Expanding the idea of design-in support, it has become evident in the areas of power and high-performance analog ICs that form the core of Linear Technology’s products that a skills gap is emerging. It is easy to look at a system with huge FPGAs and think about the challenges in digital terms while overlooking the difficulty of powering the parts. Sub one-volt cores, consuming 10-20Amps require good transient response of <50mV over all conditions making loop response, layout, voltage reference accuracy and load monitoring critical. Add in higher PCB densities, increased clock speeds, concerns over conducted emissions and operating temperature considerations and these quickly become analog and RF problems. The number of designers specializing in these areas has diminished while the need for solutions is on the increase, hence the skills gap. A good supplier will address this with applications support, evaluation boards, circuit analysis models and with products.
An example is LTC’s range of μModule voltage regulators. Resembling a surface mount IC, each product includes system-in-a-package solution that aims to simplify design and minimizes external components. Internally, the layout and design are optimized for electrical and thermal efficiency. Built to the industry’s highest standards, these modules offer reliability approaching that of standard ICs.
Another factor in the selection of components at the design stage is that of cost. First, it is important to distinguish between component price and solution cost. Using the example of the μModule voltage regulator above, the solution cost equates to the unit price, whereas looking only at the DC-DC controller and MOSFET unit costs would not take into account the passive components, magnetics, design time and technical challenges associated with developing a high-efficiency switching regulator. Again the skills gap may play into this calculation. Furthermore, the lifetime cost for a given solution will be heavily dependent on the long-term availability of the component. If a redesign is required, the costs could easily dominate the lifetime cost as the example below shows. Component 1 has to be redesigned twice whereas component 2 remains available for the lifetime of the program.
Clearly, even a reduced unit price with each redesign is insufficient to offset the redesign costs, which may be far more substantial than those indicated here in reality.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.