As Graham says, if the impedance exceeds what's needed for 0.4s, the fault current would be too low to guarantee the operation of the magnetic mechanism, so we have to rely on the thermal mechanism and at those currents it can take longer than 5s at those sorts of currents.

There is an exception to that though - look at the details for type D MCBs - the current to operate the "instantaneous" magnetic mechanism is so high (20x In) that currents that would operate the thermal mechanism within 5s aren't interrupted by the magnetic mechanism first - so for those we do indeed have separate 0.4s and 5s figures. The tables along the appendix 3 graphs perhaps illustrate the point (e.g. 3A6).

(Note there is some dispute about type D MCBs as it seems that the product standard only guarantees opening within 8s rather than 5s for certain ratings), but the general principle holds the same).

    - Andy.

Ok. I got a bit more confused now....

When an MCB trips at 0.4s is the thermal or the magnetic part of the MCB that's tripping?

I always thought that because the tables are about fault currents, we have to do with the magnetic part of the MCBs and that the thermal is only for overloads

Ok. I got a bit more confused now....

When an MCB trips at 0.4s is the thermal or the magnetic part of the MCB that's tripping?

I always thought that because the tables are about fault currents, we have to do with the magnetic part of the MCBs and that the thermal is only for overloads

It's a mix of both.

For overload, think of your car's engine when you try to set off in third gear. The engine will try to drive the vehicle forward but will stall. Same with a motor under mechanical load starting from rest. It may take some time for the thermal part of the mcb to heat up before it trips, but what about a L to N short circuit? Lots of heat generated very quickly, so the thermal part of the mcb gets hot very quickly. The magnetic part will act faster in the 2nd scenario though.

An MCB (almost) never actually takes exactly 0.4 seconds to trip. That is a time, related to the human heartbeat period, that gives a good chance of surviving a 150V shock - assumed on a TN system with near equal CPC and live resistances and is used to decide what sort of breakers or other protection we need.
(for a 250V shock , as you may get in TT or other high Zs systems, or 400/690 3 phase, that maximum permitted shock time is usually reduced to 0.2 seconds.)

It is a horrible kludge as humans very enormously in conductivity, contact area, heart muscle condition etc, but these are figures that are known to save most folk most of the time,  and allow a set of rules for ADS  to be created.

If the magnetic part fires it does so near instantly, certainly in a few cycles and in any case almost certainly in less than 100msec for an MCB size mechanism in good working order.

The magnetic part is set to several time the thermal part for B (between 2-5 times) and C types (5-10 times) and is the bit that operates  when there is a dead short fault or a massive overload.

If the thermal part fires you have to wait while it heats up, and that time depeneds critically on the degree of oveload, as it is a race between (I2R) heat being generated by the heating element in series with the load, and the heat escaping out of the sides of the MCB. For this reason the exact cut-in point of the thermal part is quite loosly specified, and may go a bit early in a warm environment, or late or maybe even not at all, if it is very cold.

The thermal part is for moderate overloads, and protects the cable and fittings against cooking slowly. If it is a cold day the breaker may be late, but the cable will not be so hot either, so it sorts itself out.

for D types and beyond, the magnetic part is set so high,  that the thermal part may be heating pretty fast without the magnetic part operating, then the thermal times actually do become useful - at least they may pick up the 5 second operating event before the magnetic.
Mike.

When an MCB trips at 0.4s is the thermal or the magnetic part of the MCB that's tripping?

Almost certainly the magnetic part for 0.4s (it'll actually disconnect in <<0.1s so the thermal part won't get the chance to warm up by much).

For 5s it's also the magnetic part for types B & C - but for D types, currents that will trip the thermal part in 5s wouldn't guarantee  tripping the magnetic part, so a lower fault currents (higher Zs) are acceptable for 5s compared with 0.4s.

   - Andy.