By Simon Holdsworth, Product Architect, Integrated Electrical Systems Division, Mentor Graphics Corporation
The field of vehicle electrical system design (ESD) is repeating history written years ago in the realm of IC and printed circuit board (PCB) design. The complex technical nature of these disciplines began to outpace the unaided comprehension of the engineers doing the work. There were simply too many variables to juggle and rules to implement. Today, it is vehicle designers who are confronting an endless array of options, variants, versions and derivatives designed to satisfy the demands of consumers and regulators. Whether the vehicle is a two-seater sports car or a jumbo jet, burgeoning electrical system complexity is often cited as today’s number one challenge for vehicle designers.
For yesterday’s IC and PCB designers, salvation came in the form of electronic design automation (EDA) software. And today’s vehicle designers can now rely on powerful new computer-based tools that embody some of the same constructs and deliver the same benefits of speed, efficiency, and accuracy as their EDA predecessors.
In this article we shall see how complexity’s challenges can be resolved by modern wiring design toolsets. Option codes and expressions can model the variable content of a vehicle program. A full-featured suite of tools can capture best practices and rules to be used in the form of constraints. And automation can help designers manage complexity in a coherent, consistent, and cost-effective manner.
From Definition to Delivery
Figure 1 is a simplified view of a typical vehicle electrical design flow. The steps enclosed within the dashed line are the focal point here. Consider this the “Design Block.” On the input side, the platform definition and content requirements feed into both the mechanical layout and the electrical design. On the output side, the process issues fully defined and costed harnesses for all planned customer configurations. It also generates data for service manuals that can support all of these variants. Historically the physical wiring design has been done manually, but that is changing as option and variant complexity increases and outstrips human capacity to manage it.
Figure 1: Simplified view of the traditional harness design process.
The Automation Paradigm
Compare the automated flow in Figure 2 with the manual process in Figure 1. The automated wiring synthesis process uses four separate inputs, all of which are derived from the same platform definitions shown on the left side of Figure 1.
Figure 2: An automated wiring synthesis process as performed in the Mentor Graphics Capital design environment.
The signal connectivity input tells the synthesis tool how outputs and inputs must be connected, but only in general terms. It is a conceptual “sketch” that might depict, for example, both a CD player and a CD changer connected to an amplifier input. But because the design and configuration rules automatically prevent both playback units from being configured in the same end product, just one of the two will be allowed per harness variant. The signal connectivity input is meant to encompass all potential connections that result from the platform definition and features documents, but it does so independently of the vehicle’s physical constraints. Grounds and possibly other connections may be left uncommitted since their placement is best determined by the location of a related device such as an ECU. This permits the same design to be used in multiple vehicles with differing ground points.
Of course, the system design device connectivity must be mapped to the physical reality of a car or airplane to guide the eventual production and documentation of harness products. The Mechanical Constraints input in Figure 2 defines the physical routing channels for the wiring. The underlying information can be developed manually, but in modern integrated design tools it comes from the MCAD environment, where 3D models are used for space reservation and digital mock-up.
Using a full-featured design platform, the 3D MCAD models can be flattened into a 2D topology that more clearly represents the relative bundle lengths in the harness. It is also possible to create the 2D representation manually and then synchronize the bundles from the MCAD model. In either instance, the 2D image is purely a visual aid that maintains its link to the 3D MCAD model. The goal is to ensure that the bundle lengths in the topology are as accurate as possible, since these will be used later to calculate the lowest-cost wiring implementation. It is a simple matter to position and connect the 2D elements together to form the full harness. For example, the passenger door harness, the dash harness, and the driver door harness can be joined via inline connectors to other harnesses imported from separate MCAD models. The result is a cabin harness whose bundle lengths and positions are consistent with the MCAD data, as shown in Figure 3.
Figure 3: Harness topology as seen in a view produced by Mentor Graphics Capital Integrator tools.
From Development to Management
The remaining two inputs are every bit as important as the connectivity and mechanical inputs. They are concerned with the management of the design variations that are the essence of the complexity challenge.
Each individual feature in the vehicle (such as fog lights or a CD player) is represented by an option code. In a typical vehicle program there may be dozens or even hundreds of these codes, which are usually imported into the design tool environment from another system. Variable vehicle content is defined using these option codes, and objects in the system design and the 2D topology are tagged with boolean combinations of these individual codes. These option expressions are used throughout the process to indicate the applicability of each object to a specific vehicle configuration.
What are the relationships between all these vehicle features? Recall the example from the connectivity data. A CD player and a CD changer were both shown as “routed” to the same amplifier input. But the toolset stipulates that these two configurations are mandatory and mutually exclusive. This means that every built vehicle must have one these options, but only one.
The final input is the group of design rules that defines exactly how the automated synthesis should implement wiring using the enterprise’s best practices. This know-how is built into sets of configurable constraints that are defined and assigned to the objects within the design. Constraints can control anything from the physical placement of devices within the topology, to the exact implementation of a CAN network. Rules take the form of the following examples:
• Wire Spec: Wire Color is G where Signal matches Type= Ground
• Don’t route signals matching Class=Power through here
• Cost of splice = 100.0 where Signal matches Class=Safety
With the data from all four process inputs (Figure 2) the design toolset is ready to associate the system connectivity with the vehicle’s topology. This is done automatically, of course, and the tool displays all of the devices which are defined on the system designs. The result doesn’t specify the actual placement of all the devices (which range from tiny fuses to airbags and CD changers) in the system but thanks to the pre-defined placement constraints, this step can be simply automated (a process known as “auto-place”).
