[Introduction] This article provides an overview of the system requirements for the Open Compute Project Open Rack Version 3 (OCP ORV3) Battery Backup Unit (BBU). The article emphasizes the importance of efficient, smart BBUs that can provide power during power outages. Furthermore, this paper presents analog and digital design solutions, electrical and mechanical solutions, and their architecture developed to meet written specifications.

 Summary

This article provides an overview of the system requirements for the Open Compute Project Open Rack Version 3 (OCP ORV3) Battery Backup Unit (BBU). The article emphasizes the importance of efficient, smart BBUs that can provide power during power outages. Furthermore, this paper presents analog and digital design solutions, electrical and mechanical solutions, and their architecture developed to meet written specifications.

introduction

Data centers power the Internet, connecting communities around the world. Social media companies such as Facebook, Instagram, and Nearly every major company and government agency around the world requires reliable data center capabilities to operate and maintain their primary business functions with intelligent computing, storage, and search. As the number of users increases every year, data center capacity continues to grow at an astonishing rate to accommodate demand and technological advancements. In order to keep up with the growing demand, the system architecture of the data center must be constantly updated and upgraded.

OCP is an organization that shares data center designs. Its system architecture definition is based on the Open Compute Project Open Rack Version 2 (OCP ORV2), where the backplane voltage is nominally 12 V and the system power is 3 kW. On the other hand, increased usage leads to increased power requirements, which makes the power requirements of 12 V systems too high, which in turn is detrimental to overall system performance. To address this issue, the backplane voltage was increased to 48 V while system power remained constant, minimizing the current and copper traces required and reducing the heat dissipated by the backplane. This change improves overall system efficiency and reduces the need for complex cooling systems. This is the basis for the new Open Rack Version 3 standard (OCP ORV3).

                                                                                                                                         Figure 1. OCP ORV3 power architecture.

The reliability of the data center is a basic condition for ensuring normal operations. Adding BBUs to the system provides system redundancy. If a power outage or brownout occurs, it takes time for the system to become aware of the situation, save important data, and switch operations to another data center server (most likely in a different data center facility and location). These response operations must be completed in a seamless manner. Each rack uses a backup power system to regulate delayed power to the system. This requirement is clarified in the latest standard ORV3 BBU: Based on the power stored and regulated by the lithium-ion battery, each BBU unit needs to provide 15 kW power output to maintain the system operation for 4 minutes.

Under the guidance of this specification, Analog Devices worked with the OCP organization to complete and produce a reference design solution, which includes: a bidirectional power converter for dedicated charging and discharging operations through a single circuit, a battery management system (BMS) device, a Firmware and GUI support on-board design system host microcontroller as well as hardware amplification.

Design requirements and hardware implementation

The concept and design requirements required to meet the BBU module standard are outlined in the specification provided by the OCP organization (version 1.3). The BBU module reference design is based on the ORV3 48 V proposal and consists of a battery pack with BMS, charge/discharge circuit and other functional blocks, as shown in Figure 2.

                                                                                                                      Figure 2. OCP ORV3 BBU block diagram.

In addition to circuit requirements, the BBU module also needs to have several main operating modes during its service life, as follows:

►Sleep mode: The BBU module is in shipping or inventory status, or is not connected to the active bus. At this time, the battery discharge current is minimal to extend the storage time. BBU monitoring or reporting functionality is not available in sleep mode. When the bus voltage is detected to be higher than 46 V and the duration is greater than 100 ms but less than 200 ms, and the PSKILL signal is low, the BBU will wake up and exit sleep mode.

►Standby mode: The BBU module is fully charged and operating normally, and continuously monitors the bus voltage to prepare for discharge events. The BBU module operates in this mode for most of its life. The status and parameters of the BBU module are displayed on the upstream rack monitor via the communication bus.

►Discharge mode: When the bus voltage drops below 48.5 V for more than 2 ms, the BBU module discharge mode is activated. The BBU module is expected to take over the bus voltage within 2 ms, with a backup time of 4 minutes.

►Charging mode: When all conditions are met, the BBU module enables its internal charger circuit to charge its battery pack. Depending on the last depth of discharge of the battery capacity, the charging current can be anywhere between 0 A and 5.5 A. It also allows upstream systems to override the charging current via the communication bus. There should be a charger timeout control scheme based on calculated charging current.

►Operating status check (SOH) mode: The BBU module routinely tests the battery pack capacity by forcing the battery pack to discharge. The BBU module should perform an SOH test every 90 days to determine the EOL status of the battery.

►System control mode: The BBU should allow the upstream system to control charging/discharging operations through the communication bus.

