"17 edition"consumer units still being sold.

Thanks (the mention of regulations for type-B).  That type, I know, is sometimes required. My particular interest is more whether the latest regulations put hard requirements on using A instead of AC in almost every situation nowadays.


"DC":  this causes plenty of confusion, as people use different meanings. One is pure DC: a constant value over multiple AC cycles, possibly arising 'gently' so as not to cause any trip. Another is that the mean value isn't zero, or (more extremely) that it's always in one direction like a half-wave rectified current, which is a form of "pulsating DC" in the words of RCD standards. 


The distinction between type-AC and type-A is about responding to pulsating DC such as half-wave rectified current.  Pure DC is not the issue: it could affect either of these types according to the standard, as they're only required to tolerate the 6 mA steady DC level.   Handling even pure DC is the feature of type-B.


It is not reliably true that a type-AC will operate with half-wave rectified current (pulsating DC) even if this is made, say, twice the normal rated value.  It might operate.  Or it might not operate even with multiple amps flowing. It depends on the design.  Don't be too confident based on an example measurement.  I've seen more variation between type-AC than type-A.  In the best case, a type-AC basically seems to be type-A under another label.  In the worst case a 30 mA type-AC can permit amps of "pulsating DC" without tripping, at least in one direction.  It depends on the design details, which can vary a lot within the bounds placed by the standard. See the final paragraph.


The non-tripping of type-AC with half-wave rectified current is worrying because: tripping can fail even with high currents; and the type of fault is not at all so improbable (in my estimation) as some others that could make pure DC; and UK systems sometimes rely on the RCD for ADS.  I consider that worse than the type-A vs type-B issue, since a type-A with big enough AC fault current (compared to any pure DC current) does trip, according to all my tests; being made less sensitive is not desirable, but it's better than being anaesthetised.


A TT earth electrode in a system with 30 mA type-AC RCD could well have a resistance of 20 ohms.  That would limit the current through a L-PE fault in a diode to be a value that the diode in a larger device could tolerate and that wouldn't trip an OCPD.  One could say similarly for a diode in a smaller device with a higher electrode resistance, e.g. 80 ohms (still plausible).  But the potential on all the PE relative to (real)earth would be half-wave ~230 V, unless source earthing were also weak.  As said above, some type-AC RCDs would actually respond to this, but some wouldn't: they're not required to.  I consider this a higher risk than some other dangers for which special RCDs and other special measures are pushed.  The good reliability of many appliances is one saving grace: having any fault, let alone through a diode, is very unusual.


In case anyone wants to understand how a 30 mA RCD (type AC) can manage fail to operate even on e.g. 30x  I_{\\delta n} of half-wave rectified current, here's a quick explanation.  Take a current-transformer with magnetically hard material (cheap), as is typical of a true type-AC RCD.  Start applying the current.  As the current moves away from zero, the core becomes magnetised in this direction, and likely goes into saturation if the residual current is well above its rating. The non-saturated relative permeability of RCD cores is pretty high - tens of thousands - so losing most of this during saturation means there's relatively negligible change of flux. And when the current comes down near to zero again, the hardness of the core prevents the flux going down much either, when there isn't any reversal of the current (as with real AC) to push the magnetization the other way.  So the change of flux after that first few milliseconds is really very small, whereas with full-wave AC it would be much more.  On the first edge of the current away from zero there's a chance of tripping as the flux goes between zero and saturation, but usually only for one direction of the current!  That's because the sensitive trip-mechanism of non-powered RCDs normally uses a permanent magnet to hold a magnetic circuit closed, and a coil to demagnetise this magnetic circuit: one direction of the induced current demagnetises, and the other doesn't.  (That's why, on type-A too, you can often see a ~10 ms change in trip time when varying between 0 and 180 degrees. With some types of powered RCDs this dependence can be avoided.)   I've recently started a set of measurements on RCDs that I've had planned for some months: as promised once before, I will soon put up a webpage of results, along with the actual device models.

