By Andrea Polti, Global Product Manager – Magnetics
Great strides have been made in recent years in improving the efficiency of power converters using the latest semiconductor technology, with broadband devices in single-phase resonant converter topologies now allowing efficiency figures in excess of 99%. With a stable reduction of static and dynamic losses in semiconductors, attention is increasingly focused on the dissipation of passive components – especially magnetic ones.
One of the desired benefits of the efficiency of large converters is a dissipation and the packaging generally smaller and lower cost. However, to match this, magnetics in the form of energy storage inductors, filters, and transformers must also decrease, and this is facilitated by increasing the switching frequency. Filters and storage inductors that pass net DC typically require less induction as switching frequency increases, which may allow smaller cores and/or fewer turns for the same flux density. This causes little or no increase in total magnetic component losses if the AC component of the current is small. For transformers, core size can also decrease with increasing frequency for the same number of turns and flux density. However, being all AC transformer current, the eddy current and core hysteresis losses also increase substantially as frequency increases. Also, dynamic semiconductor losses increase with frequency, so there is always something to trade off between system frequency, efficiency, temperature rise, and size.
However, transformers can be very efficient and losses at medium and low powers and low frequencies are often ruled out. At higher powers, however, even fractions of a percent inefficiency can cause significant power loss, with correspondingly high average temperatures and Hotspot of the transformers. This can be problematic, particularly if the advantage of the small size of the magnets has been achieved by increasing the frequency, giving up a small overall transformer surface area for heat dissipation to the surroundings. High temperatures can damage insulation and safety or, at best, force the use of unnecessarily high temperature ratings for wire insulation and coils to gain agency safety certification. The ohmic resistance of the copper turns also increases with temperature, which causes even more losses and, in turn, at even higher temperatures.
One approach to minimizing temperature rise in transformers is to provide controlled paths for heat to move away. Ferrite cores used at high frequency have relatively poor thermal conductivity, typically 2 to 5W/m K, compared to 400W/m K for copper, so temperature differentials across the ferrite can be high and thus thermally insulate the interior of a transformer effectively. Thick turns with multiple turns can be used as are typical in a "planar" construction to carry heat away, but the approach is not effective for inner turns which can often be high voltage primaries with relatively high amounts of more wire turns. fine.
A new approach reduces internal temperatures of transformers
Murata has recently made progress in turning arrangements for high-power, high-frequency transformers with its patented PDQP technology, which interleaves coils in a novel way to minimize leakage inductance and the effects of skin and proximity. The PDQP technique offers useful loss reduction, but the company has now patented a new technique to offer better control of transformers' internal temperatures by embedding heatsink plates in the core and turn structure. This method is suited to high power transformers where temperature rises can be high and the core is typically assembled with “U” or “U” and “I” core combinations. The Figure 1 shows the general approach. In this example, eight U7 – U8 cores form the assembly with interspersed metal heatsink plates in blue and red.
La Figure 2 shows the internal construction, in this case using 12 “U” cores but the top six have been removed for clarity. The thicker center plate acts as a conduit for heat and can be attached or glued to an external housing or “cold wall” to provide heat dissipation for the interior of the assembly. The thinner plates in red can be glued to the center plate, or can stick out of the assembly to attach to the external heatsink. The entire assembly can be glued or clamped, but pressure and small clearances are not critical, except for the faces between the top and rear U-cores. All other interfaces are not in the path of the magnetic field and small holes are not material, although thermal matching is better with closer contact.
Similar to the steel laminations of 50/60Hz transformers, the thin plates in red do not form complete conducting loops and the F currents would still be induced through the thin dimension of the plate in the first place. The eddy current is proportional to the area of the induced current loop and the power loss is proportional to the square of the current, so both are minimized by the thin plates. The thickest central plate theoretically has no eddy currents if it is symmetric to the turns, since the magnetic field of each core opening cancels out. The material of the plates can be copper for the highest performance or aluminum by a factor of two, but their electrical conductivity is higher by a similar ratio, so any residual eddy current would produce lower losses in aluminum.
To confirm the performance of this approach to transformer dissipation, a 50kW 24kHz converter with the embedded boards was simulated and compared to a version without the boards. A real transformer was then assembled and loaded and temperature measurements were taken. The converter is typical of an EV battery charger with a 700VDC input bus and 417V output at 122A. The Figure 3 (left) is a simulated temperature map of the transformer with the heatsink plates included (external view) while the image on the right shows a cross section of the part with the heatsinks. hotspots internal. The ambient temperature stood at 31°C and a maximum internal temperature of 56,2°C is indicated, an increase of about 25°C.
The same transformer was simulated without the additional dissipation and the Figure 4 (left) and (right) are the two equivalent drawings, showing a maximum internal temperature rise of 39°C, more than 50% higher than with the heatsink plates. It should be noted that the temperature scaling is different between Figures 3 and 4.
Practical measurements confirm the simulation (Figure 5), with embedded thermocouples registering a Hotspot maximum internal temperature slightly above 58°C, within 1,5°C of the simulation.
Murata's proprietary transformer dissipation approach, as described, promises to enable higher powers of a transformer assembly or lower temperatures for the same power with a corresponding increase in reliability and lifetime. The margin of safety to material temperature limits is improved and agency certification is facilitated without unnecessary resort to specialized and expensive high temperature insulation systems. The combination of simulation and practical measurements confirms the value of the approach.