The previous, complementary whitepaper provided a more in-depth analysis into multisourcing, what it is, how it can be interpreted, and some considerations (positive and negative) for performing a proper assessment of one’s resources and making an informed decision. In short, the different considerations really seem to boil down to a focus on mitigating technical risk versus pricing leverage. As a further reminder, the objectives of stakeholders (internal/external, engineering/supply chain the like) are typically in conflict, both in execution as well as the priorities driving them. Here, we shall take that assessment and expand upon it to turn the analysis into more actionable strategies for driving results.
Assuming the final decision has been made to implement some kind of multisourcing strategy, the first step is for Design Engineering to determine the “critical components” list for analysis. The term is in quotes because the characterization of what is critical can be highly subjective and variable. Even trying to simplify with rules like “only the power components” or “only the hottest components” or “only the components most critical to safety” will often lead to grey areas and points of dissent. Regardless, it is important to work closely with teammates (especially those in Component/Reliability Engineering and Supply Chain Management) to negotiate through the tradeoffs of performance, cost, time-to-market (TTM), and AOS.
Component/Reliability Engineering is typically the least concerned with cost, which means they can dictate many qualification/life tests that can be extremely expensive, time consuming, and require 3rd-party resources (each time one takes direct control away, it adds risk to the development and schedule). First negotiating with these stakeholders on the minimum list of critical components, with proper justifications for either accepting or eliminating items on the critical components list, is an excellent origin point. An example of a compromise proposal may be if a component can be assessed with more virtual/statistical methods (such as Monte Carlo analysis [1] or vendor random sampling data) instead of the far more comprehensive, thermal/electrical stress and accelerated life testing.
Any such “real” (e.g. – physical/environmental) tests will not only need to consider the qualification on the individual component, but also the validation in the actual system(s) it is targeted for so it should be fairly apparent how determining the number of devices/units under test (DUT/UUT) can be challenging with the number of printed circuit assembly (PCA) bill-of-material (BOM) combinations that will only grow exponentially with each extra source added to the mix. Some key questions to consider are as follows:
- Has the program manager allocated (and therefore also budgeted) a sufficient number of (prototype) UUTs to accomplish the requisite testing?
- How long will there testing take and at what stage in the development schedule?
- If the schedule is very tight, then is there any contingency planning for when a proto fails highly-accelerated life testing (HALT) or exceeds electromagnetic interference (EMI) class limits?
- Does this align with the power solution release target AND the system release target?
Supply Chain Management personnel are likely to be driven more by opportunities for cost reduction and tend to view multisourcing as a helpful tool for driving AOS. The latter point is debated in the previous whitepaper and will not be repeated here, but please be sure to investigate component availability and lead times on a line-by-line basis. The determination/qualification of a “critical” component was discussed above, but that is still a different discussion from how a second-source component is determined to be considered “equivalent” to a primary source. Again the quotes are not to be cheeky, but to emphasize the point this can be a very relative term and the semantics of which are inexorably tied to everyone’s success. Unfortunately, it is all too common these days for equivalence to be determined by some very fundamental figures of merit (FOM) and cost (typically, with highly inequitable weighting).
While this point is most salient in components that tend to be more sensitive to environmental/application factors (i.e. – semiconductor devices), it can be just as applicable to the simplest of passives (i.e. – resistors). Two different sources for a metal–oxide–semiconductor field-effect transistor (
MOSFET) may have the same package/footprint, drain-to-source (a.k.a. – blocking) voltage, and gate voltage, but can still have drastically different gate charge or input/output capacitances. While this may be irrelevant for a small-signal, switching application, it can make all the difference in the world when used in a power FET application. Two different 1k resistors in the same package style and thermal rating may still have slightly different footprints, material composition, or termination styles. The different sources may still likely be characterized as dual sources under the internal part number in the company’s approved vendor list (AVL), which means either source is considered qualified to be used interchangeably and leads to major performance risks.
It is natural to want to define rules to help simplify complicated developments and implement processes for what may be perceived as optimizing for time/cost, but multisourcing is one of those particularly touchy areas in which being too generic or restrictive can be highly counterintuitive to meeting ultimate project goals. A simple example of this is that low-volume designs should have a totally different approach than high-volume designs. Low-volume designs tend to be more specialized, can have more aggressive/ruggedized specifications, and be less cost sensitive. High-volume designs tend to have more stringent quality requirements, have increased risk from AOS, and have the overall economic viability be tied to tightly-controlled component, manufacturing, and qualification costs.
A last, key topic in this discussion about how to effectively manage supply chain stakeholders in Commodities Management revolves around business continuity planning (BCP) support [2]. BCP is all about contingency planning against major interruptions, such as is the case for disaster recovery and mitigating enterprise resource planning (ERP, [3]) bottlenecks. While these typically involve tasks for those Commodities Management groups and the resources they associate with, it is always good practice for the power design resource/owner to keep abreast of these processes and the planned actions in the case of a disaster, whether it be from an act of God or a severe supply issue (i.e. – regulatory/embargo/customs). A worst-case scenario can be when an entire factory (or even region) is taken out by a natural …
