Three-device Coordination
Utility → main → feeder → branch on a real one-line.
The real working pattern for selective coordination isn’t two devices. It’s three — sometimes four. A service entrance breaker feeds a switchgear bus, which feeds a feeder breaker, which feeds a motor branch. Each device has to coordinate with both the one above and the one below.
This lesson walks the cascade.
The one-line
A bolted fault on the load side of the branch should open the fuse and nothing else. A fault on the bus between the feeder and the branch should open the feeder. A fault between the main and the feeder should open the main. Each level isolates only its own zone.
The starting state
The widget below is seeded with three devices in series and overlap detection on. The main’s settings are deliberately wrong in two ways — its S delay is shorter than the feeder’s (a backwards step that makes the main trip first at S-band currents), and its S pickup is low enough that the main enters its S region inside the feeder’s operating range. The feeder, in turn, has its S pickup low enough that it overlaps the branch fuse’s clearing region for some fault currents. Red ✕ markers should be visible in both bad zones.
The settings ladder
Three-device coordination is solved one step at a time, from the bottom up:
Step 1 — Branch vs. Feeder
The branch fuse is fixed (200 A RK5, no adjustable settings). Tune the feeder so its curve stays above the fuse’s max-clear at every current.
What to change on the feeder:
- S pickup — push it above the fuse’s current-limiting threshold (≈30× rating = 6,000 A). Try 8× sensor = 6,400 A. Below that, the feeder stays in its long-time region, which is slow and safely above the fuse’s clearing.
- I pickup — turn it OFF or push it well right. The fuse goes current-limiting around 6,000 A — its clearing time goes vertical there. The feeder should not race the fuse at that point.
Step 2 — Feeder vs. Main
Now the main has to coordinate with the new feeder curve, not its starting state.
What to change on the main:
- S delay — set it at least 0.1 s longer than the feeder’s S delay. If the feeder S is 0.15 s, the main needs ≥ 0.25 s. The step in S delay is the selectivity margin that gives the feeder time to fully clear before the main starts.
- S pickup — push it above the feeder’s S pickup. Otherwise both LVPCBs pick up their S elements at the same fault current, and the smaller S delay step on the main is the only thing keeping them coordinated.
- I pickup — already OFF (good — the service main has the short-time withstand to ride through any fault while the feeder clears).
Step 3 — verify the long-time
The slowest part of the cascade is at low overload currents — the long-time region. Check that:
- Main L delay > Feeder L delay > Branch fuse max-clear at any overload current the system might see (≈1.5–3× the main rating).
- L pickups don’t matter as much for coordination (they’re separated by the device ratings) — but they do matter for protection. A main with L pickup at 0.9× of a 2000 A sensor will start its long-time trip at 1800 A, which is what you want for a service that’s never going to carry more than its nameplate.
A real engineer’s shortcut
Once you’ve done a few coordination studies, you’ll notice that a 1.5–2× separation between S pickups, combined with a 0.1 s step in S delays, will coordinate almost any LVPCB-to-LVPCB pair in practice — provided the available fault current isn’t off the chart.
The shortcuts don’t replace the curve check. They just give you a starting state that’s usually close, so you spend less time adjusting.
When the cascade can’t coordinate
There are real systems where selective coordination across three devices isn’t achievable with off-the-shelf trip units:
- Tight bus available faults that exceed the upstream device’s short-time withstand — you can’t turn instantaneous OFF, so the upstream races the downstream on every bolted fault.
- Mixed device types where a fuse upstream of an LVPCB is asked to coordinate — fuses go vertical at current-limiting, breakers don’t.
- Series ratings on the downstream device that bind it to a specific upstream — the downstream can’t be applied at its full AIC without the upstream in series.
In those cases the engineer’s options are: choose different hardware, accept a coordination exception (and document it), or move to a system topology that separates loads more aggressively (parallel service entrances, more transformers, separate bussing).
What’s next
The whole exercise so far has assumed the equipment downstream of each device can survive the fault long enough for the upstream device to clear. Module 3 brings cable and transformer damage curves onto the plot — the limits beyond which equipment burns up regardless of how the breakers fire. That’s where the coordination study turns into a real protection study.