In addition to BBU module operation requirements, OCP also specifies standards for battery pack capacity, cell type and battery pack configuration. The specific instructions are as follows:

►Battery pack capacity: The BBU module can provide 3 kW backup power for no more than 4 minutes over a 4-year period.

►Cell type: The BBU module should be lithium-ion 18650 type, with a cell voltage of 3.5 V to 4.2 V, a battery capacity of at least 1.5 AH, and a continuous rated discharge current of 30 A.

►Battery pack configuration: The battery pack configuration of the BBU module is 11S6P (6 series combinations are connected in parallel, and each series combination is composed of 11 cells connected in series)

 In addition, the BBU module requires a BMS to provide battery charging/discharging algorithms, protection, control signals and communication interfaces. The BMS is also responsible for establishing the cell balancing circuit to keep the cell voltage in the battery pack within a ±1% (0.1 V) tolerance.

 The reference design block diagram (see Figure 3) shows the selected devices and the various components integrated to accomplish certain tasks. They form a circuit that provides uninterruptible power, determines module health and faults, and performs module communications. LT8228 is a bidirectional synchronous controller located in the BBU module. The device provides power conversion during line power interruptions and battery charging during non-fault operation. The LT8551 is a 4-phase synchronous boost DC-DC phase expander that works in conjunction with the LT8228 to increase the discharge power delivery capability to 3 kW per BBU module. In addition to the power conversion IC, the BBU module also contains the MAX32690, an ultra-low-power Arm® microcontroller responsible for the operation of the entire system. The LTC2971 is a 2-channel power system manager that enables precision sensing and fault detection of the power path, as well as critical voltage droop functionality. The MAX31760 is a precision fan speed controller used to perform system cooling functions during charge and discharge operations. EEPROM is used as a data storage device, allowing the user to recover any useful data while the BBU module is available. In addition to the power converter and the microcontroller responsible for general management tasks, the BMS IC is included in the design. The ADBMS6948 is a 16-channel multi-cell battery monitor used to monitor battery voltage levels, while its inherent coulomb counter is used to determine state of charge (SOC) and SOH levels for cell balancing and battery life expectancy calculations. The battery operating status monitoring program is completed by the ultra-low-power Arm microcontroller MAX32625. Both microcontrollers have been carefully selected to reduce overall power consumption and thereby extend battery life during BBU sleep mode operation.

In addition to the devices provided, this reference module provides and builds a BBU module (see Figure 4a) and a BBU shelf (see Figure 5) to house and demonstrate a reference design compliant with the OCP ORV3 BBU module and shelf mechanical specifications . The BBU shelf includes 6 BBU module slots, so a single BBU shelf can provide up to 18 kW of backup power as needed.

             Figure 3. ADI OCP ORV3 BBU block diagram

                                                            Figure 4. (a) 3D rendered mechanical overview of the ADI BBU module, (b) airflow simulation.

Mechanical rendering and airflow simulation are two architectural advantages of the BBU module reference design. First, it supports visualization, providing accurate and attractive representations. Mechanical structural analysis can help identify design issues and potential changes early, which helps the entire design process. Last but not least, it reduces the need for time-consuming and expensive physical prototypes. In addition, airflow simulation can provide performance analysis to help identify potential problems and improve design efficiency. It is also responsible for thermal management, helping to identify hot spots, optimize heat loss, and enhance overall system reliability. In addition, it reduces risk by planning the battery pack space according to safety and compliance requirements. See Figure 4b for more information.

                           Figure 5. 3D rendering of an ADI BBU shelf with six BBU modules inserted.

Data and results

Test results presented below include steady-state performance measurements, functional performance waveforms, temperature measurements, and operating mode transitions. The following configurations were tested using the BBU module reference design:

Table 1. ORV3 BBU module parameters

Performance data

Efficiency and power loss

The BBU module reference design demonstrates its ability to achieve higher efficiency and lower power loss while meeting the constraints of the ORV3 BBU specification. Discharge and charge limits are set to 97% and 95% respectively. During discharge operation, the average efficiency measured at half load (31.6A) was 98.5%, while the average efficiency at full load (63.2A) was 98%. Lower MOSFET drain-source on-resistance and carefully chosen switching frequency will help improve efficiency and reduce ripple current due to the larger inductance. In addition, the BBU module achieves a high average efficiency of 97% during charging operation with a 5 A load. When operating at a switching frequency of 400 kHz using the same inductor value, efficiency is improved and power losses are substantially reduced. High efficiency and lower power losses will help extend battery life cycles and reduce fan speeds required for cooling. See Figure 6.

