Power supplies have limited the minimum size that electronic systems can attain, relying as they do on large transformers with large ferrite cores and magnet wire windings. By their very design, planar transformers ease this limitation and allow designers to achieve the low profiles required for pc-board mounting in space-constrained applications. In addition, their construction endows them with more unit-to-unit repeatability, high power-density capability, higher-frequency operation, and high efficiency. While the disadvantages are few, it is important to understand the device's basic construction to fully appreciate its capabilities and potential drawbacks.
Planar transformers are so compact because they are made from copper leadframes and flat, continuous copper spirals instead of copper magnet wire wound around conventional ferrite cores. The spirals are etched on thin sheets of dielectric material and stacked on flat, high-frequency ferrite cores to form the transformer's magnetic circuit (Fig. 1). Next, the core material is bonded with a low-grain-diameter epoxy to minimize core losses. High-temperature (130°C) insulators, such as Kapton, within the stack of spirals ensure high isolation between windings. While the leadframes are used to mechanically secure the transformer in place (because it is of low mass), for high-shock/vibration applications, the flat ferrite cores can be bonded to the pc-board using double-sided polyester tape. Connections to the outside circuit, such as the power semiconductors, are made by standard pc-board pins.
It is this construction that gives the planar transformer its characteristically low profile, which usually ranges from between 0.325 to 0.750 in. This makes them especially attractive to power-supply manufacturers working within tight space restrictions.
The planar transformer's pc-board construction means that once the circuit-board components are designed and stamped for a planar device, the windings of subsequent transformers in a production run will be spaced exactly the same distance from each other (Fig. 2). This design allows planar transformers to be manufactured with automated assembly equipment, greatly improving device unit-to-unit repeatability and yield in production runs with tight specifications. Conventional transformers are manufactured with copper wire wound around ferrite cores. Irregularities in the spacing of the windings, along with the vagaries of manual assembly, can contrive to produce wide variations in device performance.
The uniformity and predictability of planar transformers has the added advantage of making them simpler to model than conventional transformers. This especially applies when using computer-aided-engineering (CAE) tools such as SPICE modeling.
With excess weight an on-going problem in the typical power-supply design, the planar's ability to reach weights as low as 0.6 oz. per 100 W has made it a key component in many lightweight designs.
While the planar transformer has many advantages, its development was initially hampered by the need for custom cores, pc-board windings, and isolators. However, this attitude is changing as the devices gain acceptance.
Planar transformers offer efficient operation at high switching frequencies, typically reaching 97% efficiency at switching frequencies through 500 kHz. Their maximum operating frequency can reach as high as 1 MHz (with low flux density). The flat windings are the key to their high efficiency and high-frequency operation. The windings also greatly improve the device's power-density capabilities.
Because conventional transformers generally rely on round-wire windings around a ferrite core, the copper conductor is not efficiently used. This is due to a phenomenon known as skin effect. Skin effect occurs when induced currents and magnetic fields cause current in a round conductor to concentrate near the thin outer surface—or skin—of the wire, especially at higher frequencies. As a result, the total current-carrying area is less than the full wire area, making the ac resistance greater than the dc resistance by an amount determined by the skin thickness.
In a planar transformer, however, the "windings" are actually flat conductive traces formed on copper-clad pc-boards. As a result, the current tends to concentrate toward the outer edges, but it still flows through the entire conductor, with improved overall current density compared to a cylindrical (wire) conductor. The end result is that a planar transformer, with its flat windings, can achieve higher efficiency in much smaller sizes than conventional wire-wound transformers.
Planar construction also minimizes parasitic reactances, such as interwinding capacitance and leakage inductance (typically under 0.5%). The low leakage inductance is achieved by splitting, which puts part of the primary winding in one place, such as the top, part of the winding at the bottom of the stack, and then evenly sandwiching the secondary windings on both sides of the stack.
In wound designs, leakage inductance is difficult to control. The low stray capacitances and leakage inductances go a long way toward reducing high-frequency ringing in a planar transformer's output voltage. This construction—with conductive circuits stacked on dielectric sheets—also allows a planar transformer to achieve good primary-to-secondary and secondary-to-secondary dielectric isolation. The devices can accommodate a wide range of input voltages, and can be specified with one, two, or three outputs. They also meet or exceed the performance requirements of offline converters
Designing With Planars
Because current-carrying capability is a concern in an SMPS, planar transformers typically employ 4-oz. copper-clad circuit boards for their internal winding forms. The same grade of circuit board is a minimum requirement for other sections of an SMPS. The copper used in a 4-oz. copper circuit board is 5.6 mils thick, or 2.8 mils from the center to the surface. In a circular conductor, the current skin depth for copper at 70°C can be obtained from the equation:
where S is the skin depth in centimeters, and f is the operating frequency. This works out to be a little over 5 mils at an operating frequency of 250 kHz. It appears to leave little room for error with 4-oz. copper boards.
However, the planar cross section of 4-oz. copper must be converted to circular mils (as with wire tables) to make for a more meaningful examination of skin depth and current density for a given operating frequency. A circular mil is the area occupied by a circle with the diameter of 0.001 in. Just divide the area in square inches by 0.785 x 10-6.
At higher current densities (and output power levels), 4-oz. copper may not be robust enough. Most circuit-board manufacturers offer heavier copper cladding, usually as a special order. Pc-boards can also be paralleled to double the wire size.