Johan Strydom, David Reusch, Steve Colino, Alana Nakata
Efficient Power Conversion Corporation’s (EPC) hyper-fast enhancement mode Gallium Nitride (GaN) power transistors offer performance improvements well beyond the realm of silicon-based power MOSFETs. Standard power converter topologies can greatly benefit from the added performance and leap to performance not attainable with current MOSFET designs; improving converter efficiency, while maintaining the simplicity of converter designs.
Using eGaN FETs is very similar to using modern power MOSFETs. However, due to the significantly better performance, there are additional design and test considerations to make certain devices are used efficiently and reliably.
In an effort to make the transition from power MOSFETs to the new generation of power management devices as easy as possible, this paper describes the general operation of enhancement mode GaN devices, gate drive techniques, circuit layout considerations, thermal management techniques, and testing considerations.
General Description of eGaN FETs and GaN-on-Silicon Technology
Figure 1 – EPC’s GaN power transistor structure.
A device’s cost effectiveness starts with leveraging existing production infrastructure with a process with few production steps. Depreciated, low resolution CMOS foundries are used in the fabrication of eGaN FETs. EPC’s process begins with silicon wafers upon which a thin layer of Aluminum Nitride (AlN) is grown to isolate the device structure from the substrate. On top of this, a layer of highly resistive Gallium Nitride is grown. This layer provides a foundation on which to build the GaN transistor. Aluminum Gallium Nitride (AlGaN) is applied to the GaN. This layer produces a physical strain, and because GaN is a piezoelectric material, the strain attracts electrons to the interface. This concentration of electrons is called a two-dimensional electron gas (2DEG). Further processing forms a depletion region under the gate, and metal layers are added to connect the three terminals, Gate, Drain, and Source. A cross section of this structure is depicted in Figure 1. This structure is repeated many times to form a power device. The end result is a fundamentally simple, elegant, cost effective solution for power switching. This device behaves similarly to silicon MOSFETs with some exceptions that will be explained in the following sections.
Figure 2 – Resistance vs. breakdown voltage.
To obtain a higher voltage device, the distance between the Drain and Gate is increased. As the resistivity of the 2DEG is very low, the impact on resistance by increasing blocking voltage capability is much lower when compared with silicon. Figure 2 shows the theoretical tradeoff between device on resistance and blocking voltage for GaN, SiC, and Si. After 30 years, silicon MOSFET development has approached its theoretical limits. Progress in silicon has slowed to the point where small gains have significant development cost. GaN is young in its life cycle and as we have seen with EPC’s first four generations of eGaN FETs, progress in this technology is occurring very rapidly.
Figure 3 – transfer characteristics curve for EPC2001
Figure 4 - RDS(on)
for various current for EPC2001.
EPC’s eGaN FETs behave similarly to silicon Power MOSFETs. A positive bias on the gate relative to the source causes a field effect which attracts electrons that complete a bidirectional channel between the drain and the source. When the bias is removed from the gate, the electrons under it are dispersed into the GaN, recreating the depletion region, and once again, giving it the capability to block voltage. Figures 3 and 4 show the transfer characteristics and RDS(on) vs. VGS of the EPC2001 respectively. The transfer characteristics show the current capability as VGS varies. This is very similar to MOSFETs except the transconductance (did/ dvgs) is much higher for eGaN FETs of similar RDS(on). The RDS(on) vs. VGS curve shows that RDS(on) becomes reasonably flat beyond VGS = 4 V.
Drain to Source Maximum Voltage Rating
The maximum drain to source breakdown voltage (BVDSS) is specified in EPC GaN transistor datasheets.
