoint of information processing, the physical domain is of interest:

    1. Level shift (adding a constant)

    2. Frequency shift (modulation)

    3. Time-shift (memorization)

    4. Location shift (transportation)

  9. Change the power of a signal

    In the context of information processing, it is also meaningful to consider the two types of variables associated with physical signals: across variables and through variables. Examples of across variables are voltage, force, and pressure. Examples of through variables are current, velocity, and flow. The product of an across variable and its associated through variable at a certain time instant is the instantaneous signal power. The ratio of an across variable and its associated across variable shows a relation with the impedance of the signal source or load. For maximum power transfer from a source to a load, the load impedance needs to be the complex conjugate of the source impedance. The maximum power that can be delivered by a signal source is called the available power of that source.

    Basic functions related to the power transfer are:

    1. Impedance transformation (optimization of power transfer)

    2. Attenuation (reducing the available power)

    3. Amplification (increasing the available power)

  10. Solving

    Solving an equation is considered an essential mathematical operation often applied in control theory. In network theory, the nullor provides this function. A nullor consists of a nullator and a norator. The nullator sets a condition and a norator that provides the dependent variable to be solved.

The set of functions listed above is not a minimum set of orthogonal functions. As an example, consider frequency shift. This function can be resolved in multiplication of the signal with a frequency reference

and selection in the frequency domain. Another example is the level shift. This function can be replaced with addition and a level reference.

Bottom-up definition of basic functions#

A basic object is a physical implementation or the materialization of a basic function. There exists not necessarily a one-to-one mapping of basic functions onto basic objects. The materialization of basic functions in basic objects strongly depends on the physical operating principles in the available technologies. Moreover, a physical operation principle in some technology may contribute an attractive function not listed above. Bipolar transistors, for example, exhibit an exponential relation between their base-emitter voltage and their collector current. This relation holds over many decades. Resolving the functionality of a system into \(\exp(x)\) functions may then result in a straightforward mapping of the functional decomposition on the physical composition in bipolar technology. For example, the design of integrated circuit analog multipliers is based upon the application of the exponential function and its inverse. In this example, multiplication is decomposed into lin-log conversion, addition, and lin-exp conversion.

Implementation of basic functions in basic objects#

There are many ways to materialize basic functions. As an example, consider the implementation of impedance transformation. Both transformers and impedance matching networks can be used for this purpose. These two implementations use different operating principles. A transformer uses the principle of electromagnetic induction, while the operation of a matching network is based upon resonance.

Another example is the selection of time intervals. This function requires a switch. Switches can be realized with nonlinear electronic devices or with electromechanical devices, such as relays. These two implementations use different physical operating principles and technologies.

In this book, we mainly focus on the design of application-specific amplifiers. The amplification function increases the available power of a signal. Its embodiment requires a power source and a mechanism to control the power transfer from this source to the load with the input signal.

Performance, costs, and FOM#

As discussed in section Basic design process, the performance and the costs of the object are specified in the Object Performance Specification (OPS). An in-depth treatment of the structure of the Object Performance Specification for electronic information processing systems is beyond the scope of this book. Companies usually use their own templates for it. Fig. 4 shows an example of the contents of an Object Performance Specification for a Printed Circuit board Assembly (PCA).

In section Basic design process we also introduced the Figure Of Merit as the most compact representation of the performance-cost ratio. Throughout the design process, design decisions can be based on this FOM.

An example of a FOM for electronic information processing systems is the amount of information that can be processed per unit of energy and per Euro:

\[FOM=\frac{\text{bits}}{\text{Joule}\cdot\text{\euro}}\]

In this book, we confine ourselves to the circuit design of amplifiers. It is impossible to define a figure of merit for amplifier design because the relevant performance aspects and the relevant cost factors strongly depend on the application and the technology.

Error-reduction techniques#

In most cases, the materialization of a function in an object proceeds stepwise. This is because the performance-cost ratio of a first physical design proposal is not as required and needs improvement. The performance-cost ratio can be improved by applying error-reduction techniques. We distinguish two groups of error-reduction techniques.

  1. Improvement of the performance-cost ratio while maintaining the way the information is coded in the signal.

    Techniques that belong to this group are:

    1. Optimization techniques

    2. Compensation techniques

      Examples of compensation techniques are:

      1. Balancing techniques

      2. Frequency compensation techniques

      3. Error feedforward techniques

    3. Negative feedback techniques

  2. Changing the way the information is coded in the signal.

    Examples of techniques in this group are:

    1. Sampling

    2. Quantization

    3. Modulation

    4. Coding

../_images/IDEF0-SEDextended.svg

Fig. 8 Detailed process model of the implementation of functions in objects according to Structured Electronics Design.#

Fig. 8 shows a more elaborated model of the physical design. It shows that the application of an error-reduction technique may change the functional decomposition.

Differences with traditional analog design#

Traditionally, the design of analog electronics predominantly consists of analysis and improvement of the behavior of known circuits. These improvements, in turn, usually consist of applying commonly known changes, and the impact of such changes on the cost factors often remains unclear. In addition, these circuits and their improvements often carry the name of their inventor or their topology. As a result, it is difficult or impossible to recognize the intended functionality, the applied operating principles, and the applied design methods with their intended effects. Therefore, the traditional design approach is a rather heuristic process.

A clear separation between functions, the applied operating principles, error-reduction techniques, their intended effect, and their technological implementation is the distinguishing difference of Structured Electronics Design. Moreover, Structured Electronics Design uses circuit analysis for setting-up design equations for well-defined performance parameters and cost factors, and numeric simulation is considered a verification method rather than a design method.

The advantages of Structured Electronics Design for analog design automation and design education are evident, and presented below.

  1. There exist two different approaches to design automation:

    1. A big-data oriented approach

      This approach searches a circuit database circuits for viable circuits that solve a set of design problems. However, new problems cannot always be solved with known solutions. In many cases, these solutions require modifications. Modification of existing solutions without knowledge of the underlying principles is risky. It may result in show stoppers, and it requires extensive simulations to prove the viability under all possible conditions. The above can only partly be solved using parameterized designs.

    2. An algorithmic approach

      Structured Electronics Design essentially is an algorithmic approach. It is very well suited for design automation. This approach identifies show stoppers at an early design stage and requires less extensive simulations because it considers the viability of a solution throughout the design process. Moreover, an algorithmic approach can invent new solutions because it does not start with an existing solution but builds up from well-defined concepts.

  2. Design education requires an approach that facilitates the internalization of newly acquired knowledge. Therefore, the structured design approach is well suited for design education. Each step of the design process can be motivated because it uses principles from physics, signal processing, control theory or network theory, and last but not least: principles familiar to the student.

Contents