Find a downloadable version of this story in pdf format at the end of the story.Isolated brick converters are widely used in telecommunication systems to provide operating power to network equipment. These bricks come in a variety of standard sizes and input and output voltage ranges. Most can operate over large input voltage ranges of 2:1 or even as high as 4:1. Their modularity, power density, reliability and versatility has simplified and to some extent commoditized the isolated power supply market.
As these brick converters are of a strictly defined size, designers are forever coming up with innovative ideas to increase their output power (and power density). Although these ideas are numerous and varied, they are all related to system efficiency. Consider an eighth brick converter as an example — although there are numerous input and output voltage configurations, topologies and output range tolerances (regulated, semi-regulated, unregulated), they all have very similar maximum power loss numbers at full power (i.e. between 12-14 W). This is a physical limit based on the fixed volume of the converter and the method of heat extraction. Thus, for an eighth brick converter that is 90% efficient (η = 0.9) at full load, the maximum output power (assuming 14 W loss) is shown in Fig. 1 and represented by the equation:
If the efficiency can be improved by just 2%, the output power is increased to 160 W — 28% more output power (Fig. 2)!
As shown in a previous Shoot-Out article , it is possible to reduce the power loss in the magnetic components (up to a point) by increasing the operating frequency. However, this is not normally done because the increase in frequency-dependent semiconductor losses outweighs the potential improvement. On the contrary, the operating frequency is typically reduced to the point where the magnetic structure size is maximized within the overall brick size constraints.
In previous Shoot-Out articles, eGaN FETs and silicon MOSFETs were compared using devices in identical circuits and using power MOSFETs with similar RDS(ON) values as the eGaN FETs. This simple “apples-to-apples” comparison makes evaluating results straightforward and conclusions can readily be drawn from the resultant data.
When it comes to isolated brick converters, however, this simple yet effective approach breaks down. Even when limiting our comparison to regulated 12 V output, eighth-brick converters only, there are still a significant number of variations between commercial designs. Over time, advances in devices, materials, construction and other innovations have allowed greater output power. The resultant structure, layout and topology changes with power level and even between manufacturers for the same power level as what is considered optimum is interpreted differently. Determining the best solution is an iterative process. Even so, the efficiency achieved in a specific brick converter — as good as it may be — can easily be improved simply by allowing the converter to increase in size. This is clearly shown in Fig. 3 by comparing eighth-brick and quarter brick efficiency for the same generation products.
An eGaN FET based converter was developed that is not necessarily an optimal solution. The design goal was to deliberately push the operating frequency much higher than current commercial systems to show that eGaN devices could enable someone skilled in power supply design to develop state-of-the-art next-generation products.
For the 48 V to 12 V eGaN FET based eighth brick converter, a phase-shifted full-bridge (PSFB) converter with a full-bridge synchronous rectifier (FBSR) topology was chosen as shown in Fig. 4. A more complete schematic is shown in Fig. 5. The aim was to show that, due to their relatively small device size, a significant number of eGaN FETs can be used within the restrictive eighth brick size limitations. The choice of transformer turns ratio (6:3) meant that, at 75 Vin, the secondary side winding voltage would be 38 V (too close for 40 V devices) and therefore 100 V devices were used on both the primary the secondary sides. The actual prototype is shown in Fig. 6 and compared side-by-side to a similar  silicon-based converter.
EIGHTH BRICK SHOOT-OUT
The eGaN FET-based prototype eighth brick converter can be compared against a similar 48 V to 12 V fully regulated production converter shown in Fig. 6. Efficiency and power loss results are shown in Fig. 7 and Fig. 8, respectively. Despite the eGaN FET prototype operating at 33% higher frequency, it is able to produce 15% more output power for the same power loss. Also of note is the full-bridge synchronous rectifier, using 100 V eGaN devices, could be changed to a center-tap with two devices in parallel (similar to the MOSFET based design). Instead of the output inductor current flowing through two devices in series, it would then flow through two devices in parallel. This would reduce the secondary-side device conduction losses by 75% (1.3 W, or roughly 10% of total power losses) at 14 A output current.
The same eGaN FET-based prototype was operated at 500 kHz with the efficiency results shown in Fig. 9. It shows that even at twice the switching frequency of the similar MOSFET-based converter it still has equal or better performance at 36 V IN and even at 60 VIN, the full load efficiency is still within half a percentage point of the production eighth brick converter. Although 500 kHz is not an optimum operating frequency for this converter, it emphasizes the reduction in switching losses gained by using eGaN FETs.
NEXT GENERATION EIGHTH BRICK
A recently released  next-generation MOSFET-based eighth brick converter has an output power increase of 67%, to 240 W, with a peak efficiency two percentage points higher than the converter used in our comparison with eGaN FETs. This impressive performance was achieved through multiple improvements (Fig. 10 shows a visual comparison between these two converters.) Some key changes are:
- Switching frequency wasreduced by 30%, to 180 kHz. Core cross-sectional area for both the transformer and output inductor were increased to accommodate the lower frequency.
- SSecondary side center-tap synchronous rectifier MOSFET device voltage was reduced to 60 V from 100 V. These new MOSFET devices have about half the COSS × RDS(ON) product than the previous generation devices and 25% lower RDS(ON).
- SPrimary side topology was changed to full bridge (FB) from half-bridge (HB).
- SPrimary side MOSFET devices were doubled in number (for FB). Also, a smaller die size was chosen to have ½ the COSS and about 1/3 the QGD of the current eighth brick converter devices, but double the RDS(ON).
