Critical Success Factors in Defining a Simulation Strategy

SIMPLIS makes the process of virtual prototyping a practical option for the design of switched mode power supplies in a way that has not been available in the past.  While simulation will never completely replace hardware testing and qualification, Virtual Prototyping with SIMPLIS can detect over 50% of design errors that would otherwise go undetected until a power supply design is committed to hardware.  We know of no other design quality improvement technique that can begin to match this capability.

The most productive users of SIMPLIS are using it to verify the performance of their designs before they begin their first PCB or IC layout.  The benefits that they have been seeing have been quite dramatic and, to date, we have not had a single design group achieve this level of execution and then abandon the use of SIMPLIS for another alternative.

So the good news is that SIMPLIS makes virtual prototyping a practical and attractive option for real-life power supply design.  The challenge is that SIMPLIS may be relatively new for your organization.  And the challenge of introducing a new methodology into an organization is not to be taken lightly - especially if you want to see the benefits show up across a broad portion of your organization in a relatively short period of time.  A key success factor, from our experience, is how efficiently the organization can develop a simulation strategy that is appropriate for its market and then communicate that strategy to the design team.

Broadly speaking we are seeing successful organizations choose one of two approaches:

1. Internal Expertise Drives Project Launch (Examples: Delta/Artesyn/Astec)   These organizations often do not start out with an established simulation culture and they decide to have one or two experts develop a simulation strategy, gain significant expertise by trialing this strategy on several real life designs, and then train rest of the team once the strategy has been sufficiently refined.  This takes longer (several years in most cases), but it has the advantage of implicitly developing a simulation support infrastructure that is needed when you are trying to roll out the new methodology across several large design groups.  As a result, there are always some experts available that can help out when people get stuck and who can address modeling efficiency issues like the creation of IC controller models.  There is always significant senior management support of this effort administratively, but often the design managers are not actively leading the technical simulation activities in the early stages.
    
2. External Expertise Drives Project Launch (Examples: Vicor/Power Integrations)   These organizations already have a significant simulation culture, but are not familiar with SIMPLIS.  So, they have a strong vision of how they want a virtual prototyping process to work; they just had no way to get there before.  Typically, they are very familiar with Spice.  These organizations are often much smaller (10-20 engineers) than in approach #1  (100-300 engineers).  They set up a customized training program where they invite outside expertise to help them launch the process of turning their vision into a clearly defined simulation strategy.  Their training sessions have very specific objectives which include articulating in detail how they will quickly create and communicate a simulation strategy for their products.  There typically is significant management participation at the technical level.  These managers are very tuned into the business realities of both the need for simulation and the costs and technical challenges of simulation.   The managers typically are the ones that give the design engineers "permission" to exercise the 80-20 rule and thus avoid increasing the complexity of the simulation modeling effort when the business benefits do not support the additional costs in time or money.  This approach can achieve quite remarkable results in 6 - 8 weeks.
    

In the end, both of these approaches tend to converge.  In the first approach, the first level design managers do need to get involved and begin to own the leadership of the process.  In the second, there does need to be some intentional efforts to create and support a simulation infrastructure so that every engineer does not need to model all their own ICs, for example.

There are two critical dimensions of a successful simulation strategy - a technical dimension and a business dimension.  Essentially, the technical dimension of the simulation strategy describes in detail the simulation objectives of the organization for each unique class of product.  In order to achieve meaningful results for each simulation objective, there must be a clear definition of the desired quantitative measurements and the level of modeling detail required for each of the critical circuit components.  The technical portion of the simulation strategy must be summarized and clearly communicated to all members of the design team in order to achieve consistently reliable results.  Simulation results must be tested against measured product waveforms for each new class of product in order to confirm the technical validity of the strategy.  There is no substitute for simulation expertise in creating a successful simulation strategy.  Training sessions can be a very effective way to quickly build an organization's expertise in SIMPLIS.

The business dimension of a successful simulation strategy is equally as important as the technical dimension.  The business challenge is to figure out how, given your resources, skill sets and market requirements, you can get the most benefit from building SIMPLIS prototypes in the shortest practical time frame.  In all cases, it is important to make a set of choices that result in some significant business success very early on in the process.  Then you build on that success to increase the scope and "coverage" of your simulation efforts.

The key to early success in SIMPLIS prototyping is to choose the right problems to tackle in the beginning while the organization is finding its way in this new world.  Once an organization tastes significant early success, most of the mechanical problems of figuring out how to push the buttons recede to the background.  If the first two projects are executed successfully, people become very skilled at jumping on the bandwagon.  Once they are convinced that their life can be better if they learn how to do this, the average design engineer is pretty good at working through the various simulation details that can otherwise slow things down.

Picking the right simulation objectives for the right project in the beginning is very critical to early organizational adoption of this process.  This is where the early project leadership plays a critical role.  It requires good judgment and an appreciation of the business value and the technical challenges of simulation.

In defining both the technical and the business dimension of your simulation strategy, there is a great temptation to be too ambitious or ask for too much detail too early in your simulation project.  The number one, two and three reasons why organizations get bogged down in their virtual prototyping efforts is committing to climbing too many learning curves too early.  In reality, there is almost always very juicy low hanging simulation fruit that does not require simulation heroics, but rather thoughtful technical business insight.

Virtual Prototyping is now a practical option available to switch mode power supply designers using SIMPLIS.  Early organizational success in virtual prototyping depends on the development of a simulation strategy that is appropriate for your market and the effective communication of that strategy to your design team.  A key part of this strategy is the wise selection of initial project objectives so that significant business benefits may be captured early on without requiring an overwhelming learning curve on the part of design engineers.