The MCB tripped on its thermal response because the circuit was overloaded.
You need to understand that there are two different classes of over current: overload current, and fault current.
Overloads are caused when a circuit is operating normally, but the total load from the appliances connected exceeds the nominal rating of the circuits protective device. Say for example, drawing 40A from a circuit with a B32 MCB. The effects of this are to cause gradual heating of the circuit wires and accessories. If left unchecked it could result in cable damage or reduced life expectancy of the cables. The heating will take time to reach a damaging level. So MCBs include a trip mechanism based on a bi-metal strip, that heats in a way analogous to that of the rest of the circuit. It will tolerate small overloads for a very long time, and then ever reducing durations as the overload increases. So for some overloads they may run for minutes or hours before tripping.
Fault currents however are a different class of fault. Here you have something causing a short circuit, and the current that flows can be
100s or 1000s of amps. This results in very rapid heating of the circuit wires (in some cases even explosive heating) - there is also no time for this heat to be dissipated to the surroundings (i.e. its adiabatic heating). This class of fault needs the kind of immediate response that the bi-metal strip of the MCB can't provide. Hence it includes the magnetic response that will react with the speed you would expect from a fuse.As you would expect
You made the claim that you were not drawing anything close to 140A. I agree with you, you weren't. I suggest that if however you were to drive a nail through one of the circuit cables (i.e. to introduce a fault) and then measure the current draw, you will see a *significantly* larger current - hopefully only briefly.
Its may just be electrons, but that does not mean you can handle situations where you are drawing 5A too much in the same way you handle those where you are drawing 500A too much.
This is why circuit designers consider both scenarios, and the equipment manufacturers design kit that behaves in an appropriate way to cope with both.
They produced an overload. However the voltage drop you witnessed during this episode does cast doubt on the ability of the circuit to correctly deal with faults.
No it won't happen at 40A or 50A. Those are not fault currents, those are overloads.
You can get an estimate of the time to trip by looking at the response curve:
The vertical bits of the curves represent the magnetic trip - that for dealing with fault currents. The curved bits are the thermal response. If you find 50A on the 32A curve you will see it intersects the time axis at around 1000 secs. Real world conditions may mean you don't see this - but you can be reasonably confident it will ultimately trip, but it will likely take tens of mins to do so. If you are running on a circuit also suffering an undervolt then the times will be longer.
The occurrence of a fault current.
The thermal one.
I can't say with any certainty what you have installed. I have never seen it, or tested it. I can offer only educated guess work. I can say I am suspicious that all is not well based on what you have told us, if what you have told us is correct.
The fact that you have a voltage reduction device in place will make the case more borderline. The installers *should* have checked that the impedances of the circuits already installed were low enough for this to be safely installed. I would not be surprised if this was not done.
There is a limit to the maximum length of cable that can be installed for a circuit. This may have been observed as originally installed. However it is not uncommon for extensions to be made later.