Power Blog Homepage

Assessing the Reliability of Redundant Power Supply Configurations

May 6, 2024 by Bruce Rose - 5 Minute Read

As mentioned in a recent CUI blog post, the reliability of a power delivery system can be increased by either connecting the outputs of power supplies in a parallel configuration or in a redundant configuration. The previous article discussed parallel output configurations and in this one we will discuss redundant power supply configurations.

Review of parallel connected power supply configurations and reliability

As a brief review, when power supplies are connected and operated in parallel the resulting power delivery system reliability can either be increased or decreased due to the multiple power supplies. The resultant power delivery system reliability depends upon the power density at which the supplies are operated. The article also mentioned that in a parallel configured power delivery system, the failure of a single supply can cause the entire power delivery system to fail.

Figure 1: Diagram of parallel power supply system with n number of power supplies

In the previous article we also discussed how the FIT (Failure In Time) rate of the system is desired to be low for high reliability (high MTBF, Mean Time Between Failures). The discussion also covered how paralleling power supplies can increase the system FIT rate and thus decrease the MTBF of the system.

Redundant configured power supply systems

In contrast to a parallel configured power delivery system, in a properly designed redundant power delivery system the power delivery system will continue proper operation when there is a failure of one or more of the power supplies. The design of a redundant power delivery system is such that when a power supply begins to fail (the output voltage exceeds and acceptable operating range) the failing supply is electrically removed from the system and a replacement power supply is electrically inserted into the system. In many redundant systems there will be a notification that the power supply has failed, and the faulty power supply can then be physically replaced. The design of a redundant power supply system should employ power supplies that individually have power output ratings sufficient for the load power and that have acceptable reliability ratings for the system requirements.

Redundant power delivery systems configured in 1+N configurations

The simplest redundant power configuration is a 1+1 topology. In a 1+1 configuration two power supplies are configured such that one supply normally provides power to the load and the second supply is electrically switched in to provide power to the load (and the first supply is electrically switched out) when a failure of the first supply is detected. The reliability of a power delivery system with a 1+1 redundant configuration is the sum of the reliabilities of each supply separately. In many applications it is reasonable that a repair person could replace the failed power supply soon after the failure is detected, thus ensuring high reliability system operation. A system with multiple redundant power supplies (1+N configuration) may be required in applications where either support from a repair person is difficult to guarantee or the continuous operation of the power delivery system is absolutely required. A 1+N redundant system is like a 1+1 redundant system, but with multiple redundant supplies so that if more than one supply fails there are multiple other supplies available to replace them.

N+1 redundant configurations

A N+1 configuration is often referred to as a redundant topology but is instead a complex mixture of parallel and redundant topologies. In this configuration multiple supplies are connected with the outputs in parallel and an additional supply is placed in a standby configuration to replace one of the parallel supplies if it fails.

In an N+1 power supply configuration, a method will need to be employed to ensure balanced current sharing between the N active supplies, the same as a paralleled output configuration. For the +1 redundant operation, a method is required to disconnect the output of the faulty supply, connect the output of the redundant supply, and control the redundant supply so it properly shares the power delivery system output load current.

Switches to Create Redundant Configurations

Passive diodes

One of the simplest methods to implement a redundant power supply design is to place passive switches (diodes) in series with the outputs of two power supplies and then to connect the cathodes of the diodes together, this is known as diode ORing.

Figure 2: Diagram of a redundant power supply system incorporating diode ORing

While passive diode ORing is an inexpensive solution for creating a redundant power delivery system, there are some challenges. If the power supplies, diodes, layouts, and operating environments are identical then the two power supplies will share the load current equally, which is not the desire in a redundant power delivery configuration. Some method should be implemented to cause all the output power to be delivered by the primary supply and none by the redundant supply, until there is a failure of the primary supply. The quality of the output voltage regulation in a diode ORing configuration will be degraded unless remote voltage sensing is employed to sense the voltage at the cathode of the ORing diode.

Another issue with passive diode ORing for redundant power supplies is the power loss in the ORing diode will degrade the power supply conversion efficiency.

Active FET switches

Another method to electrically isolate redundant power supplies is to use active switches (FETs) rather than diodes placed in series with the outputs of the power supplies. The main difference and advantage of using FETs is the FETs can be selected such that the voltage drop, and thus power loss, caused by the FETs can be much less than that caused by diodes. The operation of the FETs can also be controlled such that one supply can be selected to provide all the load current until it is deselected, regardless of the relative output voltage of the two power supplies. A drawback of using FETs as the isolation switches is circuits must be implemented to control the operation of the FETs. It should be noted that silicon power FETs include parasitic body diodes connected between the source and the drain of the FET. In some applications two FETs may be placed ‘back-to-back’ so the configuration of the two body diodes serve to block current flow in either direction when the FETs are not configured to be ON.

Figure 3: Diagram of a redundant power supply system using FET Oring with back to back FET’s and body diodes blocking current in both directions

Whether using diodes or FETs, the current rating of the isolation switch components must be large enough to ensure they can conduct the power supply output current without becoming damaged.

Summary

Hopefully, now it is clearer that two power supplies can be operated in a 1+1 or 1+N redundant configuration to ensure uninterrupted power delivery operation. The concept of N+1 redundancy is the merging of parallel operation and redundant operation and requires the design considerations applicable to both situations.

Categories: Fundamentals