5 Steps to Make Your Electrical Engineering Analysis Faster and More Reliable

Step 1) Turn Your Design Parameters into Variables That Can Be Easily Updated

Design parameters can change mid-way through a project for many reasons. For example, the customer may determine the original requirements were missing an important limiting factor or scenario, and the sales team agrees that the new constraints must be adopted. Alternatively, when supply-chain shortages affect the availability of some components, a different part or parts may need to be substituted in, changing the tolerance values, the bill of materials and also the circuit schematic.

It is very time-consuming to manually update values multiple times in engineering calculations – and long documents make it more likely that data-entry errors or misapplied units will occur. Whilst most engineers define their parameters at the start of their reports, documents can be improved by using a calculation tool that supports variables instead of constants, and by adopting a standardized format with subscript notation.

Defining each parameter as a symbolic variable gives a letter or symbol to represent each appearance of the parameter. This makes it easy to reference and to trace the usage of the values throughout the document. This also supports data provenance through the report. Explanatory notes can denote the source and development of the variables to help a colleague to trace the parameters and see that the equations are correctly applied right up to the point that they need to be evaluated.

When design constraints change, the symbolic variable can be updated once at the first reference point, with the updated value flowing through the pages of equations to keep all the other assumptions and data the same. The calculations and results are then updated automatically. This means your worksheet can be developed as a template and be reused in the future when a similar project occurs.

Step 2) Use Powerful Calculation Software to Rapidly Perform Math- based Analytical Methods and Run Monte Carlo Simulations

Performing tasks such as simulations, mathematical modeling, and analysis of the electrical system’s behavior under different conditions can be extremely time intensive. The calculations are certainly valuable by contributing towards the safety, reliability, and longevity of electrical equipment, but the number of factors increases the mathematics involved and requires more computational effort to include them in a model. Simultaneously managing multiple design requirements quickly leads to complex non-linear mathematical expressions, especially when considering all the component parts in a circuit. This precludes performing the calculations by hand, but does mean that sufficient detail, such as temperature variation, can be added to truly reflect the real-world operation.

Engineers can save time by using math-based techniques during their calculations. Command-line solvers available in the calculation software can be called on from within the analysis to speed up Component Trade- off assessments and quickly find ranges of operating conditions that simultaneously satisfy multiple design requirements.

Due to imperfect manufacturing processes, real-world components do not have fixed values for output current, resistance, capacitance, etc. and so need to be designed within some tolerance interval. Component ratings are defined in terms of an average value with some variation. This variation can be defined as a percentage (for example, with resistors) or as a magnitude (for integrated circuit datasheets).

The following mathematics-based analytical methods can be used to perform a functional verification, i.e., to compare the designed circuit’s performance against the design specifications, and quantify the variation.

• Extreme value analysis (EVA) – This is an estimate of the most extreme limits of the circuit’s components and function. The behavior of a circuit is simulated for every permutation of extreme component parameters and each component is varied to obtain the absolute worst-case values of the circuit performance.
• Sensitivity analysis (SA) – This determines how changes in a set of input variables are related to the change in some target variable or function. The goal is to identify which input variations produce the largest variation in the circuit’s output. In calculation software you can calculate the symbolic or numeric partial derivatives of the circuit with respect to each component parameter. These can be used to perturb the circuit equations. Engineers typically use one of two methods to numerically estimate the likelihood of the output variations, namely Root-sum-square and Monte Carlo analyses:
• Root-sum-square (RSS) analysis - This uses a statistical approach, assuming that most of the components follow a normal distribution and fall to the mid of the tolerance zone rather than at the extremes. This is a simplified model that takes less computational effort but loses accuracy when the circuit contains arrangements of non-linear elements found in complex electrical systems (i.e., that utilize feedback) and can even lead to instability of the system.
• Monte Carlo Analysis (MCA) – This method includes the performance variation of each component as part of the model and runs a series of scenarios for possible results by substituting a range of values (via probability distributions) for any factor with inherent uncertainty. Next, it calculates results repeatedly, utilizing different sets of arbitrary values from the probability functions. In this way, parameters are randomly selected from a distribution, and the circuit simulated anywhere from 1,000 to 1,000,000 times. This provides a more robust method when non-linearities are involved. A mean, variance and confidence level for the result can be calculated, and used to compare with the required design specification. These thousands of permutations are evaluated alongside the probability that they will happen. Although these calculations are computationally intense to perform, modern calculation software allows engineers to rapidly run the Monte Carlo simulations.

