The law has the same problem, many of them are not clear, and are then subject to interpretation by the courts. As we do not wish to involve lawyers or courts any more than is necessary, a clear guide is needed.
I agree that the light dimmer example is a bit poor, but there is a dilemma as to whether to IR test properly or just test between LN linked and Earth. In fact, light dimmers will generally allow Zs testing without problems, but some devices will not, particularly VFDs, etc. which completely reform the mains supply. The same may be said for UPSs, and battery to mains inverters. These will all become more common with "green" supplies, but clearly BS7671 is not equipped to cover them. Power electronic devices have to deal with overload situations because semiconductors are often known as the fastest fuses known to man! Because the active areas are very small, the rate of heating is very high, and heatsinks do not help because that heating is adiabatic, and we all understand that. This means that transient high current must be avoided very quickly indeed, and this is done by switching the power devices off as fast as electronically possible from real-time current measurement. If a fault causes a power device to fail short-circuit then the normal CPD will provide protection to the cables, which should be sufficient for safety.
We need to look at why disconnection time is important and realise that we are not analysing shock protection, it is fire protection which can be much slower. It is true that some shock risk can be found under some circumstances, but these can be removed by equipotential bonding if considered significant. If 5 seconds is considered safe for submains, why is it unsafe for installed motor drives? Really it only matters for Portable appliances which may develop live-to-case faults in circumstances where other Earth potentials may be present, for example, a kettle in a kitchen. Here the use of RCDs as additional protection is a good idea, but is this required in a factory full of machine tools where they are unlikely to work properly? In this case bonding to nearby machines is easy, and very effective. Blanket RCDs everywhere is not the correct Engineering answer to this problem of electronics everywhere. Many modern tools have 6-10 VSDs in the control system, 30mA RCDs are not the answer.
The law has the same problem, many of them are not clear, and are then subject to interpretation by the courts. As we do not wish to involve lawyers or courts any more than is necessary, a clear guide is needed.
I agree that the light dimmer example is a bit poor, but there is a dilemma as to whether to IR test properly or just test between LN linked and Earth. In fact, light dimmers will generally allow Zs testing without problems, but some devices will not, particularly VFDs, etc. which completely reform the mains supply. The same may be said for UPSs, and battery to mains inverters. These will all become more common with "green" supplies, but clearly BS7671 is not equipped to cover them. Power electronic devices have to deal with overload situations because semiconductors are often known as the fastest fuses known to man! Because the active areas are very small, the rate of heating is very high, and heatsinks do not help because that heating is adiabatic, and we all understand that. This means that transient high current must be avoided very quickly indeed, and this is done by switching the power devices off as fast as electronically possible from real-time current measurement. If a fault causes a power device to fail short-circuit then the normal CPD will provide protection to the cables, which should be sufficient for safety.
We need to look at why disconnection time is important and realise that we are not analysing shock protection, it is fire protection which can be much slower. It is true that some shock risk can be found under some circumstances, but these can be removed by equipotential bonding if considered significant. If 5 seconds is considered safe for submains, why is it unsafe for installed motor drives? Really it only matters for Portable appliances which may develop live-to-case faults in circumstances where other Earth potentials may be present, for example, a kettle in a kitchen. Here the use of RCDs as additional protection is a good idea, but is this required in a factory full of machine tools where they are unlikely to work properly? In this case bonding to nearby machines is easy, and very effective. Blanket RCDs everywhere is not the correct Engineering answer to this problem of electronics everywhere. Many modern tools have 6-10 VSDs in the control system, 30mA RCDs are not the answer.
