If a power supply bill-of-materials (BOM) item is not available, then the power supply cannot be delivered to the system and no power means no system for anything that is powered, which is everything in the world of electronics. As obvious as this sounds, it is important to emphasize and internalize these high-level points because it trickles into every aspect of the successful delivery of power solutions (and therefore systems), technical or otherwise. To make the extreme point, if a particular 0402 resistor is not available (because of special packaging, tight tolerances, higher quality, or just plain long lead times), then it does matter if it costs $0.001 or how small and trivial it seems if it prevents you from building and shipping a $100,000 system in a qualified design. The emphasis on a qualified design is made here because shortcuts (i.e. – aftermarket components, unqualified seemingly equivalent components, improperly multisourced components, etc.) can be taken in desperation, but rarely work out for the better…no field recall and long-term brand damage is worth the short-term relief and revenues.
Especially for
power supplies and powered solutions, AOS will typically come from a well-vetted vendor (or equivalent, internal resources in the case of larger enterprises). There are far too many technical, logistical, manufacturing, and business dependencies to thoroughly cover at a low-level in this whitepaper. But high-level categories, identification of key stakeholders, and methodologies for how to vet such resources can get one well down the path, particularly at the appropriate level of invasiveness a design engineer will directly have (as opposed to the management of delegated tasks that fall directly to other stakeholder groups such as Commodity Managers, Component Engineering, and Project Management).
It is very important to internalize the point that the relationship between the engineer and power resource is a two-way relationship. Power solutions can be incredibly complex, have BOM component counts that rival the entire systems they are integrated into, can fail certifications with the slightest discrepancy from an established process (i.e. – fail
electromagnetic compatibility, or EMC, targets because a power transformer wire crosses over another wire instead of under) and are the most critical component for allowing your system to turn on and operate properly. Pricing can also be just as prohibitive to your AOS as we will see in a moment in the subsequent discussion on multisourcing. The customer is not always right if a different customer is willing to pay more for your resource!
A more controversial strategy for ensuring AOS in recent years is the push for multisourcing. The theory is that by procuring solutions/ components that are form/fit/function equivalents from competing sources, this enhances AOS by mitigating the risk of depending on a single supplier (a.k.a. – sole-sourcing). The exception to this is when there is only a sole source of a component, such as with a specialized, application-specific integrated circuit (ASIC).
Just take a moment to consider all that is involved in designing and qualifying a power solution, both standalone and in the end-application or system. From a design standpoint, does your design require considerable tolerance stack-up analyses? What of the vigorous accelerated life, stress, and long-term qualification testing done on assemblies and systems? How many cycles did Engineering need to put into deciding on a single component and getting it entered onto the company’s approved vendor list (AVL)? Now take just this handful of considerations (let alone the many left out) and consider how these will all be accomplished when there is more than a single option for each BOM component!
Even in a relatively low-count BOM and only dual-sourced parts, one can quickly see how the permutations of plausible manufacturing scenarios can quickly get out of hand. Even if an organization has the time/resources to perform virtual assessment (e.g. – Monte Carlo analysis [1]), the number of combinations must typically be constrained to perform the assessment in a reasonable amount of time, hopefully also with a reasonable degree of confidence. To address this conundrum, a design engineer (typically in conjunction with Component/Reliability Engineering) might identify what they consider “critical” components as candidates for multisourcing to try and keep the validation work reasonable for a realistic number of permutations. But then this begs the question of “What is a critical component?” so this strategy is likely to end up being a far more subjective than anyone will prefer.
From a financial standpoint, Supply Chain Management groups and stakeholders tend to be big fans of multisourcing because it can greatly increase the leverage a customer has to drive favorable pricing (including periodic price reduction over time). In particular, this is typically accomplished via what is known as a “share split” or established ratio of what volume of a specific component will be procured from VENDOR A vs. VENDOR B (or A through M), and that ratio can be quite disproportionate (think more 70/30 or 80/20 instead of 50/50). Vendors may also spend quite a bit of money and resources to qualify a power solution so this can sometimes leave them feeling scorn if splits are not equitable so when the vendor with the 80 % share has an issue and cannot deliver, the leverage has suddenly flipped and the vendor with the 20 % share can decide to your savior, charge a major premium, or simply leave you hanging to twist in the wind with no solution.
Hopefully, it is now clear why multisourcing can be a somewhat controversial topic. Do you hear a clear argument for whether multisourcing helps or hurts AOS? Maybe both?!? We did not even broach the topic of how multisourced components are evaluated and considered equivalent to a primary source, which is yet another risky and costly process, potentially also at the detriment of AOS. Or even worse, what happens when your Component Sourcing department wants to cut-in a new component source to a shipping design with many units already in the field?
Now to focus on the key part, which is how to assess a vendor’s/resource’s abilities to accomplish AOS.
