In data centers, servers usually have redundant power supplies, which makes it possible to plug them to two different UPS units. This way, the server continues to function correctly if one of the PSU goes down, or if one of the UPS units is shut down for replacement.
Some servers don't have redundant power supplies. In order to be able to still connect them to two UPS units (in order to be able to replace one unit without shutting down the server first), special devices called transfer switches are used: they can receive input from multiple sources, and provide power for single-corded equipment. A transfer switch is usually expensive and contains lots of electronics inside. Basically, automated transfer switches provide power from one source, and when they sense that the source is down, they fallback to another source.
What makes it impossible to have just a cable which would plug into two sources on one side, and to one server on another side? In other words, something like that:
but the other way around, i.e. not a cable with one male and two female connectors, but a cable with two male and one female connectors?
In other words, why is it possible to wire two DC power sources in parallel, but one cannot (or shouldn't) wire two AC sources in parallel? Is it because the waves may not be synchronized and could then cancel out each other? Or something else?
simulate this circuit – Schematic created using CircuitLab
Best Answer
When two DC sources are paralleled, blocking diodes prevent back feeding from one supply to the other. The supply with the highest voltage will always supply current without being affected by the other.
There is no comparably simple and effective provision for paralleling two AC supplies. Two AC supplies from the utility can be obtained with matching phases and voltages, but if one goes down, it must be disconnected to prevent back feeding. Even dealing with small voltage or phase differences is not particularly easy.