On the other hand, the conduction losses of the control and synchronization MOSFETs affect the overall power loss during BBU discharge and charge operations.

                                     Figure 6. Respective efficiency and power loss during discharge and charge operating modes.

Output voltage drop

Another requirement of the ORV3 BBU specification is to consider voltage drop during discharge operating mode. Voltage droop refers to the intentional reduction of the BBU backplane voltage when driving system loads. The BBU backplane voltage will change in real time based on the system load current measured by the LTC2971 online DAC. Therefore, the backplane voltage drop from no load to full load remains below the ±1% limit required by the ORV3 BBU. See Figure 7.

                              Figure 7. Output voltage drop during discharge operating mode.

switching waveform

Examining switching waveforms can provide valuable information for performance evaluation, failure analysis, efficiency optimization, EMI reduction, and safety considerations. It allows engineers to identify and solve problems, optimize system performance, and ensure reliable and efficient operation of data center BBU modules.

The switching operation of the BBU module is critical during discharge operating mode, converting the 30 V to 44 V battery pack voltage to the 48 V backplane voltage. This is accomplished through a synchronous power MOSFET, which is accurately regulated by the LT8228 pulse width modulation (PWM) signal, and the accompanying LT8551 repeats the operation of the LT8228. The switching frequency and current sharing of each phase cause the voltage to increase and are important factors affecting its operation. The switching waveforms of the main converter and its polyphase expander at full load are shown in Figure 8. In charge operating mode, the bidirectional converter operates in single phase, reducing the 49 V to 53 V backplane voltage to 44 V to charge the battery pack. It works by rapidly switching a synchronous power MOSFET and ramping up the inductor current. The switching waveform of the bidirectional converter under 5 A load is shown in Figure 9.

          Figure 8. Switching waveforms of the main controller and expander during discharge operating mode when operating with a 44 V input and 63.2 A output load.

                                              Figure 9. Master controller waveforms during charge mode when operating with a 53 V input and 5 A output load.

Thermal properties

Thermal performance and efficiency must be carefully balanced. BBU modules must be able to withstand high temperatures and continue to operate without overheating, while also operating at ideal efficiency to convert as much input power into output power as possible. In Figure 10, the worst-case board temperature measured during discharge operating mode (approximately 4 minutes at full load) was only 40°C to 60°C. In charge mode, the temperature of the synchronous MOSFET is below 50°C. A properly constructed air cooling system can reduce the heat generated by components and prevent thermal runaway. To avoid overheating of the battery stack, proper cell spacing and proper airflow are required. See Figure 11.

Working mode conversion

The operating mode transition of the BBU module is critical to ensure uninterrupted power supply during power interruptions or changes. This process includes the smooth transmission of battery pack energy to the backplane of the data center, ensuring that important systems and equipment remain operational for 4 minutes. The BBU module continuously monitors the backplane bus voltage. When the bus voltage drops to the BBU module activation level (48.5 V) within 2 ms, the BBU module backplane voltage must ramp up to provide full power to the bus within 2 ms. During the entire conversion process, the bus voltage must not drop below 46 V. After the BBU module detects that the bus voltage exceeds 48.5 V and continues for more than 200 ms, it exits the discharge working mode. See Figure 12.

                                              Figure 10. Thermal performance of the circuit board when running at full load in each of the discharge and charge modes of operation.

                                                 Figure 11. Battery stack gap design.

                                                                        Figure 12. Transition from steady state to power interruption state.

Summarize

To save energy, data centers are moving to 48 V systems. Due to smaller currents, copper losses and smaller power bus sizes, 48 V server racks are more efficient than 12 V server racks in terms of power consumption, cooling, size and cost. A high-efficiency, unregulated front-end circuit stage followed by a voltage regulator that adjusts to the appropriate load makes the design ideal for data center server microprocessors and memories. This high-level thinking, coupled with OCP’s latest innovations, lays the foundation for more efficient power distribution and smart backup battery unit designs that can support continuous and seamless operations.

Selecting and implementing appropriate devices for BBU modules and laminates can simplify the overall design, extend battery life, shorten long engineering development cycles, and effectively reduce engineering and production costs. In addition, providing mechanical simulation simplifies the prototyping step, obtains data that can be used to improve thermal dissipation and thermal management, and enhances design reliability. Finally, providing appropriate and well-designed firmware algorithms and sequences ensures that the BBU operates easily and smoothly.

The second part of this series will introduce the various main microcontroller functions and operations of the BBU module, specifically designed for general BBU management tasks. Additionally, Part Two will provide a more in-depth overview of how to monitor useful information and how to use this information to build and execute correct workflow routines.