Type A rcd . EICR coding ? etc

What I found interesting about the Blakley paper was that the problems with the RCD weren't necessarily related to the load on the circuit the RCD protected. If we had d.c. flowing around our earthing system for any reason then any circuit could be compromised as soon as it developed an earth fault.


I can't help wondering if things like the d.c. pilot, communication and earth monitoring gubbins on EVSE might be a worry there.

 

TT installs are a rather special case, and only require a single fault to be dangerous

I agree it's less likely to be an issue on TN systems - the rectifier arrangement is likely to either limit the fault current to something relatively modest for an earth fault (perhaps a couple of tens of amps) or blow itself to smithereens in short order (ADS after a fashion) - which with a EFLI in the region of an Ohm or so isn't likely to produce a hazardous voltage difference - but still probably not ideal to let it persist. TT will be quite different though - such currents could raise the entire earthing system to very hazardous levels so you really don't want the one device that providing both ADS and supplementary protection to be disabled. I think we have enough TT installations even in the UK for it to be a worry (especially if we've been busy creating new ones for EV charge points). Most manufacturers and of course ENs cover regions where TT is the norm (France and southern Europe for example) so I can see why it's being raised.


   - Andy.

What I find most worrying with type AC is that there is no guarantee that it will even trip for half-wave rectified current, i.e. a fault via a single diode. That's not a strange type of fault.  I have tested some RCDs that are marked AC yet behave largely as A. But I've also met AC ones that will not trip even for many times the rated residual operating current when it's half-wave rectified in one direction: the other direction will typically work (I've a feeling I already explained the reason for this sometime in the spring). Type AC is not required (by IEC6100[89]) to trip for any half-wave waveform.


Given that many loads contain diodes connected to their inputs, I think it's very desirable to protect against accidental contact with such parts or against earth faults from such parts in TT systems.  A plausible range of earthing resistance in a TT system could result in such a fault continuing without tripping an RCD or breaking the diode.


If it cost a lot more to be sensitive to half-wave rectified AC, there'd be a justification for considering more of the cost/benefit of what harms are caused by earth-fault-through-diode situations.  However, the cost difference is negligible between AC and A ... this claim will probably raise some eyebrows, so let me explain more: when regulations are sloppy, as previously in the UK, almost everyone does the cheapest thing, so ready-made CUs come with AC, retailers stock mainly or only AC, and A becomes a more specialised device with a higher price-tag (because they can, or because the admin is divided between a few sales, or whatever). If none of type AC are sold, type A becomes practically the same price as AC was.  The difference is basically in the magnetic material used. I made a study of the prices a year ago, comparing prices in the UK and France (which widely use AC) with Germany and Sweden (which don't use AC).  There's some difficulty in comparison due to the fact that one bloc would commonly use 2pole RCDs of higher current rating, and the other would commonly switch to 4pole for loads above 16A or so.  But basically, you can get an AC or A RCD for around the £25 mark.  More recently I see many RCDs in the UK becoming type A, such as some BG(manufacturer) ones I bought recently, without increased price from the older AC models that have quietly been replaced. 


So, I maintain that among all the safety choices in recent regulations, getting protection from through-diode earth faults for negligible extra cost is pretty much top of the list, compared to more sophisticated RCD types, EV worries (UK specific), and let alone AFDDs!  However, I take the point that there might not even be a demonstrable case of harm that would have been avoided by type A instead of AC. I'd be surprised, but I can't point out a case (anyone?).


As indicated above, it is not the case that all of Europe is happy with type AC.  On the contrary, Germany doesn't recognise them for

 RCD protection of anything. Sweden doesn't have them to find on sale. (It's a puzzling distribution, where countries more known for using TT systems, e.g. France, appear less averse to type AC.)


Moving from type AC was the main thing I thought good in the recent proposal.

Those are reasonable points Nathaniel, but there are wider issues. First, is what happens to present installations? Type AC is perfectly safe in 99% of cases, yet I can see many code C1s generated if the type is simply removed from the regulations. If manufacturers are changing the type anyway (and this may be because type A is widely used elsewhere, and I'm not sure that most are not of this type already although marked as AC) because it is not a requirement not to respond to unipolar (DC) faults. MFTs give one a setting to change the fault phase from positive cycle first to negative first, which often changes the trip time. Why this change happens is not very clear, and maybe a type AC phenomenon, but adding a new test to give a rectified waveform of either polarity is not difficult. In fact, I will try this next time I have my RCD testing jig out as the result would be interesting, and I will report back. There is very little electronics that uses half-wave rectification, even low power items, and LED light bulbs seem to have full-wave rectifiers at the few watts level. There is also a case that Earth faults inside electronic items are very unlikely, because of the insulation and creep distance required from mains components, and failure of the EMC filter capacitors can never be DC (rectified AC or whatever). Whilst a rectified type of fault current cannot be totally discounted, it does seem very unlikely. Changing every type AC for a type A, as suddenly being dangerous is clearly completely unacceptable to customers, as in many cases we would end up with many more CU changes for unreasonable reasons, for example, a type-tested type A is not available for the existing CU.

