Cutting design costs: A contrarian view

The industry has been under tremendous pressure during the past few years to reduce the cost of design for semiconductors and all electronic equipment. Driven by ever-decreasing prices in the end markets and the expanding global economy, companies are closely examining their research and development costs. This has led to decreased R&D expenses as a percentage of revenue.

Unfortunately, too many companies have focused on the wrong elements in their attempts to reduce design cost. The following are five simple organizational and process changes that can dramatically reduce design cost.

Pace your group
The first key to reducing design cost is to accelerate the design pace. There is a direct correlation between development schedule and development cost. Even if the total man-hours are estimated to be the same, a program with a longer schedule will invariably cost more. A corollary is that a late program is also an over-budget program.

To increase design speed, take this basic step: Assemble an experienced team with the required skills and one whose members have previously worked together. It is critical for the group to have this collaboration experience; otherwise, there will be delays and misunderstandings as the team dynamics evolve.

An essential element in pulling together such a team is the maintenance of a senior, stable engineering group. Obviously, this cannot be done in an atmosphere of high turnover. Many companies do not fully take into account the inefficiencies created by hiring on the upturn and reducing head count on the downturn.

It takes time to build a high-performing engineering team. If the team's makeup changes significantly from year to year, a great deal of inefficiency is introduced into the process.

It also takes time for engineers to feel ownership for their project, for key relationships to form and for the team simply to learn the various processes used in the organization. All of those elements are necessary to form a highly effective team.

In addition to turnover in the engineering organization, another common problem occurs when companies move engineers frequently from program to program to meet short-term schedule needs. While this is sometimes necessary, every attempt should be made to ensure that these assignments are rare and temporary.

Find key players
The current trend in which companies hire cheap engineering, typically offshore, is the antithesis of building a senior, experienced team. If offshore organizations are engaged, it is critical that they maintain senior, stable engineering groups as well. This can be a real struggle in some emerging high-tech areas, such as Bangalore, where companies such as Electronic Data Systems Corp. have reported professional staff turnover rates as high as 40 percent.

Although experienced engineering teams will command higher salaries, the resulting high-quality design, faster execution and fewer errors more than offset the cost. Mistakes are extremely expensive in a complex equipment program. A design that does not meet performance or misses the cost of goods sold targets will need to be redesigned, causing delays and increasing cost.

When such mistakes are made, one of the most common causes is that a junior engineer has been assigned a problem that's too complex. This may have been done with good intentions--it gives that individual an opportunity to learn, for example, or maybe no one with more experience was available to take the project.

Problems arise, however, when the necessary support structure is not present. For instance, if a senior engineer tasked with helping is overloaded and cannot spend sufficient time reviewing and coaching the design process, the result is a poor design that does not receive appropriate attention until late in the development cycle. At that point, the product may already be committed to a customer, so delaying shipment would not be acceptable.

If an immature design is shipped into the field, the higher costs are absorbed by the support organizations. While an experienced team will certainly make mistakes, those errors will be fewer in number and more quickly addressed. An inexperienced team will take much longer to even realize there is a problem, and therefore will take much longer to resolve it.

Watch your speed
Accelerating the design phase also reduces the chance that changes will be requested. While no program is immune to new customer input, new marketing direction or a change in corporate strategy, a program that takes longer to be completed will inherently face more demands to change. Changes, especially late in a program, dramatically affect the schedule and increase development cost.

Given that most semiconductor and other equipment development programs last for a year or longer, it is almost certain that either the market requirements or the competitive positioning of the product will change. Such changes compound the schedule problem, delaying it if they have not been incorporated into the schedule in advance.

To meet customer requirements and revenue targets, engineering management is under tremendous pressure to take on new programs. But since the flexibility of the development organization is crippled when operating at overcapacity, this can be extremely detrimental to engineering performance: Any change in priorities is exponentially more difficult to accommodate when the organization is overloaded (see figure below). Consequently, it is essential to clear the pipeline before adding in new assignments. Once engineering loses credibility by missing scheduled milestones, it is extremely difficult to get it back.

Companies often build in many process checks to catch poor designs before they move too far through the development cycle. The phase-gate new-product development process is one such system common in the semiconductor equipment industry.

Checks and balances
While well intentioned, intensive checks can dramatically slow the design team. The development system can grind to a halt if the support organizations become gatekeepers vs. enablers of the design: It is always easy to find reasons the design might not work or needs more testing.

An inexperienced development organization can be easily misdirected by such criticism. If an immature design is released into the support organizations, the development team's credibility is damaged. The support groups will insist on more process checks if they don't believe the design will meet their requirements.

An experienced design team has the advantage of implicit trust with the constituencies it must satisfy, however. With such a team, these checks can be minimized and primarily used to inform the other groups about what will be needed to move the design along quickly. The support groups should be able to focus on what they need to do to move the design into production, confident that the development organization will deliver a mature design.

Creating a design process that minimizes handoffs will reduce development cost. A handoff occurs when one specialist passes the design to another, adding only that person's part of the design. This typically happens when a systems designer passes the design to a detail designer, who then passes it to document control, who then hands it off to purchasing, who then hands it off to manufacturing.

Even if all those exchanges are crisply executed, it takes much longer than having a senior engineer handle it all the way through. Most organizations believe this is a poor use of the senior engineer's time, but the reality is that every handoff slows things down and introduces errors into the process.

Close the loop
The testing function handoff is another area to watch in the design process. While separate test organizations should have a strong role, they should not be the first line of testing for the design.

Some organizations let the design engineer hand off design verification testing to manufacturing or applications personnel. While this well-intended strategy assumes that the design engineer should design and manufacturing or application personnel are more knowledgeable about how the design will be used, it can dramatically slow the development effort.

User testing is not the same as design-verification testing. The result is that the feedback loop is delayed and critical weaknesses may not be discovered until very late.

Engineers should be encouraged to follow their design all the way through the process. This includes sourcing the parts, interfacing with the suppliers, inspecting the parts when received, supervising the assembly process and testing the result.

Too often, organizations want to insulate the suppliers from the design engineer, afraid that the engineer will compromise the organization. Such isolation, however, reduces the effectiveness of the design team by limiting its understanding of the manufacturing/supply chain issues.

While the design team members do not need to deal with every supplier, they should be able to interact closely with those that are important to the design effort. If you want the design team to have the ability to go fast, give it the option to take the design from beginning to end in those cases where time is critical.

The steps described above are simple to execute and will dramatically reduce design costs and improve results. Following this process will not only reduce program development costs but also improve your design team's morale and employee retention rate.