Special considerations have to be taken when switching inductive loads. These types of loads present a possibility of the drain voltage exceeding the maximum rating due to inductive “kickback”. This phenomenon will cause the drain voltage to increase beyond the breakdown and dissipate the energy from the inductor in the device. EPC’s GaN transistors are not rated for avalanche mode operation, but they do have an overshoot rating of 20% over the BVDSS for 10,000 cycles of 5 ms or less in duration. If the devices will be subjected to higher voltages or more repetitions, proper active or passive clamps/snubbers must be used to limit the rise in the VDS to a safe level. Also, proper layout techniques must be used to limit the parasitic inductance in the circuit and hence limit the stray inductive energy present in the system.
eGaN FETs differ from their silicon counterparts because of their significantly faster switching speeds and consequently have different requirements for gate drive, layout, and thermal management which can all be interactive.
Do Not Exceed Gate Drive Maximum Ratings
Figure 5: EPC GaN power transistor gate structure.
As seen in Figure 4, full enhancement of the device channel can be achieved by applying 4 V or greater between the gate and source while the absolute maximum gate-to-source voltage is 6 V. Similar to the Power MOSFET, the equivalent gate circuit consists of a small gate resistance and a small gate capacitance (Figure 5). When used in a half bridge configuration, care must be taken to not over-voltage or under-voltage the gate.
Figure 6: Effect of dv/dt on a device in the off-state and requirements for avoiding Miller-induced shoot-through.
Figure 7: The Miller ratio of 2nd and 4th generation eGaN FETs and state-of-the-art Si MSOFETs for drain-to-source voltages at half their rated voltage.
Figure 8: Impact of a positive di/dt of an off-state device with common-source inductance.
Figure 9: di/dt induced turn-on (shoot-through) of an off-state device with under-damped gate turn-off power loop.
A high, positive-voltage slew rate (dv/dt) on the drain of an off-state device can occur in both hardand soft-switching applications, and is characterized by a quick charging of the device’s capacitances as depicted in Figure 6. During this dv/dt event, the drain-source capacitance (CDS) is charged. Concurrently, the gate-drain (CGD) and gate-source (CGS) capacitors in series also are charged. If unaddressed, the charging current through the CGD capacitor will flow through and charge CGS beyond VTH and turn the device on. This event, sometimes called Miller turn-on and well known to MOSFET users, can be very dissipative.
To determine the dv/dt susceptibility of a power device, a Miller charge ratio (QGD/QGS1), as a function of drain-to-source voltage, needs to be evaluated. A Miller ratio of less than one will guarantee theoretical dv/dt immunity . In Figure 7, the large reduction of Miller ratios in EPC’s latest generation eGaN FETs is shown, reduced by at least a factor of two and resulting in the entire product line falling below a value of 1 at half their rated voltage. Also plotted, as triangular dots, in Figure 7 are Miller ratios for current silicon MOSFETs which in general are much higher.
A rising current through an off-state device, as shown in Figure 8, will induce a step voltage across the common-source inductance (CSI). This positive voltage step will induce an opposing voltage across CGS. For a rising current, this causes the gate voltage to be driven to a negative value and, with insufficient damping of the off-state gate loop LCR resonant tank, this initial negative voltage step across the gate could induce positive ringing and cause an unintended turn-on and shoot-through as shown in Figure 9.
It is possible to avoid this type of di/dt turn-on by sufficiently damping the gate turn-off loop, although some level of undershoot may be preferred as described in the dv/dt immunity case above. However, increasing the gate turn-off power loop damping through an increase in gate pulldown resistance would negatively impact dv/dt immunity. Thus, adjusting gate resistance alone for devices with marginal Miller charge ratios may not be enough to avoid di/dt and/or dv/dt turn-on.
A better solution is to limit the size of the CSI through improved packaging and device layout. This is accomplished by separating the gate and power loops to as close to the GaN device as possible, and minimizing the internal source inductance of the GaN device, which will remain common to both loops.
For a more detailed discussion on the impact of inductance on circuit performance please see our white paper: Impact of Parasitics on Performance.
Below are the key characteristics of preferred gate drivers for EPC GaN transistors. For a controller to be functional with eGaN FETs, its driver would also need the following:
- Ground bounce immunity: The gate driver design should be made with the assumption that the driver ground and the controller ground can differ significantly, and the input logic pin must be immune to noise-induced changes in logic state.
