The evolution of magnetics in power electronics
Esoteric theories of relativity and field manifestations of fundamental particles give an insight into deeper connections between magnetism and electrostatics, but a satisfactory intuitive explanation of magnetics remains elusive. Despite this, practical magnetic components are ubiquitous and have evolved in their construction, albeit at a slower rate than other electronic components. Today, the emphasis is on miniaturisation and power density to keep pace with the newest technologies such as the Internet of Things (IoT). The old standard of round wire on plastic bobbins is giving way to low-profile planar construction and integration of magnetic components at substrate and even chip level. Read More
James Clerk Maxwell built on the work of Gauss, Faraday and Ampere to generate the mathematics that describes electromagnetism, and when valves and transistors appeared on the scene, magnetics became core components in power conversion. To this day, little has changed in the method of use of these components, and engineers still joke about the ‘black magic’ of inductor and transformer design – perhaps as a reason not to have to delve into the arcane world of ‘flux’ and ‘magnetising forces’.
In essence, magnetic component theory has changed little over the decades, and we still use 19th-century maths in routine calculations for the most leading-edge designs. The science is seen as quite fundamental, with little hope of radical change without shaking the fundamental laws of physics. This leaves magnetics as a lagging technology, a major cost in materials and labour in power conversion applications, and a major barrier to miniaturisation. Construction techniques and the materials used for cores have however advanced significantly, following extreme pressure to implement more and more compact power conversion functions in applications such as mobile phones, IoT and ‘wearables’.
Round wires have given way to square and flat formats and on to foil and film windings. Core materials have improved, not in their fundamental operation, but in their losses when subject to high frequencies, allowing switched-mode power conversion to operate in the MHz region. High-frequency operation is key to lower cost and smaller size, because 19th century mathematics tells us that winding turns – and hence size and consequent copper loss – and weight scale down with frequency, all other things being equal.
As small-size and low-profile formats have evolved, the construction of magnetic components has progressed from through-hole bobbin types to surface-mount, on to planar magnetics integrated into circuit boards, and towards the goal of chip-scale inductors and transformers embedded in silicon. Through this evolution the cost has decreased, particularly the labour element. Bobbin-wound transformers using magnet wire can be assembled with a degree of automation, but there is always some handling overhead and also variability of the final component’s performance. Even toroid winding can be automated, with core sizes down to a few millimetres in diameter with wires of perhaps 0.05 mm, but the electrical termination of such assemblies is even more variable.
With shrinking sizes, transformer designers come up against another fundamental force of nature; the international safety agencies. The isolation needed, often rated at thousands of volts, must be withstood by the insulating materials between windings. Creepage and clearance distances also set a practical limit when the transformer is providing a safety barrier against lethal voltages. Exotic wires have been developed which provide sufficient voltage isolation to satisfy the agencies, typically using multiple layers of insulation so that defects and pin-holes are unlikely to line up and cause breakdown. Otherwise, substantial solid insulation or large isolating gaps must be used, reducing coupling between windings and degrading performance.
Losses in magnetic components fall into several categories: there is the ‘skin effect’ in the wire where at high frequencies the current ‘crowds’ on to the surface of the conductor, leaving it with effectively higher resistance; there are proximity and eddy current effects where the magnetic field from windings couples into the structure of the component, causing unwanted circulating currents with their consequent ohmic losses; and there are core losses. Figure 1 shows how core losses vary in different core material types due to the work done in cyclically aligning magnetic domains in the material in alternate directions. Note that for a particular material and frequency, the loss is expressed as watts per unit core volume. This has the welcome effect that smaller cores naturally dissipate less under given conditions. Figure 2 shows the power attainable with some exotic materials such as ferrite polymer and granular film, which promise potential operation with reasonable losses at 100 MHz and beyond. The throughput power at these frequencies is limited, however.
Planar transformers, with their low profile and windings typically formed with circuit board tracks or metal stampings, are the choice for commercial products at high power. Currently available standard core materials enable operation up to around 30 MHz, and this construction, although complex and relatively expensive in materials, does give the benefit of very repeatable performance, which can be proven in simulation. The flat construction with wide faces also allows efficient heat removal. Figure 3 shows a comparison between bobbin and planar transformer heat spreading under similar conditions. Downsides to planar construction are the relative difficulty in achieving large creepage and clearance distances for safety isolation, and parasitic capacitances between the wide flat windings can be high. Implementing screens between windings is easy, however, so EMI effects can be simulated and controlled.
Planar magnetics typically use discrete flat ‘E’ cores and a stack of PCB windings, either as a separate assembly or sometimes as part of a motherboard. A natural next step is to integrate the magnetic material itself into the layers of the winding. Various techniques are in evidence, including moulding ferrite around wire windings to maximise core volume and power handling such as in the Murata FDSD inductor series; embedding film magnetic material in PCBs; and embedding discrete cores inside the PCB layers and forming windings with tracks and microvias. With careful design and close process control, embedded transformers in substrates can have very high agency-rated voltage isolation. At the chip scale, die-inductors are incorporated on to silicon substrates and sit alongside wire-bonded chips. The ‘Micro-DC-DC’ non-isolated DC-DC converter takes this a step further by actually using the magnetic core as the substrate for placement of the control electronics. This gives very small size and excellent EMI characteristics because of the short electrical path lengths.
Chip-scale magnetics is an enabler for IoT and other technologies such as energy harvesting, but at high power there is also pressure to realise the promise of wide band-gap switching devices such as silicon carbide (SiC) and gallium nitride (GaN) MOSFETs. The switching speed of these components can be hundreds of MHz, with low losses at high power and high temperature, but magnetics are struggling to provide a match, at least with low losses. Practical designs have been done with SiC, for example at more than 100 kW at relatively low frequencies to keep the ferrite material cost reasonable and with low core losses. The currents that go along with these high powers certainly point the magnetics designs towards planar types with, typically, stamped copper windings and with the assembly actively cooled with water plates or fans. Planar magnetics at high power and high frequency also allow close control of parasitic characteristics such as leakage inductance. This can be vital in conversion topologies at high power and high frequency that use leakage inductance as part of a resonant network to obtain low-loss switching.
Evolution rather than revolution is the watchword for magnetics components, but they do put a little magic into the world of power conversion.
Disclaimer: by clicking on this button, you accept that your data might be communicated to this company. If you do not want us to communicate your data, please update your details on your profileDownload White Paper