Inside the type A RCD, the core may be larger, and the peak detector may be more symmetrical in its response to positive and negative peaks, but in all other respects there is not a lot different inside, certainly not much to justify a large price hike.

I too have done some informal testing, but it is far from comprehensive - I can see the wisdom in recomending installing only  type A in future, but I share the view that the existing AC types will not suddenly rise up and strangle the user, indeed I think the majority, if not all  of them could safely be rebadged as A with no other changes, just that performance was never tested and cannot sensibly be guaranteed, and some designs as type A may be a bit more insensitive.


The hard one is the smooth DC, but even there some limits can be set - a a DC small compared to the test limit of 30mA will have a small effect, and in general a desensitised RCD will still always trip once the AC amplitude exceeds the blocking DC, what is hard to say is if it will trip before that .


As solar panels and EV batteries and so on proliferate, there is more scope for odd faults the put a true DC into the system, but these need managing case by case.

regards

Mike

True - I avoided the wider issues of already-installed RCDs in my earlier comment. This admittedly is the thread's main purpose: how to handle existing type-AC in inspection reports. But I think I should leave it to others, as I don't have much experience of these codes or their implications to installation owners. I do think that a lot of tolerance should be given to installations that are according to the rules of their day, unless there's good reason to suppose them dangerous due to statistics or to knowledge about new use-situations (such as the very different loads).  My main reason for posting was earlier comments about Europe and about whether preventing type AC for new installations is a bad idea. Now, thanks to interesting comments (Dave and Mike), I'm stimulated to give some description of general AC/A RCD features that might interest a few people here and be good background for your tests.


Yes, I agree that in the majority of cases the type AC would do the job, since earth faults (including accidental contact) would typically not have a diode present in the loop.  However, note that it's not just devices with a single-diode rectifier that could give a half-wave rectified residual current. A bridge rectifier can do this too, from either of its dc outputs. So half-wave rectified residual currents can arise from the wires and components that come after a rectifier inside an appliance, if these touch the frame or if a person directly contacts such parts. See the residual ('fault') component of current in cases 3,4,5,(&7) in the figure that Andy J posted earlier in this thread.


The RCD standards just specify tests that must be passed, without demanding how the device is to be implemented. However, the design of any moden, passive ('voltage independent') RCD of types A or AC that I've seen follows much the same pattern. An explanation of this design makes clear a lot about their behaviour and the polarity dependence, which was discussed in earlier comments. I'll try to give such an explanation below. The focus is on passive RCDs, known in the standards as 'voltage independent', meaning that they'll be able to trip even if there's no voltage between the conductors. Most RCCBs in Europe are of this type, although many recent RCBOs aren't.


-- Types A and AC


It's the trip mechanism that is polarity dependent. In order to trip with the very small power available from a small current-transformer, these mechanisms are not like normal relays or solenoids in which iron parts move in a magnetic circuit that includes air-gaps. Instead, they have a closed magnetic circuit, where a small permanent magnet holds the magnetic parts together against a spring. A coil of fine wire is wound around this circuit: a current through the coil in one direction will just add to the permanent magnet's field, but in the other direction it will counteract this field, and with enough current the spring will be able to separate the iron parts. I haven't seen any exception to this design in modern voltage-independent RCDs of types A or AC. The difference in trip-time when tripping on full-wave ac starting at 0 versus 180 degrees (affecting types A and AC, but typically with more delay for AC as can be inferred below) is a result of the polarity dependence of the tripping mechanism.


An RCD of type A needs a current transformer with a soft magnetic material. Consider a half-wave rectified residual current, i.e. pulses in one direction. In the good case, the first half of the half-wave pulse, rising from zero to peak residual current, induces a current in the trip coil in the right direction for tripping, so the RCD trips after whatever further delays the mechanical parts have. In the bad case, the first half of the half wave induces the 'wrong' (non-tripping) direction of current in the tripping coil. However, when the residual current returns to zero the magnetization of the soft material will also return largely to zero, inducing a current in the opposite direction in the tripping coil, so it still can trip.