- High dv/dt immunity for high-side drivers: logicisolators or level-shifters used to transfer the control logic signal to the floating high-side device need to be immune to high dv/dt rise and fall times without changing the logic state. For 100 V and below, 50 V/ns immunity should be sufficient, while higher voltages require increasingly higher dv/dt immunity.
- Low inductance surface mount package and optimized pin-out: The gate drive and high-speed GaN device need to be closely connected with the interconnection impedance minimized. This requires pin-outs and packaging options that complement the GaN transistor. QFN/DFN or WLCSP packages are preferred. Complementary pin-outs have the VBS/HG/VSW next to each other and VCC/LG/PGND next to each other.
- Gate power loop inductance minimization: The gate driver should be designed to minimize the inductance between the VDD supply capacitor and the actual gate driver power devices (sink and source devices). This will minimize gate driver rise time and maximize driver di/dt. Driving the smaller Generation 3 devices with input capacitance in the 50 pF range are best served by a driver with rise and fall times in the 500 ps range or below.
- Gate drive strength: For a general purpose GaN driver, the speed of the driver needs to be matched to the size and speed of the device being driven. This flexibility requires a low-resistance gate driver with the option of additional external resistors. Gate drive strength in the 1 to 3 ohm range for pull up/ pull down is preferred.
- Gate drive strength for pull up / pull down should be in the 1 to 3 ohm range.If it is larger it will be too slow (or limited to lower power applications), while much smaller can induce ringing issues - especially when in combination with high inductance wirebonded packages.
- Regulation of gate drive supply voltage: Both lowside drivers, and especially high-side drivers, need to regulate the gate drive supply voltage to avoid an over-voltage condition on the transistor gate. Gate drive specification should be 5 V ±0.5 V max, 5 V ±0.25 V is preferred.
- Dead-time: Minimizing dead-time will limit the loss component of the ‘body-diode’ forward voltage. Dead-time of 20 ns or less is preferred. For a more detailed discussion of dead-time management for eGaN FETs please see the white paper: Dead-Time Optimization for Maximum Efficiency
- High Frequency Operation: eGaN FETs are capable of switching frequencies greater than 10 MHz. An optimized gate driver for high frequency operation would include low minimum on-time, minimal internal capacitance between ground and the high side supply as well as a bootstrap supply with minimal reverse recovery, such as available through the use of an external schottky diode. For a more detailed discussion of driver impact on high frequency operation, please see the article Hard Switching Converters at High Frequency
For a more detailed discussion please see eGaN® FET Drivers and Layout Considerations
For a current listing of enhancement-mode GaN compatible IC’s, please see eGaN Drivers and Controllers
Due to the fast switching and high current-carrying capability of EPC’s GaN transistors, special considerations should be taken into account when designing printed circuit boards utilizing these devices. To illustrate this we will look at the example of a buck converter.
Figure 10: Synchronous buck converter with parasitic inductances
In a practical buck converter, there are two major parasitic inductances, as shown in Figure 10, which have a significant impact on converter performance:
1. Common source inductance, LS, is the inductance shared by the drain to source power current path and gate driver loop.
2. High frequency power loop inductance, LLOOP, is the power commutation loop and comprised of the parasitic inductance from the positive terminal of the input capacitance, through the top device, synchronous rectifier, ground loop, and input capacitor.
Figure 11: (a) Parasitic inductance impact on power loss (VIN
= 12 V, VOUT
= 1.2 V, IOUT
= 20 A, FSW
= 1 MHz) Top Switch: EPC2015
, Synchronous Rectifier EPC2015 (b) eGaN FET LGA package
The common source inductance, LS, has been shown to be critical to performance because it directly impacts the driving speed of the devices -. The common source inductance is mainly controlled by the package inductance of the device, and varies from package to package , . For the eGaN FET, the LGA package (Figure 11b) offers low common source inductance, reducing loss, as shown in Figure 11a.
The high frequency loop inductance, LLOOP, impacts the switching commutation time and the peak drain to source voltage spike of the devices. The high frequency loop inductance is controlled by the PCB layout and package inductance. In applications utilizing low package parasitics, such as with the eGaN FETs’ LGA, the PCB layout dominates the high frequency loop inductance -.
