“While the world is very excited about the technological advancement of 5G, designers are now busy working on how to provide power supply for base stations. The core of 5G’s future success is whether it can achieve the three desired goals: huge data throughput, ultra-low latency, and large-scale connections (see Figure 1). 4G base stations can provide a powerful downlink, but the uplink needs to be improved.
Authors: Giuseppe Bernacchia, Senior Chief Application Engineer, Infineon Technologies, Moshe Domb, Director of Application Engineering, Infineon Technologies
While the world is extremely excited about the technological advancement of 5G, designers are now busy working on how to provide power supply for base stations. The core of 5G’s future success is whether it can achieve the three desired goals: huge data throughput, ultra-low latency, and large-scale connections (see Figure 1). 4G base stations can provide a powerful downlink, but the uplink needs to be improved. In contrast, only the improvement of 5G base stations in terms of uplink capabilities requires more power, coupled with the higher data throughput and the number of users supported by each base station, it will only further increase the design engineer’s face Challenge. Most importantly, there are also some practical problems in the layout of related infrastructure.
Figure 1: The goal of 5G is to support huge data throughput, ultra-low latency, and a huge number of connections.
Many base stations are installed inside or above specially rented properties. There are strict restrictions on footprint, equipment volume, and allowable installation weight. Telecommunication equipment also needs to face these restrictions during the transition to 5G in the industry. In addition, additional micro base station transceivers (BTS) and active antenna systems (AAS) covering sub-GHz and millimeter wave spectrum are required to provide the coverage and capacity of large-scale connections (see Figure 2), which requires the rental of new Site. But these new sites also have strict weight and volume restrictions, and may also have an upper limit on available power. In the future, it is expected that there will be a trend of wall-mounted and pole-mounted installations, and will be integrated with existing public facilities such as street lighting, which is directly related to investment costs and budgets. The rent of the selected site has a significant impact on the overall operating expenses (OPEX) and will account for more than one-third of the expenditure. If you can design smaller, lighter, and more energy-efficient base stations, you can achieve cheaper rental costs and lower electricity bills, then more capital expenditures (CAPEX) can be spent on important radio and power equipment .
Figure 2: Large-scale connections require small base stations BTS and active antenna systems, all of which have demanding power requirements.
Studies have shown that in mature markets, energy costs account for 10% to 15% of OPEX. In developing markets, especially those markets that are far away from the grid and rely on battery-powered sites, this proportion will rise to nearly 50%. Of the energy provided, only 15% is used for data transmission between base stations and mobile phones, 35% is used for redundant systems, and 15% is consumed by power supplies and fans. Personnel costs related to site maintenance are the second most important part of operating costs. China Mobile believes that 70% of downtime cases are related to power supply issues. Therefore, it is necessary to focus on the robustness and reliability of the system.
Although the performance of silicon power devices has gradually improved over the years, the emergence of wide band gap (WBG) technologies such as gallium nitride (GaN) has immediately provided power converter designers with huge room for improvement and can be applied to their product designs.
GaN High Electron Mobility (HEMT) transistors provide extremely low and temperature-independent RDS(ON). Coupled with the output capacitor (EOSS) In the lower energy, low leakage source (QOSS) And gate charge (Qg) and almost zero reverse recovery charge (Qrr). Compared with silicon devices, GaN has a significant improvement in the figure of merit (FoM). Compared with silicon MOSFETs and IGBTs, GaN can also operate at higher switching frequencies. It not only improves the efficiency of power factor correction (PFC) and DC-DC converters, but also allows the use of smaller passive components. This has a significant impact on the weight and size of the power converter. This improvement can be translated into higher power density and smaller size, or the existing design can be converted to flatter or fanless operation.
Here, two common power converter designs are replaced with wide bandgap devices to evaluate the actual improvement that Infineon’s CoolGaN technology can provide over existing silicon power devices. 48V to 12V DC-DC ¼ bricks are the main products in the telecommunications industry, usually designed with 80V silicon MOSFETs, including the OptiMOS 5 series.