In fact, there is no longer any reason to use RSS to simplify assumptions especially if that introduces error into the Sensitivity Analysis expressions. The error introduced may at first look small in most cases, but when thousands of calculations are performed it quickly grows to have a big impact on the circuit function.

Step 3) Collect Your Component and Reference Data into a Central Repository

Electrical engineers have many sources for project constraints. This can be supplier provided information in the form of component property tables, or customer provided, such as tolerances and operating conditions. Designs may need to follow assumptions and best practices that have been adopted internally by the company or include informational reference material. This data can be presented in raw number tables, or can be structured with dependencies.

Engineers can greatly benefit from reusing and recycling the data from component lists and design requirements, but only if they are stored centrally and are used consistently across projects.

The first step is to recognize the types of data that is reused across projects or even multiple times in the same worksheet. To prepare for the analysis, the circuit models are expressed as equations and the parameters, derating factors, limits and Min/Max/Nom Tolerances are gathered.

The source data may include units or be structured into equations or tolerance ranges or codebases. If the data fits into a table, then it can be setup as a csv or Excel format before being parsed into matrix form in the calculation software.

When these data assets can be used directly, then their value as a shared resource becomes higher, and this will support a standardized solution across a team of engineers. When the supporting information, such as the system of units or tolerance range, is not passed into a calculation tool, the engineer has to spend effort and time to restate this in later definitions, and there is an opportunity for errors to occur.

The engineer has the responsibility to assess the reliability of each source – for example, if the information is provided by a manufacturer, there should be some check on the validity of the derating tests and the provided model. Supplier models typically come with disclaimers that the product is not guaranteed to behave the same as the model, and suppliers rarely publish any model-to-measurement correlation data. This can be solved by working directly from the trustworthy items listed on component datasheets, and applying them to simple models that fit the datasheet specifications.

Once an established derating approach is selected, the use of a repository for component properties and derating values will encourage consistency and avoid duplication of effort.

Step 4) Add Automation to Change Analysis Steps into an Agile Process

The sequence of designing and developing an electrical system involves multiple stages, and often hand-off of findings between tasks and even teams.

The initial design will typically include a schematic diagram and common circuit development platforms can turn this into a “netlist” file to represent the hierarchical relationship in textual form, and develop a bill of materials listing the system components.

Once the netlist, subsystem variables, and source load information are available, engineers often resort to manually transferring this information into separate calculation software to complete the analysis. This takes time and introduces the risk of data-entry errors, plus also creates a challenge for reviewers to ascertain whether the numbers used in equations and mathematical analysis were correctly derived.

When conducting Worst Case Circuit Analysis, traditionally engineers will perform the calculations and pass the results between tasks. For example, the Extreme Value and Stress Analysis results are compared to the Design Margin and other requirements and assessed as Pass or Fail. If any of these results show that the product failed a requirement, the design has to be changed and the full analysis rerun using different parameter values, resulting in a lengthy linear process.

Lack of automation in the worst-case analysis process makes it difficult to keep the analysis in sync with design changes. The analysis should be easily repeatable when components, design, operating conditions, or requirements change.

The process can be made more productive by switching to an agile process that can respond quickly to upstream updates. This can be achieved by parsing the structured netlist data into a format that can be manipulated with calculation software. By connecting the circuit design netlist to the calculation software, then the parameters, and component list can be automatically derived, and through scripts the Bill of Material and the model expressions can too be aligned to the circuit as portrayed in the netlist.

Step 5) Perform Your Analysis and Report Your Findings in the Same Tool

The workflow of creating electrical engineering analysis can be greatly simplified by reducing the number of calculation tools used to pass results back and forward. If the engineer defines the parameters and conducts all the stress analysis and derating limit values in the calculation software worksheet, then the results and key design information such as the Bill of Materials can be presented together in a report in the same tool.

Clear presentation of the data and analysis makes it easier for peers to review or to share project work with other teams. Using standard math notations, integrated text and graphical displays, calculation software can automatically generate readable documents that are easily understood up and down the management chain.

Project documents are more valuable if they can be easily adapted for future design assignments. Having the analysis and results together in a single worksheet allows the design documentation to be used as a template which promotes easy knowledge transfer to colleagues.