With the significant reduction in package related inductance provided by the eGaN FET, the common source inductance is minimized and is no longer the major parasitic loss contributor. The high frequency loop inductance, controlled by PCB layout becomes the major contributor to loss, making layout using the eGaN FETs critical to high frequency performance.
Optimal eGaN FET Layout for Reduced Parasitics
Figure 12: Proposed optimal power loop with eGaN FETs (a) Top view (b) Top view of inner layer 1 (c) Side view
To enable the high switching speed available from the eGaN FETs, devices were developed in land grid array (LGA) packages that not only have low internal inductance, but enable user to design ultra-low inductance into their board. To provide the benefits of reduced loop size, magnetic field self-cancellation, consistent inductance independent of board thickness, a single sided PCB design, and high efficiency for a multi-layer structure, an optimal layout is proposed for eGaN FETs. The design utilizes the first inner layer, shown in Figure 12b, as a power loop return path. This return path is located directly underneath the top layer’s power loop, Figure 12a, allowing for the smallest physical loop size combined with field self-cancellation. The side view (Figure 12c) illustrates the concept of creating a low profile self-cancelling loop in a multilayer PCB.
The improved layout places the input capacitors in close proximity to the top device, with the positive input voltage terminals located next to the drain connections of the top eGaN FET. The eGaN FETs are located in the same positions as the lateral and vertical power loop cases. Located between the two eGaN FETs is a series of interleaved switching node and ground vias arranged to match the LGA fingers of the synchronous rectifier eGaN FET. The interleaved switching node and ground vias are duplicated on the bottom side of the synchronous rectifier. These interleaved vias provide three advantages:
(1) The via set located in between the two eGaN FETs provides a reduced-length high frequency loop inductance path leading to lower parasitic inductance.
(2) The via set located beneath the synchronous rectifier eGaN FET provides additional vias for reduced resistance during the synchronous rectifier eGaN FET freewheeling period, reducing conduction losses.
(3) The interleaving of the via sets with current flowing in opposing direction allows for reduced eddy and proximity effects, reducing AC conduction losses.
A high density power device must not only be more electrically efficient by generating less heat, but also enable superior heat conduction properties. The thermal efficiency of a package can be determined by comparing the two parameters, RθJC and RθJB, normalized to the package area. RθJC is the thermal resistance from junction-to-case, this is the thermal resistance from the active part of the eGaN FET to the top of the silicon substrate, including the sidewalls. RθJB is the thermal resistance from junction-to-board, this is the thermal resistance from the active part of the eGaN FET to the printed circuit board. For this path the heat must transfer through the solder bars to the copper traces on the board. In Table 1 is a compilation of thermally related characteristics for several popular surface mount MOSFET packages as well as two popular eGaN FETs.
Figure 13: RθJB
(Junction-to-Board Thermal Resistance) for several package styles listed in table 1, eGaN FETs represented by blue circular dots and Si MOSFETs represented by red square dots.
Figure 13 shows a plot of the junction-to-board resistance (RθJB) for each of the packages given in Table 1. Red square dots represent the MOSFET packages, and blue circular dots represent the eGaN FETs. The majority of the sampled packages fall on a single trend line indicating that performance for this element of thermal resistance is determined primarily by package size, and not technology. In contrast, in Figure 14 shows a plot of the thermal resistance from junction to case (RθJC). The CanPAK and double-sided cooling SO8 packages are far less efficient at extracting the heat out of the top of the package than either the Blade package or the eGaN FETs. The eGaN FETs, however, are over 30% lower than even the Blade  when normalized to the same area. This makes the eGaN FETs the most efficient thermal package for double sided cooling and most suitable for high density power designs.
TABLE 1: Comparison of Package Area and Thermal Resistance Components RθJC
Figure 14: RθJB
(Junction-to-Case Thermal Resistance) for several package styles listed in table 1, eGaN FETs represented by blue circular dots and Si MOSFETs represented by red square dots.