When using the OptiMOS 5 series to implement a 1kW fixed frequency LLC, the measured peak efficiency is 97.58%. Although the on-resistance of such devices is specified at ID Remains relatively stable within the range, but this requires a 10V gate drive (VGS) Can be realized. Moreover, although the output capacitance (EOSS) Compared with previous generations, it has been reduced, but its further improvement is limited by the physical characteristics of silicon devices.
We changed the design to CoolGaN and made other modifications to adapt to these transistors, such as replacing the gate driver approach. The same 1kW LLC design achieved a peak efficiency of 97.70% (see Figure 3). Although this increase in absolute peak efficiency may seem small, it must not be forgotten that it means a 5% reduction in heat generation. In addition, the low-load end of the power supply range has the greatest improvement, which helps the telecommunications industry to achieve the desired stability and high efficiency in the 30% to 100% load range.
Figure 3: Efficiency comparison of 1 kW fixed frequency 48V to 12V ¼-brick DC-DC LLC when using silicon MOSFET and CoolGaN.
Compared with silicon MOSFETs, applying CoolGaN in this isolated DC-DC design requires a different gate drive method (see Figure 4). RC coupling circuit is usually used, but this may lack negative pressure drive under certain circumstances. In addition, the switching dynamics may be limited by the duty cycle of the signal being driven. An EiceDRIVER series dedicated to GaN HEMTs ensures optimal control. They can provide the continuous gate current of a few milliamps required to maintain the “on” state. Due to fast switching transients and inherently low threshold voltages, they can also provide negative “off” voltages that are independent of duty cycle.
Figure 4: Fixed frequency 1 kW DC-DC LLC used to compare the efficiency improvement of CoolGaN over silicon MOSFETs.
Design a 3.6kW, 385V to 52V DC-DC LLC in the same way. If the silicon MOSFET is changed to CoolGaN, higher efficiency can be achieved. Although the silicon MOSFETs and GaN transistors used show similar RDS(ON) Value, but the power density of CoolGaN™ here is 160W/inch3 The design has a peak efficiency of 97.83% (see Figure 5), which reduces heat loss by approximately 15% compared to a silicon-based design.
Figure 5: Efficiency comparison of CoolGaN and silicon MOSFET in 3.6 kW DC-DC LLC.
Based on CoolGaN’s high-efficiency totem pole PFC, the further improvement in efficiency is mainly due to the addition of medium-voltage CoolGaN to the two full bridges of the secondary synchronous rectifier, which can provide the best load distribution (see Figure 6). If the cooling fan and internal management power supply are removed, the GaN-based design can achieve a peak efficiency of nearly 98.5%.
Figure 6: 3.6 kW DC-DC LLC full bridge to full bridge design, with two transformers in series on the primary side and parallel on the secondary side to force load sharing on the secondary side.
Of course, it is not just the innovation of semiconductor technology that is provided at the application level. Miniature surface mount packages with low inductance leakage can simplify the design, especially at high switching frequencies. The size of PG-VSON is only 5 × 3 × 1.075 mm, which helps to keep PCB wiring short. At the same time, its construction is suitable for passive heat dissipation and thin or narrow rack or wall-mounted installation solutions (see Figure 7).
Figure 7: The PG-VSON surface mount package has low inductance leakage and is very suitable for high-frequency switching CoolGaN applications.
Although GaN is generally considered to be an important part of the realization of 5G radio frequency transceivers, it is clear that its role is much more than that. The actual requirements of replacing the existing 4G BTS hardware to support 5G and purchasing and renting new sites for AAS and small cell BTS means that factors such as size, volume and weight and system functions need to be considered. Infineon’s CoolGaN HEMT transistor product portfolio is used in combination with EiceDRIVER™ driver products to ensure improved power converter efficiency within the required load range. Through innovative heat dissipation and construction methods, the weight and volume of the system can be reduced, while ensuring the robustness and reliability of the system, and minimizing maintenance and downtime.