Proliferating “Levels”
Before synthesizing production wiring it is necessary to decide how many build levels are needed for each harness family. As the number of levels increases, there is a corresponding increase in the cost of managing these individual harness parts. There is an optimum point where cost and complexity are both brought to their practical minimum relative to each other.
The process of calculating this point begins with determining the number of unconstrained harness levels. The majority of the options affecting a harness come directly from the devices attached to it, but wiring synthesis makes it possible to consider the effect of pass-through options as well. A pass-through occurs when a wire required for a particular feature “passes through” a harness in order to take a signal from one part of the vehicle to another. For example, a switch in the driver’s door may operate the passenger window motor, so this feature may pass through the instrument panel.
When both device options and pass-through instances are unconstrained, the number of separate variants can become impracticably large—potentially hundreds of variants to support in manufacturing (including formboards, version control, etc.), service, and documentation. Giveaways can help balance this by “writing off” the cost of some wiring and leaving it in the vehicle even if the customer doesn’t order the associated features. Astute giveaway choices can reduce the number of levels by 50% or more, saving more cost than the value of the sacrificed wires.
The Composite Wiring Synthesis process, the final step, considers all the possible vehicle configurations and creates wiring to satisfy all of them. This automation can significantly reduce build errors and is typically much faster than a manual process. Every wire and its splices is calculated as the optimal implementation of each signal, based on the defined constraints. Moreover, wires are automatically assigned appropriate option expressions and associated with the appropriate harness levels.
The finished wiring diagrams provide valuable service documentation and material for design reviews. Here again, automated tools can generate the wiring designs using configurable styling that conform to enterprise standards and preference.
Metrics Can Support Optimization
Consider some of the elements that can be controlled by configurable rules and constraints:
• Characteristics and properties of synthesized objects such as wires
• The placement of devices, with multiple or variant placement of devices at several positions within the topology.
• The placement of ground connections (typically to minimize the wire length)
And there are even several constraints for controlling synthesis itself, for example defining the trade-off between splices and additional wire length.
A change in constraints can cause a corresponding change in the synthesized wiring. Suppose it were possible to quantify such changes and measure the impact of the design decisions? Today’s most capable design platforms have just such a capability. They use configurable metrics to quantify the “goodness” of a design decision.
In the field of harness design, “goodness” may mean cost, weight, or other crucial characteristics chosen by the designer. Among the most common metrics are those that plot the financial cost effect of any design change. Figure 4 illustrates the kind of issues that might be explored when optimizing a harness design. All of them include cost as a key factor, but weight and other impacts can be similarly evaluated.
Figure 4: Metrics help designers answer these questions and more.
The overall cost of the wiring is made up of wire cost, connector cost and splice cost. The bar chart in Figure 5 summarizes the proportions of these costs for a harness in the baseline configuration of the car above. Imagine a proposal to eliminate a splice by routing a particular power signal from the battery to the power distribution box via an inline connector through the instrument panel (IP). The practicality of the change cannot be divorced from its cost, so a comparative evaluation is in order. Simply by changing a constraint from the default setting which allows splices to one that prohibits splices (in the harness of interest) it is possible recalculate the routing without a splice. The “spider” diagram on the right in Figure 5 maps the impact on the various harness families and graphically reveals the implications of the change. Clearly the IP harness has seen a significant increase in cost. This information is detailed in a spreadsheet-style document similar to that shown at the bottom of Figure 5.
Figure 5: Using a series of iterative changes, the impact on cost, weight, and other key variables can be observed and documented.
The increase in the cost of the IP harness has outweighed the slight decrease in the cost of the engine harness. Other metrics such as weight, or wire length, can be similarly compared. It is even possible to measure the number of harness levels required for a design in order to predict complexity.
Each metric is actually calculated as the sum of several elements. The foregoing discussion is based on a separate cost element that encompasses connectors, splices and wires. But this is completely configurable and a user can add more elements as required. To calculate the cost impact of an individual connector type, one would either acquire the value from a parts library or build up a set of estimating criteria. The estimate can be as simple as multiplying the number of cavities by a constant to create a “placeholder” figure that can be fine-tuned as more information becomes available.
Conclusion
The notion of using rules and constraints to guide a series of “change vs. impact” calculations brings this article to its logical conclusion. Today’s powerful computer-based electrical system design automation tools make it possible to create a virtual environment in which every harness component, bundle, and connection is defined and tested long before the first hardware prototype need be assembled. And within that environment it is easy to explore the impact of changes while optimizing the harness for both performance and profit.
Simon Holdsworth has more than 20 years experience in the EDA industry and has worked within the wire harness electrical distribution sector since 1998. He joined Mentor from Innoveda and previously Transcendent, where he worked on the development and deployment of a new generation of electrical design tools with a major automotive manufacturer. During this time, Simon has held primarily technical roles in development, marketing and consultancy and is currently a technical specialist focused on wire synthesis and automation. Simon started his career as an Electronics Engineer and has an MA in Engineering from Cambridge University.
