Improving Communication Power Performance Using Synchronous Rectification and Current Sharing cPCI Interface Secondary Power Controller

Compact PCI (cPCI) power supplies have been widely recognized for their applications in the computer, industrial and telecommunications fields. To optimize system applications, a cPCI power supply uses standard industrial mechanical structures and high-performance connection technology. However, traditional cPCI power supplies use diode rectification technology, which causes large power losses and limits the available output power. This article will introduce a cPCI power supply using synchronous rectification technology, which utilizes a unique secondary synchronous rectification and an integrated power controller SC4910 with output current equalization function. We can see that this type of cPCI power supply will have great improvements in efficiency and performance

The standard for cPCI power supplies is PICMG 2.11. This standard mainly defines the electrical and mechanical requirements of cPCI power supplies and also defines the mechanical interface and signal interface between the power supply and the system backplane. Mechanically, cPCI power supplies must conform to standard rack dimensions and their panels are compatible with IEEE 1101.10. A standard 47-pin connector from Positronic is installed in the power supply device for input/output power and signal interfaces. Electrically, cPCI power supplies must meet electrical performance requirements such as voltage and current, output current sharing, and output remote detection. 3U and 6U racks are used in PCI systems. 3U units generally provide approximately 200 W to 250 W of output power, and 6U units generally provide approximately 400 W to 500 W of output power.
For telecommunications and networking applications, the input voltage of cPCI power modules is usually 48V. PICMG 2.11 does not specify the maximum load, full load, and minimum load requirements for each output; nor does it specify that a 3U power module must be equipped with 200W of power and a 6U power module must be equipped with 400W of power. The total power in the 3U and 6U racks depends mainly on the efficiency of the cPCI power supply and the available cooling methods for the PCI system. The current trend is to integrate more power in a 3U unit and minimize the space occupied by the cPCI power supply in the system rack, thereby freeing up more space for cPCI application circuit boards.

Traditional cPCI power supply circuit

The cPCI power supply used in telecommunications has a DC input voltage of +48V. The circuits of the AC input cPCI power supply and the DC input cPCI power supply are very similar, except that the AC input power supply requires additional diode rectification, power factor correction circuit (PFC) and EMI. The AC power supply also has more stringent safety requirements than the DC power supply. Figure 1 is a block diagram of a traditional DC input voltage cPCI power supply.
As shown in Figure 1, a traditional cPCI power supply usually includes three parallel power converters in a 3U or 6U rack. The output voltages of the three power converters are +5V, +3.3V, and +12V, respectively, and the -12V output is generally obtained from the +12V power converter. The power hot-swap circuit is generally located at the DC input end to control the inrush current caused when the cPCI power supply unit is inserted into the backplane with an input of +48V. This hot-swap circuit is usually mainly controlled by a dedicated hot-swap controller. The cPCI power supply needs to provide N+1 redundancy for the system, and each output requires a load current sharing circuit to provide current sharing for multiple cPCI power supplies in parallel. The most common topology of power converters is the forward topology, which operates at a switching frequency of about 150 to 200kHz. This type of traditional cPCI power supply uses forward and freewheeling Schottky diodes on the secondary side, and uses low-voltage Schottky diodes as output redundant diodes. The feedback of each power converter is electrically isolated from the input and output through an optocoupler. In addition, a dedicated current sharing control chip is required. The efficiency of such a cPCI power supply is usually around 75%. If the output is 200W, there is about 66W of power loss, and the temperature in the 3U rack will rise significantly at an ambient temperature of 50℃ and an airflow of 200 to 400LFM. In fact, in the cPCI system rack, in order to ensure the reliability of the power supply, it is necessary to provide users with a power reduction curve similar to Figure 2. Although each power converter is designed to provide a higher output current, the total output power of the power supply is limited by the operating ambient temperature and the total airflow in the system. At 400LFM and 50℃ ambient temperature, the maximum output power of a 3U power supply with 75% efficiency will generally not exceed 200W. In order to increase the output power of a 3U rack, the only way is to reduce power consumption. This can only be achieved through the synchronous rectification technology discussed later.