A further advantage of EPC’s GaN transistors is the capability for thermal resistance improvements to both thermal heat paths as shown in Figure 15. First, the PCB-to-ambient thermal resistance can be reduced through the addition of thermal vias, but with the thermal vias connecting to internal and external copper layers to spread the heat laterally. Secondly, the thermal interface material (TIM) resistance can be reduced in two ways:
Figure 15: Thermal diagram of board-mounted LGA GaN transistors with dual-sided cooling showing top-side heatsink and thermal vias with heat spreading through PCB.
1) By reducing the thickness of the device-to-heatsink interface: typically some form of spacer between heatsink and devices are required together with a TIM, as any mounting force applied to the heatsink cannot be transferred to the device for fear of cracking. This spacer determines the minimum distance between the heatsink and GaN devices.
2) By placing the thermal interface material on all sides of the devices and not just the top (case): this reduces thermal resistance, as the surface area of the device perimeter side walls is larger than the top surface, as shown in Figure 16.
Figure 16: Diagram of an LGA GaN transistor showing the die surface area and area of die perimeter.
Further improvements are possible by utilizing dual-sided heatsinking, forced-air cooling and through the use of exotic PCB materials, such as DBC (direct bonded copper) , or IMS (insulated metal substrate).
Using Device Models to Simulate Circuit Behavior
Although enhancement-mode devices are made to operate similarly to silicon MOSFETs, they cannot be readily modeled with traditional physics-based MOSFET models, as the physics of the GaN transistor is significantly different. The models developed for EPC’s enhancement mode GaN transistors  are a hybrid of physics-based and phenomenological functions to achieve a compact spice model with acceptable simulation and convergence characteristics. Temperature effects have also been included for conductivity and threshold parameters. The basic equivalent circuit for an enhancement mode GaN transistor is shown in Figure 17. The main components are: a voltage-controlled current source ID, capacitors, CGD, CGS, and CDS, and the termination resistors - RS, RD, and RG.
As a demonstration of device model and circuit considerations, a simple circuit was built and tested to compare device performance with that predicted by the model (Figure 18).
Figure 18: Schematic of basic demo circuit.
Figure 19: Comparison of simulated and measured demo circuit.
The circuit consisted of a voltage source charging a 13 μF cap through a 10 kW resistor used to isolate the voltage source from the device under test. The GaN transistor is driven with a 5 V pulse and the capacitor is discharged through a 0.8 W resistor and the device with a 0.1 W stray resistance. Comparison of the simulated results for the demo circuit show reasonable correlation with the measured values. Although not perfect, overshoot and ringing is qualitatively reproduced. Figure 19 shows the overlay of the gate and drain voltages vs. time for the measured and simulated circuit.
The models developed by EPC accurately reproduce the basic response of the devices under circuit operation conditions. Models for EPC devices are available for download from the Device Models page.
Curve Tracer and Auto-Testing Considerations
EPC enhancement mode GaN transistors generally behave like n-channel power MOSFETs. Common curve tracers, parametric analyzers, and automatic discrete device parametric testers that are used for an n-channel power MOSFET will be applicable for the characterization of GaN transistors. Below are some general guidelines for characterizing DC parameters using a Tektronix 576 curve tracer, Keithley 238 parametric analyzer, or a TESEC 881- TT/A discrete device test system.
Figure 20: Comparison of VTH
curves with and without a gate resistor for a 100 V eGaN FET.
Caution: GaN transistors are sensitive to static. GaN transistors have very low capacitances and a low maximum allowed gate voltage. Wrist straps, grounding mats, and other ESD precautions must be followed to avoid exceeding maximum device ratings.
VTH is the gate-source voltage (VDS = VGS) which produces a specified drain current on the datasheet. This test is typically done with the drain and gate shorted.
Caution for VTH curve tracer testing: If there is no gate resistor (RG) in series with the gate during the VTH measurement, you may see an oscillation on the gate which will result in a typical S curve like that shown in Figure 20. The oscillation voltage can become many times the input voltage. THESE OSCILLATIONS CAN DAMAGE OR DESTROY THE DEVICE.