The classic type AC has a much harder magnetic material in its current transformer. It's then mainly the first rising edge of a half-wave residual current that has a chance of causing a trip. That only happens if the residual current is in the right direction. After the first peak of residual current, the core will remain strongly magnetised until a residual current flows in the opposite direction. It doesn't help that the residual current just goes down to zero, as the magnetization remains. If the half-wave residual current builds up slowly, e.g. by moisture accumulation, the classic type-AC RCD would not trip in either polarity. However, as has been pointed out earlier in the thread, not all RCDs marked as AC nowadays have this classic behaviour - although some new ones I've tried in the last year still do.


Really old RCDs were big and heavy compared to modern ones (even passive modern ones). They used a big core in the current transformer in order to get enough power to operate the trip mechanism. Modern ones have not only a more sensitive tripping mechanism, but also very high-permeability current-transformers: relative permeabilities are tens of thousands. I don't know the history of these materials, but can imagine that some decades ago it might have been more difficult to obtain cores with the combination of such high permeability and magnetic softness, at reasonable price. Modern type A RCDs and their prices (when not seen as specialized devices) show us that it's now possible.


Regarding the size of magnetic cores: note that making a classic type-AC core bigger does very little to help make it more like type A. It's the material's nature that's important: it should be soft enough to give significant contribution to induced voltages during the return-to-zero of residual current. The 'pulsating-dc' issue, of AC versus A, is more about hysteresis than saturation. Saturation is relevant to steady dc preventing other signals being noticed - and there we move to thinking of type B, so let's stop...  It should also be noted that the input residual current is like a current source applied to the transformer, rather than a voltage source. Increasing an inductor's core area in this case doesn't much help to prevent saturation. (A further subtlety is the current in the secondary: with a varying residual current the induced secondary current can partly counteract the primary current, whereas steady dc gives no induced current after a while, in which case the entire primary residual current is used on magnetizing the core.)


Another feature of voltage-independent RCDs is that capacitors are often added in the trip-circuit, to increase the current through it (by resonance). This makes the device less sensitive to other frequencies than the intended one. The final common component is shunt diodes to limit the voltage applied to the tripping coil in the event of a large residual current.



-- Voltage dependent devices


Many small modern RCDs (typically the ones implemented in 1-module RCBOs) are 'voltage dependent'. They use the supply voltage to amplify the weak signal from a small current-transformer and to drive the tripping coil. I remember Dave described this a few months ago in a discussion of the power demand of such devices: a tripping coil can be in series with the electronics, across the 230V terminals. In these, the tripping coil can be a simple solenoid that pulls a magnetic 'slug', independent of polarity, as the power available to drive the coil is much more than in a voltage-independent RCD. I have seen one case where the electronic part was just a thyristor (not triac) supplied from the current transformer, so it anyway became polarity dependent, i.e. the tripping time varied between 0 & 180 degree settings of a tester). The use of a triac should allow a voltage-dependent RCD to be independent of polarity. That doesn't necessarily mean it's type A, as a gradually increased half-wave could still fail to trip in either direction if the core had a hard material. However, the demand on the core material's permeability and size is much less when its output is amplified, so it's easy to have a suitable material. Type A was already becoming common in voltage-dependent RCBOs in the UK in the past several years.


The traditional worry with a voltage-dependent RCD is if e.g. the neutral of a two-pole device is broken before the device. Then, with no voltage between conductors to drive the tripping circuit, it would be possible for large earth-fault currents to flow without a trip. In some regions the 'functional earth' connection has been used in order to have a chance to operate from a bit of current between a phase and this earth. That's not much use in a TN-C-S system unless the break is very close to the RCD, but it's helpful for other systems. TT is of course the one where the biggest danger exists from RCD non-operation. In other regions the functional earth is not popular, as the combination of neutral break and earth fault is considered too unlikely to warrant the hassle of an extra wire and troubles with insulation-testing.


Reliability of electronic components has been another criticism of the voltage-dependent devices, but the US and friends have been at it a long time with their 6 mA devices, and the reliability of very sensitive mechanical tripping mechanisms in voltage-independent RCDs leaves no strong case for the electronic choice being necessarily worse in reliability. I've found cute little baby spiders and web inside a tripping mechanism, stopping it from working.


Another issue of powered (voltage-dependent) devices lies behind Victoria's ban on some RCBO models that nevertheless fulfil the IEC requirements. The problem typically arises if supply and load sides are not connected as such, in which case some models can continue to power the tripping coil after tripping, resulting in its burning out so the device can't trip later. More about this prohibition here. I experienced this when first using the test-button of a via-ebay-imported 10 mA RCCB, before I read of the Australian experience. Mine had the clever safety feature of displaying flashing light and smoke to warn me that it probably wouldn't be on duty again.


