Star topology fast, scalable, easy-going
Once the ring backbone is in place you can connect any number of nodes, or functional units (FUs), to the switches. The result? The classic star topology.
It has some good things going for it. It’s scalable and regardless of the number of FUs, it ensures very high-speed communication. In an LV distribution switchboard, it doesn’t care whether the mounting system is fixed, plug-in, or draw-out. And FUs go on working even when others are removed
Common mode impairs dependability
There is a weak point, however, and it is do with redundancy. It’s what dependability studies call the “common mode”. The sub-systems depend on the switches they’re connected to. A switch failure and all the nodes connected to it fail, too.
Another drawback is the length and number of cables – there are as many as there are functional units in each switchboard column.
Still, the star topology delivers. Combined with a fault-tolerant backbone in ring configuration, it is a simple, fast, dependable solution – even in the most critical applications.
But all that wiring is a headache. An alternative is the daisy chain topology.
The daisy chain loop with the 2-switch ring
In an LV distribution switchboard application, you can keep wiring to a minimum with the daisy chain configuration – for every column, just one cable in the cable tray.
However, it’s only good for fixed functional units. Because if you remove one FU, or it fails, you break the communication chain downstream.
Unless you close the chain into a loop that circles back to another switch on the backbone. It’s what we call a daisy chain loop (DCL) with a 2-switch ring. With two switches, you have redundancy, so the common mode is no longer an issue. Should a switch fail, data will flow the other way. Tests show failovers take typically 500ms for DCLs with between 8 and 32 nodes (FUs).
Daisy chains that can heal themselves
What more can a switchboard ask for? A system that addresses any motor via two different paths and tolerates first faults anywhere on the sub-system. But let’s not forget. Daisy chain topologies are suitable only for fixed functional units that cannot be removed.
But in normal LV switchboard operations, there are always some drawers pulled out. And the FUs in between are disabled.
There is one way of getting round – or bypassing – that constraint. A sub-system that “self-heals” after a unit is removed. By installing separate bypass switches that close when drawers are taken out, you close the loop. There is thus no break in the data flow. And when a drawer is in place or replaced, the bypass switch opens. And because it opens and closes automatically, we have a self-healing DCL.
If you opt for a backbone ring topology you have redundancy at the general architecture level. With the self-healing 2-switch DCL, you have a second level of redundancy at the sub-system level.
Which topologies have best met your needs? And which ones haven’t? What redundancy solutions have you chosen or even devised yourself? Please share.