Every automatic disconnection design stands on one number. When a live conductor faults to earth, the fault current is decided entirely by the impedance of the loop it flows around, and that current decides whether the protective device opens in 0.4 seconds or sits there while an exposed metal enclosure holds a lethal voltage. Loop impedance is not one calculation among many. It is the one that makes the protection real.
The loop and its arithmetic
Zs = Ze + (R1 + R2)
Ze is everything upstream of the installation: the transformer winding and the supply cables, declared by the distributor. Typical maximum declared values are 0.35 ohms for TN-C-S and 0.8 ohms for TN-S. R1 is the line conductor out to the fault, R2 is the protective conductor back, and both are yours to calculate from conductor resistance per metre times circuit length.
The disconnection times you are designing against
Table 41.1: for a TN system at 230 V to earth, final circuits up to 63 A must disconnect within 0.4 seconds. Distribution circuits get 5 seconds under Regulation 411.3.2.3. TT systems are stricter at 0.2 seconds, which is why TT design leans on RCDs rather than loop impedance alone.
The maximum Zs tables, and where their numbers come from
Tables 41.2 to 41.4 publish the maximum Zs for each device, and the values are not arbitrary. For a Type B MCB, which must see five times its rating to trip instantaneously, the formula is 230 × 0.95 ÷ (5 × In). For a B32: 230 × 0.95 ÷ 160 = 1.37 ohms, exactly the tabulated figure.
That 0.95 is Cmin, the factor acknowledging the supply might be running at the low end of its permitted tolerance at the moment of fault. A calculation done at nominal 230 V without Cmin passes circuits the standard fails.
Type C devices need ten times rating instead of five, so every maximum halves: a C32 gets 0.68 ohms where the B32 got 1.37. Choosing a C curve to ride through motor inrush quietly halves your permitted loop impedance, and on a long circuit that is frequently the constraint that bites first.
A worked example
A 32 A radial, Type B MCB, TN-C-S supply at Ze 0.35, wired in 4 mm² with a 1.5 mm² protective conductor, 35 m long.
Conductor resistances at 20 degrees: 4.61 and 12.10 milliohms per metre, summing to 16.71. Conductors incorporated in a cable run hot under fault, so Table F3 applies a 1.20 multiplier for 70 degree thermoplastic: 20.05 milliohms per metre.
R1 + R2 = 20.05 × 35 ÷ 1000 = 0.70 ohms.
Zs = 0.35 + 0.70 = 1.05 ohms. Against the B32 maximum of 1.37: compliant.
Run the same arithmetic backwards and you get the far more useful number, the maximum length: (1.37 − 0.35) ÷ 0.02005 = 50.9 m. Past that length this circuit cannot meet 0.4 seconds no matter how carefully it is installed, and the fix is a bigger protective conductor, a different device, or RCD protection doing the disconnection instead.
The temperature trap in measured values
Designers calculate at operating temperature. Electricians measure on a cold, unloaded circuit. A measured Zs will therefore read lower than the calculated maximum suggests it should, and comparing a cold measurement directly against a hot limit is the classic way to pass a circuit that fails. The conventional site check compensates by testing against roughly 80 percent of the tabulated maximum, but the design calculation must always be done hot, which is what the Table F3 multiplier is for.
A detail from implementing this
CurrentProof reports the disconnection time a circuit is required to achieve, checked against Zs, rather than the time a specific device will actually clear the fault in. Actual clearance needs the manufacturer's time current curve for the exact device, and pretending to know it without that curve would be manufacturing precision. That is a deliberate limit, stated in the tool's own output.
I mention it because the difference is worth understanding when you read any software's disconnection result, ours included. Required time is the standard's demand. Actual time is the device's promise. A design is compliant when the loop impedance guarantees the first; commissioning proves the second.