Power Stage Integration: Welcome Friend or Elusive Foe?
By Guy Moxey, Fairchild Semiconductor Corp.
Considerable focus has been placed on controller enhancements for low voltage DC/DC synchronous buck designs. Ultra-high switching frequency and fast transient response are two key design considerations that can be achieved through digital or mixed signal control implementation.
Guy Moxey, Fairchild Semiconductor Corp.
By Guy Moxey, Fairchild Semiconductor Corp.
Considerable focus has been placed on controller enhancements for low voltage DC/DC synchronous buck designs. Ultra-high switching frequency and fast transient response are two key design considerations that can be achieved through digital or mixed signal control implementation.
Although this does represent a significant step forward, the complete system design will only be as good as its weakest part and the vast majority of converter loss is still generated through the discrete implementation of the power train.
Optimization of the driver and switching devices within the power stage is paramount, not only at specific thermal design points, but more so across the entire load spectrum. Advancements in discrete MOSFET silicon and innovative advanced packaging technology have enabled low voltage DC/DC switch mode power supplies such as POLs, bricks and VRMs to be pushed the present limit of power density, efficiency and thermal performance – however, this has not be achieved without tradeoffs. Increasing power density is achieved but with a cost of increasing the overall power losses and higher temperatures on the silicon junction, the device case, and the overall PCB. In the same manner, optimizing a DC/DC power supply for medium to peak currents comes at a sacrifice of efficiency in the light loads (such as standby or sleep mode), and visa versa.
So what will the future hold?. We can see in the market that Multi-chip modules MCMs are gathering momentum primarily because discrete solutions do not solve the need for higher power density nor parasitic issues at higher switching frequencies. These MCMs, for example DrMOS, have been the subject of evaluation for quite some time and it can be proven that the performance of these devices is at par with, or better than, existing discrete solutions.
Typical advantages can include low thermal impedance from the use of leadless packaging. Simulated and proven internal wire bonding designs can minimize external PCB routing, thus reducing inductive and resistive PCB parasitics. Let's not forget compatibility to controllers that enable various modes of operation, in particular, discontinuous conduction mode for light load efficiency improvements. Most importantly, there is a flexible “system solution” integration of driver, FETs, diodes and LDOs.
As consumers, we have all enjoyed the significant price decreases in the price of computers, flat panel TVs and DVD players. However, nothing is free and so the machine development teams have to design with ever tightening budgets. System efficiency is the key and integrating – hence optimizing, offers state-of-the art results that are in the region of up to 5% higher than the equivalent discrete designs. These points are taken at industry typical switching frequencies (300KHz) and if the system switching frequency is doubled or even tripled, then the benefits of using an integrated solution are even more pronounced. Higher switching frequencies, comparable efficiencies and a significant reduction in the size and cost of the output filter capacitors and inductor all make a compelling, but somewhat eluding, case.
So we can see that design changes are required and power train integration can offer significant advantages. However, embracing the change means deviating from the tried and tested years of discrete driver and FET implementation. The positive argument is that power semiconductor vendors take the ownership to simulate, study and optimize all the individual elements before implementation and manufacture, so the solution available can only yield the best performance possible for the specific application. However adoption of these advanced integrated solutions does mean a divergence away from the discrete comfort zone and are we, as power design engineers, willing to embrace new thinking and be “power train pioneers’ ?
Guy Moxey is Senior Marketing Director for Computing, Communications and Consumer Products at Fairchild Semiconductor. He holds a BEng and MSc in Power Electronics and has spent 17 years in electrical power engineering. His experience includes electrical machine/drives design, power semiconductors and power IC applications. He is the author of many technical articles and application notes.
Advertisement