- STo accommodate these 60 V devices, it is calculated that the transformer turns ratio was changed from 4:3:3 (HB:CT) to 9:3:3 (FB:CT). This requires a 100%+ duty cycle at 36 VIN to maintain regulation and, at 75 VIN, the secondary winding voltage is 50 V.
- SThe use of a digital controller reduced the required board area for control, but also enabled nearly a 100% duty cycle.
Considering 2, 5 and 6 above, there were significant advantages from going to a lower RDS(ON) secondary side devices , whereas 1 and 4 improved efficiency by reducing primary side switching losses.
To see what eGaN FETs can offer to further improve this benchmark performance, consider the comparison in Table 1. To make direct comparison possible, the equivalent eGaN FETs have been scaled to match the R DS(ON) of the MOSFETs.
Using eGaN FETs for the primary side devices, a 60% lower QOSS losses and a staggering 93% reduction in switching figure of merit (FOM) (RDS(ON) × QGD) can be achieved. The actual switching loss improvement is dependent on gate drive strength and layout. To put this in perspective, the changes in QOSS and QGD between the next-generation eighth brick and the previous version is devices are 45% and 42%, respectively. The eGaN FET reduces these numbers further by 60% and 90%, without having to increase RDS(ON). Although the improvement is QOSS doesn't offer much of an efficiency improvement, it is estimated that the reduction in switching time can reduce the primary side switching losses by as much as 2.3 W at full load. The equivalent eGaN FET has 40% lower RDS(ON) while offering 100 V capability.
The sidebar, “EPC Moves Ahead,” discusses the company's work on medium and high-voltage devices and ongoing efforts to improve the on-resistance of its eGaN FETs.
Johan Strydom, “eGaNTM FET-Silicon PowerShoot-Out Part 3: Power over Ethernet,” Power Electronics Technology, March 2010, http://powerelectronics.com/power_semiconductors/egan-fet-viable-efficient-201103/
Ericsson BMR453 series 48 V to 12 V quarter brick converter, Ericsson website, http://www.ericsson.com/ourportfolio/products/bmr453-series-quarter-brick
Ericsson BMR454 series 48 V to 12 V eighth brick converter, Ericsson website, http://www.ericsson.com/ourportfolio//products/bmr454-series-eighth-brick-intermediate-bus-converter
Ericsson PKB4000-C series 48 V to 12 V eighth brick converter, Ericsson website, http://www.ericsson.com/ourportfolio//products/pkb-c-series-eighth-brick
Picture of Ericsson BMR454 eighth brick converter taken from press release http://www1.ericsson.com/solutions/news/powermodules/2009/index.shtml
EPC MOVES AHEAD
FROM ITS START IN 2007, Efficient Power Conversion, EPC, has come a long way toward boosting the image and applications for its eGaNTM FETs. These devices are produced at Episil in Taiwan, a commercial wafer foundry also used to produce CMOS, BCDMOS, and bipolar ICs side-by-side with EPC's eGaN wafers. EPC's wafers use standard well-known CMOS processes, which provides the advantage of producing eGaN FETs at costs competitive with current MOSFETs. This “cost competitiveness” separates eGaN from other alternative materials.
Today, EPC's eGaN FETs cover the range from 40V to 200V. It will be launching 600V products in 2011. Depending on customer interest, it may launch 900V and 1200V products in 2012.
The eGaN FETs have extremely low on-resistances. Today, EPC's eGaN FETs have an RDS(ON) as low as 4 mΩ for 40V devices; by scaling the size of the device, much lower values can be achieved. EPC already has benchmarked RDS(ON) with the EPC2010 200V, 25 mΩ FET. This device has an area less than 6 mm2.
Because of the wide band gap of the gallium nitride crystal, active eGaN devices can be made to operate beyond 300°C. Special devices for high temperature operation are now in development at EPC. The present devices are rated at 125°C or 150°C, because they mount directly to a PCB with a typical temperature limitation of about 100°C.
Monolithic ICs consisting of GaN transistors integrated with GaN FETs experience no degradation in FET performance. In contrast, monolithic ICs consisting of integrated silicon transistors and MOSFETs experience degraded MOSFET performance. This has forced manufacturers to use multi-chip modules to achieve optimum performance from circuits consisting of silicon MOSFETs and transistors.
EPC does not produce eGaN subsystems, but is designing certain eGaN FET subsystem reference designs that can be used by its customers to get products to market sooner and with less engineering effort.
EPC and Microsemi have partnered together to market eGaN FETs to high reliability, defense, and space applications. EPC's eGaN FETs have demonstrated extraordinary capability to operate in high-radiation environments such as those experienced by commercial, military, and research satellites. With EPC as a partner, Microsemi is expecting to release a full range of mil spec products in the coming year.
The eGaN FETs are also capable of meeting the AEC-Q100 automotive specs but the testing and infrastructure are not yet in place. The company expects the first appearance of eGaN FETs in automotive applications in three to five years, starting with entertainment, navigation, comfort, and convenience functions and later spreading to safety and drive train applications.
EPC is planning to integrate various protection and driver function monolithically with its eGaN power FETs. Expect to see those in 2012.
The company now has two application-engineering groups in California and an EPC device characterization group also in California. Its reliability testing and reliability R&D is done in outside labs in both the US and Taiwan, and it periodically uses various industry experts as consultants and contractors for product characterization, and applications development.
EPC now has a pool of highly talented and experienced engineers and continues to hire top technical people in the field of power management. In particular it is looking for device engineers, material scientists, and applications engineers with advanced degrees.
Download the story in pdf format here.