--

That is a very good synopsys of RCD design Nathaniel. In reading it I hope that many of you find it rather primitive and that the specification seems to have been written after the device is made to work, rather than before as might be expected. What alternatives do we have to a differential current transformer based design? If we really want DC sensitivity, a transformer is difficult, so we need to consider other DC connected measuring methods. One would be a Hall sensor in the magnetic circuit of the differential core, which would measure both AC and DC differential current. Obviously this would need some amplification and level detection, but at the expense of a very small current would be easy enough. It would be important that the core could not become permanantly magnetised by a large differential fault current, something which does seem to happen to standard RCDs sometimes. A big enough core of hard magnetic material should be able to sort that problem. There is little reason for this to cost any more than a complex mechanical arrangement, and we could control the tripping characteristic fairly easily. It could be faster or slower, have adjustable trip current, and have a big a solenoid driven by a semiconductor  to ensure that "sticking" was very unlikely. This must be more reliable than a mechanical version, a million hour MTBF would not be unlikely (about 100 years).

My view is that no coding should be applied so long as the tripping current of the RCD regardless of type, is matched to the purpose it is intended to fulfill

Better a Type A/C than no RCD at all..

so we need to consider other DC connected measuring methods


Yes - we can be confident that any method that should detect milliamps of imbalance against hundreds or thousands of amps of load current will rely on the magnetic field caused by a balance of currents. The usual method to detect this field, e.g. in a type B RCD, is a magnetic current-balance core that has a circuit stimulating it with weak ac and using the relation of current to induced voltage to determine the dc level based on the different magnetic behaviour at different dc levels. 

If there is a fault of negligible impedance just downstream of a rectifier, which is close to the supply of a piece of equipment, wouldn't ADS operate?


In a system where overcurrent protection is enough for ADS, i.e. probably a TN* system, then yes; or else the diode in question might explode first. But, in a TT system with perfectly plausible earthing resistance, e.g. 50 ohm or 300 ohm, a direct L-PE fault through a diode could continue to supply a moderate but easily lethal fault current, and the entire installation's earthing would be near a half-wave 230 V potential. TT systems are surely a minority in the UK now (statistics anyone?) but that's where the biggest concerns about RCDs lie - ADS by TN* methods has very nice features for reliability regardless of waveform, and for limitation of voltage. A high resistance half-wave fault (e.g. condensation) would not give such a problem, as typical TT earthing could still hold the potential well down. 

This begs the question, "What is additional protection for?"


I don't know if there's an official view of which cases are aimed at in the standards. The additional protection provided by a good RCD could give a person a very good chance of being ok even if (1) making direct contact with a live part in a broken flex, plug, etc; (2) making indirect contact via exposed conductive part whose protective conductor has broken; or (3) fiddling around inside an appliance; or (4) exposed to a fault like a bad quality usb charger where the output has become connected to the mains by condensation or other contaminant or a faulted component. 

      In case 1, contacting a cut neutral coming back from an appliance with single-diode rectifier could produce a current that type AC wouldn't respond to, but a bridge-type rectifier in the appliance would give bidirectional current on the ac side, which would be ok with type AC.  The single-diode case is unusual, although it is found in many hairdryers and electric blankets to give the half-heat setting. So additional protection in case 1 is generally ok with type AC: it seems like a very small subset of all case-1 situations where this type would not work. 

      All the other cases, 2--4, involve currents coming from inside loads, so potentially coming from the 'dc side' of rectification, where half-wave currents can arise. Internal faults from dc-side parts to frames are not something I can say I've seen: good equipment makes this very difficult, with  classic heater elements in cookers still being the main cause of appliance faults that I see making significant leakage. I agree with Dave that most of the earth faults that occur are probably plain ac.  My bigger worry is bad equipment, such as a dodgy-imported USB charger with near-zero clearance of input and output parts and with unsuitable capacitors bridging these; or this 'camping lamp' here (notice the diode connection). 

      For myself I wouldn't touch type AC, as I don't like to compromise on protecting from several plausible types of danger for little or no reduction in cost. But if advising others I'd not be confident to justify the parts and labour cost for changing existing devices just to protect against unlikely combinations of events that appear to pose far less risk to life than most people's car use, recreational activities, etc, particularly if the people are sensible with what electrical products they use and buy.  For new installations, no question... 

It sounds as though an IET official view should be provided on the right code for a legacy type-AC in different situations, in order to avoid widely different codes being given by different people.