I wonder whether the switches are failing to break the circuit cleanly and the contact bounce is giving the MCB multiple hits of inrush while the rocker is moving to the off position. I.e. it's not the breaking of the circuit that does it. An AC micro-break switch continues to carry current while switching off until the current-zero at which the arc extinguishes and the circuit opens. It's a bit far-fetched to think that it would then re-make the circuit but if the switch action is really rough it's the best explanation I can think of without more info.
Back-emf is the release of the stored energy in the transformer core that occurs if you catch it at the wrong point on the mains cycle. However although the voltage into an open-circuit can be high, the current should not normally exceed the peak current when energised, so I am sceptical how that could trip the MCB (although it can cause chaos with voltage-sensitive things). OTOH the inrush that occurs when you switch a transformer on at the wrong point in the cycle is caused by magnetic saturation and the current can be very high because the inductance momentarily drops. Hence my looking for some explanation as to how the circuit might be getting re-made while switching off.
Barrage of questions, sorry:
Does the MCB ever trip at switch-on?
Does it always trip at switch-off?
Does the transformers sometimes make an audible thump when switching on or off?
Do all three 'pairs' behave the same?
What type of switches are in use?
How near to 10A is the MCB running?
Is there anything in circuit that might contain a VDR surge suppressor (that the back EMF is kicking into conduction)?
And one final one, if you didn't wire the switches yourself, have you checked that L2 hasn't been used to joint the neutral?