IGSS is the gate-source leakage current with the drain shorted to the source.
Do not exceed 6 V on the gate in the positive direction or 4 V in the negative direction as that is the maximum gate rating for the device.
It is very important to have a very low resistance short between the drain and the source in order to get an accurate IGSS measurement. It is not recommended to use the Autorange function during IGSS testing on an automated tester such as a TESEC 881- TT/A, as range-changes during testing can lead to voltage spiking which may destroy the gate.
RDS(on) is the drain to source resistance with VGS = 5 V. Since RDS(on) is sensitive to temperature, it is important to minimize heating of the junction during the test. A drain pulse test is therefore used to measure RDS(on). Accurate RDS(on) measurement requires the use of Kelvin Sense on both drain and source. The locations of sense points have a strong influence on RDS(on) reading. It is not recommended to use the Autorange function during RDS(on) testing on an automated tester such as a TESEC 881-TT/A, as rangechanges during testing can lead to voltage spiking which may destroy the gate.
It is not recommended to use needles on a bare die to measure RDS(on). Too high current density at the probe needle/solder bump contacts could damage the device.
IDSS / BVDSS Measurement
BVDSS is the rated voltage of the device at VGS = 0 V. IDSS is the drain current at a specified drain-source voltage which is equal or less than the rated voltage of the device, with V = 0 V.
The true breakdown voltage for a GaN device from EPC is generally well above the maximum drain-source Voltage rating of the device. Therefore a BVDSS test should not be done on the device because the maximum VDSS rating will be exceeded. Degradation of device RDS(on) may be seen if the max rating is exceeded. It is also very important to short the gate and the source to avoid floating the gate with respect to the source and accidentally turning on the device. The device could be damaged during the IDSS testing if this occurs.
As with IGSS and RDS(on) measurements, it is not recommended to use the Autorange function during IDSS testing on an automated tester such as a TESEC 881-TT/A, as range-changes during testing can lead to voltage spiking which may destroy the gate. The use of “Function BVDSS” should also be avoided because the measurement of the drain-source voltage at a fixed drain current may exceed device VDS maximum rating.
Users should first verify that there is no spiking above voltage test settings during the IDSS measurements. It is recommended to use a controlled voltage ramp to aid in avoiding voltage overshoot.
It is very important to short the gate and the source to avoid floating the gate with respect to the source and accidentally turning on the device. The device could be damaged during the IDSS testing if this occurs. It is not sufficient to set the gate to 0 V; you must have a very low resistance short from the gate to the source.
A Final Note
How easy a device is to use depends on the skill of the user, the degree of difficulty of the circuit under development, how different the device is compared with devices within the experience of the user, and the tools available to help the user apply the device.
The new generation of enhancement mode GaN transistor is very similar in its behavior to existing power MOSFETs, only much faster, and therefore users can greatly leverage their past design experience and achieve new levels of performance in their products. The relatively high frequency response of eGaN FETs, is both a step function improvement over any prior silicon devices, and an added consideration for the user when laying out circuits.
On the other hand, there are several characteristics that render these devices easier to use than their silicon predecessors. For example, the threshold voltage is virtually independent of temperature over a wide range, and the on-resistance has a significantly lower temperature coefficient than silicon.
Table 2 is a summary comparison between a silicon power MOSFET and an EPC2001C GaN transistor’s basic characteristics.
Table 2: Summary comparison between 100 V silicon Power MOSFET and 100 V eGaN FET
User-friendly tools can also make a big difference in how easy it is to apply a new type of device. EPC has developed a complete set of device models available for user download. These models provide reasonably reliable circuit performance predictions that can enhance the engineer’s productivity and the time it takes to get a product to market.
Application notes and design tips codify the collective experience of engineers over the years. EPC’s GaN Library contains an extensive collection of application notes, white papers, and scholarly journals. Also, in its second edition, GaN Transistors for Efficient Power Conversion , is the defining textbook on GaN technology and its applications.
