There are a lot of misconceptions about the over current protection (OCP) and an explanation of why this protection exists is in order.
There is an international safety regulation called IEC 60950-1 that states that no single conductor can carry more than 240 VA in computer equipments. Since computer power supplies deliver continuous current, this means that no output wire from the power supply can carry more than 240 W.
This way the ATX12V specification includes a requirement for an over current protection circuit in order to shut down any rail that pulls more than 240 W.
When it comes to the +12 V output, we have that this equals to a current of 20 A (P = V x I; therefore I = P / V or 240 W / 12 V).
Of course this relatively low limit would prevent manufacturers from building higher wattage units. So they came with the idea of breaking down the +12 V output into two or more group of wires, each group with its own over current protection. For example, two groups of wires with an OCP configured at 20 A each would double the maximum allowed power for the +12 V output from 240 W to 480 W.
Each group of wires with its own separated over current protection (OCP) is called “rail” (although we personally prefer the term “virtual rail”). So a power supply with “two rails” means that its +12 V wires are divided into two groups and each group has its own OCP circuit.
Power supplies that have only one OCP circuit (or even no OCP at all) are called “single rail.”
Currently there are several single-rail power supplies with a current limit way above 20 A on the +12 V rail. How is this possible? If you pay attention, the IEC 60950-1 requirement is per conductor. So if you get a high-current rail and spread it into several wires and make sure no wire will carry more than 20 A/240 W, then you are good.
In summary, the difference between single-rail design and multiple-rail design is the presence of more than one OCP circuit for the +12 V wires on the later.
Some manufacturers add a color stripe on the +12 V (yellow) wires from the power supply in order to identify to each rail each wire is connected to.
Low-end power supplies, however, normally lie about the presence of two +12 rails. On their labels you will see the description of two +12 V rails (and even having some +12 V wires – usually the ones connected to the ATX12V/EPS12V cable – with a stripe on a different color), but inside the power supply these units don’t even have an over current protection (OCP) circuit and all wires are connected together on the same place, thus these units are in fact single-rail products.
So how can you visually identify the presence of separated over current protection circuits? Just looking at the wires is not enough, as a manufacturer can simply add wires with different colors to deceive you.
There are two basic components necessary to build the over current protection circuit: the power supply must have a monitoring integrated circuit supporting OCP (and with the number of channels compatible with the number of rails advertised by the manufacturer) and current sensors, also known as “shunts,” which are high wattage resistors with a known very low resistance. On Figures 3 and 4 you can see the most common physical aspects of these “shunts.”
click to enlarge
Figure 3: Example of “shunts” (current sensors).
click to enlarge
Figure 4: Example of “shunts” (current sensors).
Each “shunt” represent a +12 V rail. The two power supplies portrayed above have four sensors and thus they probably have four +12 V rails. If you follow the wires you will easily find out which wires are connected to which rail.
There is one detail, though. Some manufacturers use the same printed board for products with single-rail design and multiple-rail design. So you may find power supplies with more than one “shunt” that are in fact single-rail products, because even though the manufacturer added the "shunts", they are all connected to the same circuit, instead of using separated circuits.
So if you open a power supply and can find only one (or no) “shunt,” it is a single-rail design; if you find more than one “shunt,” it is probably a multiple-rail design (the “shunt” count tells you the number of +12 V rails), but it can be also a single-rail design. You can take a look at the datasheet of the monitoring integrated circuit to see how many overcurrent protection circuits (“OCP channels”) it has. If it has only one OCP channel, obviously you are facing a power supply with only one rail.
Although theoretically required by the ATX12V specification, several power supplies simply don’t carry this protection or only install it on the +5 V and +3.3 V rails but not on the +12 V, which doesn’t make any sense.
In order for you to understand even more how this circuit works, consider the schematics from Figure 5, which is based on the popular PS223 monitoring integrated circuit, which features four OCP channels. The components marked as RS5, RS33, RS12(1) and RS12(2) are the “shunts.” Notice how in this example the power supply has only two +12 V rails, as the other two OCP channels are being used to monitor the +5 V and +3.3 V outputs.
click to enlarge
Figure 5: Over current protection.
The OCP trigger point – i.e., the value at which it will kick in – is manually configured by the power supply manufacturer, usually by choosing the value of external resistors that are installed at one of the pins of the integrated circuit (resistors ROC5, ROC33, ROC12(1), ROC12(2) and RI in Figure 4).