Most electronics features DC-DC conversion in some guise. It might be after rectification in an AC-DC converter, or it might just be to drop 5V to 3.3V with no isolation. There is a considerable amount of support to ‘make your own’ from the textbooks and component suppliers with reference designs and interactive calculator tools. It’s fair to say that it’s not hard to come up with a solution which is ‘fairly good’ for simple, low power requirements and you might convince yourself that the design time, qualification, component procurement and assembly costs are worth it for a simple function.
Increased Power = Increased Complexity
As power level increases, everything gets more difficult, and because of that increase, other parameters start to become critical. Maybe an ‘OK’ efficiency of 75% at 5W load represents little loss, but at 50W, the difference between 75% and a class-leading 95%+ efficiency makes the difference between running embarrassingly hot or cool or needing a heatsink or not. With increased power, other parameters such as EMI start to become an issue, and as current levels increase, good dynamic load-step performance becomes very important to processors and ASICs/FPGAs for example. For these parts, voltage over- and under-shoot with load steps must be tightly controlled. Additionally, higher power DC-DCs are naturally performing more critical tasks where control, monitoring, and protection becomes necessary.
When isolation is needed, and magnetics are involved, the situation gets more difficult still; magnetics rarely work optimally from paper theory or simulation and experience in design counts for more. If safety agency requirements have to be met for isolation, another whole discipline of standards for global compliance has to be called in.
You’re a talented engineer with skills in all these areas, but then the layout guy has to fit your discrete design in the space left. Moreover, procurement has to find that custom magnetic you need, all at low cost with no special assembly techniques – it's quite a big ask with lots of trade-offs and compromises.
Make Versus Buy
We’ve laid out the familiar argument about ‘make versus buy’ and of course, it’s not always clear-cut. Eventual production volume might justify the investment in a discrete design with the promise of lowest possible BOM cost but then time to market is also impacted. What is the value of maybe two months early to the market compared a cent or two of BOM cost saving? What is the financial implication of the risk and uncertainty involved in a discrete design? The decision to make or buy is, therefore, a complex one and is not static. You can do simple analysis showing the effect of design costs on payback against production volume for the different module and discrete part cost (Figure 1), but there are still unquantifiable factors such as the value of quicker time-to-market.
Example payback analysis.
Many control ICs are available for switching, and linear designs that make the designer’s job easier to some extent and new parts with more and improved features appear every day. Their reference designs can be a great starting point for a solution, but again, they have to transfer to the target product and achieve the same electrical EMI and safety performance, if it was ever even considered.
Traditionally, at current levels below 2A, a discrete solution might have been preferred. Above 10A, bought-in modules might be assumed which could bring in other benefits such as SIP mounting to save board space. Between 2A and 10A, it may not have been so clear which would be an optimum solution. The mechanical format of modules, often through-hole at these current levels could seem bulky, hard to fit and relatively inflexible.
μModules Are Another Option
There is another option to consider; the μModule parts from Linear Technology Corporation (LTC), now part of Analog Devices (ADI), are bridging the gap between the discrete component and full module solutions. The parts, dubbed ‘Power System-in-a-Package’, crucially incorporate the magnetic element of the converter in isolated and non-isolated versions while appearing as just another SMT component to place. The µModule introduction addresses that space between low and high power and pushes down the boundary in terms of current where a make-it-yourself solution might seem attractive.
μModule DC/DC regulators replace a complex PC board with active and passive discrete components and a single, simple, tiny, drop-in module. (Source: Linear Technology)
Part of the compelling attraction of the μModules is the clear evidence that the technology utilized gives advantages that just could not be achieved in discrete designs (Figure 3). The proprietary LTC silicon is integrated with magnetics, power MOSFET and passive components in a package down to 6.25 x 6.25mm for a 3A step-down regulator and with heights down to 1.82mm. The ultra-thin packaging allows the parts to be used on the back-sides of PCBs or under heatsinks covering processors for example.
Typical μModule construction (Source: Linear Technology)
There are over 100 different variants in the range, grouped into 15 families. Included are step-up and step-down converters, configurable step up/down, inverting types, battery chargers, LED drivers and digital power management. Parts with isolation are available with 725VDC rating and agency-recognized 2kVAC.