The power MOSFET is not dead, but is nearing the end of the road of major improvements in performance and cost. GaN is positioned to become the dominant technology over the next decade due to its large advantages in both performance and cost; advantage gaps that promise to widen as we plummet down the learning curve.
- Wu, T. “Cdv/dt Induced Turn-On In Synchronous Buck Regulators”, white paper, International Rectifier Corporation
- A. Elbanhawy, “Effects of parasitic inductances on switching performance,” in Proc. PCIM Eur., May 2003, pp. 251–255.
- G. Nobauer, D. Ahlers, J. Sevillano-Ruiz, “A method to determine parasitic inductances in buck converter topologies,” in Proc. PCIM Eur., May 2004, pp. 37–41.
- B. Yang, J. Zhang, “Effect and utilization of common source inductance in synchronous rectification,” in Proc. IEEE APEC’05, Mar. 2005, vol. 3, pp. 1407–1411.
- M. Pavier, A. Woodworth, A. Sawle, R. Monteiro, C. Blake, and J. Chiu, “Understanding the effect of power MOSFET package parasitic on VRM circuit efficiency at frequencies above 1 MHz,” in Proc. PCIM Eur., May 2003, pp. 279–284.
- D. Reusch, D. Gilham, Y. Su and F.C. Lee, “Gallium nitride based multi-megahertz high density 3D point of load module,” APEC 2012. pp.38-45. Feb. 2012.
- T. Hashimoto, T. Kawashima, T. Uno, Y. Satou, N. Matsuura, “System in package with mounted capacitor for reduced parasitic inductance in voltage regulators,” Applied Power Electronics Conference and Exposition, 2008. APEC 2008. Twenty-Third Annual IEEE, pp.187-191, 24-28 Feb. 2008.
- Y. Kawaguchi, T. Kawano, H. Takei, S. Ono, A. Nakagawa, “Multi Chip Module with Minimum Parasitic Inductance for New Generation Voltage Regulator,” Power Semiconductor Devices and ICs, 2005.
- A. Ball, M. Lim, D. Gilham, F.C Lee, “System design of a 3D integrated non-isolated Point Of Load converter,” Applied Power Electronics Conference and Exposition, 2008. Twenty-Third Annual IEEE, pp.181- 186, 24-28 Feb. 2008.
- D. Reusch, F.C. Lee, Y. Su, D. Gilham, “Optimization of a High Density Gallium Nitride Based Non- Isolated Point of Load Module,” Energy Conversion Congress and Exposition (ECCE), IEEE, Sept. 2012.
- Infineon Blade BSN012N03LS datasheet, www. infineon.com
- Infineon CanPAK S-size BSF134N10NJ3 G datasheet, www.infineon.com
- Infineon CanPAK M-size BSB012N03LX3 G datasheet, www.infineon.com
- Infineon S3O8 BSZ075N08NS5 datasheet, www.infineon.com
- Texas Instruments S308 Dual Cool SON 3.3x3.3mm CSD16323Q3C datasheet, www.TI.com
- Super SO8 BSC010N04LS datasheet, www.infineon. com
- Texas Instruments Super SO8 Dual Cool SON 5x6mm CSD16321Q5C datasheet, www.TI.com
- Efficient Power Conversion EPC2001 datasheet
- Efficient Power Conversion EPC2021 datasheet
- D. Reusch, “High Frequency, High Power Density Integrated Point of Load and Bus Converters,” Ph.D. dissertation, Virginia Tech, http://scholar.lib. vt.edu/theses/available/etd-04162012-151740/ 2012.
- R. Beach, A.Babakhani, R. Strittmatter, “Circuit simulation using EPC device models,” epc-co.com/ epc/documents/product-training/Circuit_Simulations_ Using_Device_Models.pdf
- A. Lidow, J. Strydom, M. de Rooij, D. Reusch “GaN Transistors for Efficient Power Conversion,” Second Edition, Wiley, ISBN 978-1-118-84476-2.
Download pdf version