Improved power supply with secondary synchronous rectification

In recent years, the performance of power MOSFETs has been significantly improved, and the price of such devices has dropped rapidly. As the on-resistance of MOSFETs has become very low, synchronous rectification technology has been used in many low output voltage application circuits. In order to improve the efficiency of the power supply, designers have no choice but to use synchronous rectification technology. Just like buying other types of power supplies, users always want to buy newer and more powerful cPCI power supplies in existing 3U and 6U racks. Traditional cPCI power supplies in 3U racks can only provide 200W output power. If the efficiency of the power supply can be improved to 85% to 87%, theoretically, a power supply with an output power of 400W can be installed. Figure 3 shows the circuit block diagram of a cPCI power supply using synchronous rectification technology.

As can be seen from Figure 3, power MOSFETs are used in place of traditional Schottky diodes. Each output secondary is connected to a secondary synchronous rectifier controller SC4910, which is used not only to control the secondary synchronous rectifier MOSFET, but also to control the primary MOSFET through a gate drive isolation transformer. This secondary controller makes it very simple to control the system load and implement secondary synchronous rectifiers and load current sharing.
Using the +5V converter as an example, let's see how the current sharing circuit works. See Figure 5.

(1) When multiple outputs of the +5V converters are connected in parallel, the converter control chip (SC4910) obtains the same ISHARE voltage.

(2) Because each converter uses current mode control, when the Vea of ​​each +5V converter is the same, their secondary output inductors will have the same peak current. Therefore, the Vea value represents the peak current of the output inductor of each +5V converter.
(3) If the current of one +5V converter (converter 1) is greater than the current of another +5V converter (converter 2), the Vea of ​​converter 1 will be greater than the Vea of ​​converter 2. At this time, the Vss of converter 1 will drop, thereby reducing its Vea until it is equal to the Vea of ​​converter 2.
(4) If converter 1 fails, the Ishare voltage of converter 2 will be readjusted to a new level to start its normal operation and share current with other operating converters.
(5) Since the peak main switch current is used for current mode control and current sharing control, there is no need to use a sense resistor to sense the average current of the secondary inductor.
(6) Since such a current sharing circuit mainly uses the peak current in the secondary output inductor of each converter to control the average current in the inductor (equal to the converter output current), the error between the output inductor values ​​of each converter will cause the error of the output current of each converter. The experimental results show that the current sharing error is generally within 3% to 7% under heavy load.
Quantitative loss analysis
Below we will do a quantitative loss analysis of the traditional diode rectified cPCI power supply (Non Syn) and the synchronous rectified cPCI power supply (Syn.). Let's look at the power loss of a 200W 3U traditional cPCI power supply and the power loss of the same 200W 3U cPCI power supply using SC4910 to achieve synchronous rectification. Both the +5V and +3.3V converters are designed for a typical 40A maximum load, while the +12V converter is designed for a typical 7A maximum load. The -12V output has a very low current and is not analyzed here.

From Figure 6 (ac), we can see that the power loss of the synchronous rectifier converter is much lower than that of the traditional diode rectifier converter. For the +5V converter, the traditional converter's loss in the rectifier and redundant diodes alone is greater than the total loss of the same synchronous rectifier converter. Figure 7 further illustrates this point. In a traditional rectifier converter, the power consumption of the rectifier diode and redundant diode accounts for about 2/3 of the total power consumption. In a synchronous rectifier converter, the power consumption of the synchronous rectifier and redundant MOSFET accounts for only 1/3 of the total power consumption of the converter. Figure
8 is a comparison of the power consumption and efficiency of 200W and 400W traditional non-synchronous rectifier cPCI power supplies and synchronous rectifier cPCI power supplies. It can be seen that the power loss of the 400W synchronous rectifier cPCI power supply is approximately equal to the power loss of the 200W traditional diode rectifier cPCI power supply. Therefore, in the same 3U rack, the output power of the synchronous rectifier power supply is twice the output power of the traditional diode rectifier power supply.