Thanks (the mention of regulations for type-B).  That type, I know, is sometimes required. My particular interest is more whether the latest regulations put hard requirements on using A instead of AC in almost every situation nowadays.


"DC":  this causes plenty of confusion, as people use different meanings. One is pure DC: a constant value over multiple AC cycles, possibly arising 'gently' so as not to cause any trip. Another is that the mean value isn't zero, or (more extremely) that it's always in one direction like a half-wave rectified current, which is a form of "pulsating DC" in the words of RCD standards. 


The distinction between type-AC and type-A is about responding to pulsating DC such as half-wave rectified current.  Pure DC is not the issue: it could affect either of these types according to the standard, as they're only required to tolerate the 6 mA steady DC level.   Handling even pure DC is the feature of type-B.


It is not reliably true that a type-AC will operate with half-wave rectified current (pulsating DC) even if this is made, say, twice the normal rated value.  It might operate.  Or it might not operate even with multiple amps flowing. It depends on the design.  Don't be too confident based on an example measurement.  I've seen more variation between type-AC than type-A.  In the best case, a type-AC basically seems to be type-A under another label.  In the worst case a 30 mA type-AC can permit amps of "pulsating DC" without tripping, at least in one direction.  It depends on the design details, which can vary a lot within the bounds placed by the standard. See the final paragraph.


The non-tripping of type-AC with half-wave rectified current is worrying because: tripping can fail even with high currents; and the type of fault is not at all so improbable (in my estimation) as some others that could make pure DC; and UK systems sometimes rely on the RCD for ADS.  I consider that worse than the type-A vs type-B issue, since a type-A with big enough AC fault current (compared to any pure DC current) does trip, according to all my tests; being made less sensitive is not desirable, but it's better than being anaesthetised.


A TT earth electrode in a system with 30 mA type-AC RCD could well have a resistance of 20 ohms.  That would limit the current through a L-PE fault in a diode to be a value that the diode in a larger device could tolerate and that wouldn't trip an OCPD.  One could say similarly for a diode in a smaller device with a higher electrode resistance, e.g. 80 ohms (still plausible).  But the potential on all the PE relative to (real)earth would be half-wave ~230 V, unless source earthing were also weak.  As said above, some type-AC RCDs would actually respond to this, but some wouldn't: they're not required to.  I consider this a higher risk than some other dangers for which special RCDs and other special measures are pushed.  The good reliability of many appliances is one saving grace: having any fault, let alone through a diode, is very unusual.


In case anyone wants to understand how a 30 mA RCD (type AC) can manage fail to operate even on e.g. 30x  I_{\\delta n} of half-wave rectified current, here's a quick explanation.  Take a current-transformer with magnetically hard material (cheap), as is typical of a true type-AC RCD.  Start applying the current.  As the current moves away from zero, the core becomes magnetised in this direction, and likely goes into saturation if the residual current is well above its rating. The non-saturated relative permeability of RCD cores is pretty high - tens of thousands - so losing most of this during saturation means there's relatively negligible change of flux. And when the current comes down near to zero again, the hardness of the core prevents the flux going down much either, when there isn't any reversal of the current (as with real AC) to push the magnetization the other way.  So the change of flux after that first few milliseconds is really very small, whereas with full-wave AC it would be much more.  On the first edge of the current away from zero there's a chance of tripping as the flux goes between zero and saturation, but usually only for one direction of the current!  That's because the sensitive trip-mechanism of non-powered RCDs normally uses a permanent magnet to hold a magnetic circuit closed, and a coil to demagnetise this magnetic circuit: one direction of the induced current demagnetises, and the other doesn't.  (That's why, on type-A too, you can often see a ~10 ms change in trip time when varying between 0 and 180 degrees. With some types of powered RCDs this dependence can be avoided.)   I've recently started a set of measurements on RCDs that I've had planned for some months: as promised once before, I will soon put up a webpage of results, along with the actual device models.