This discussion is locked.
You cannot post a reply to this discussion. If you have a question start a new discussion

DC on AC supply

I have seen a number of videos which demonstrate the problems caused by DC currents on AC supplies with respect to RCD's. An issue which is apparently increasing with increased technology applications.  Please could someone explain how DC gets back into the neutral.


Almost all PSU's in all equipment begin with an isolating transformer. So getting a significant pulsed "dc" signal back that way is not impossible but tricky. (Pulses only occurring on one side of the cycle). 


My thoughts are that the DC is most often "created" by an unbalanced waveform, with a muted positive cycle, giving an overall DC flow.


Please could you direct me to, or give examples of how DC gets into the neutral, and/or just how much of an issue it can be in some homes.
  • It would take a long time to try to answer thoroughly, unless restricting the scope. For example, is it the L-N loop (you say 'into the neutral'), or the L-PE (or N-PE) as would be more important for the influence on an RCD? And is only pulsed-dc (e.g. like half-wave rectified ac) what's meant, or steady dc? 


    Input stages of PSUs in modern equipment like a computer or charger do have isolation, but practically never directly on the input any more. Older equipment started with a transformer, then a rectifier and voltage regulator.  (And even older equipment started with the rectifier, then the valves, then the isolation, if any, was on the output.) Modern electronic equipment starts with a rectifier, which might in bigger items be a cleverer type than just diodes in order to reduce harmonics and phase-shift in the input current. Then there's more electronics to switch the resulting ~300V dc quickly into a high-frequency transformer, from which the output is rectified. The advantage is that a high frequency transformer can be much smaller and therefore much cheaper than a 50 Hz one.  That's why modern wall-warts are so light compared to old ones that started with a bulky transformer.  As you imply, it also gives the potential for unbalanced positive and negative parts of current, although most devices try to be balanced.


    Some loads can deliberately inject several amps of pulsed dc from L to N. Typically these are hairdryers and electric blankets (ok, not multiple amps for the blankets), with a design where a single diode is put in series for the half-heat setting; I've come across cheapo usb-chargers that also have just one diode. This is permitted in EN_61000-3-2 if the power is below 100 W or the device has only two wires and is intended for short use (a few minutes).  Other loads can unintentionally do a less extreme case if their positive and negative half-cycles aren't balanced. I don't see a problem for RCDs unless there's a fault that causes this current to return through PE instead of N, with a type AC RCD.


    I have briefly studied claims of RCD+dc problems in a related matter where different organizations disagreed about the basic principles and the risk. One point l I learned was to (largely) ignore videos. Some example cases were given by references to youtube. Generally there were clear lacks, like measuring only in a place that doesn't confirm the current that actually matters, or with equipment that can't be expected to give useful results. I stopped reacting to them, since they mainly wasted many minutes without firm, useful information. However, if it's a simpler question like L-N currents with a non-zero average then it should be hard to measure wrongly. Just don't trust too much about 'proving' the effects on an RCD unless the setup and measurements are very clear. 

    I hope something in the above will be useful for you.

  • There's also a particular issue with electric vehicle charge points as they use d.c. signalling between a pilot wire and PE to communicate between the charger and the vehicle (for things like agreeing the charge rate and detecting whether the c.p.c is intact) - a simple pilot wire to N fault (plausible as they run in the same flex and terminate into the same wiring enclosures) risks running d.c. around the N-PE loop via the supply's N-PE link (and hence through the N coil of any RCDs in the circuit).

       - Andy.
  • A great many devices rectify the mains before use and do not use a transformer, though most are full wave rectifiers ~(bridges) but can create a pulsed DC fault if one diode fails or there is a short from on side of the output of the rectifier to earth or neutral.

    Some smoke alarms and emergency lights have transformerless power supplies

    Other devices have the potential to rectify because they use a triac, that may not be triggering equally on both half cycles.

    Several designs of PIR sensor, certain types of motor speed control and light dimmers come into this category.

    All of this is not new, and rectified mains faults are quite rare.

    Mike.
  • A number of cheap domestic appliances draw DC current from AC mains. Hair dryers, hair straighteners, and small cooking appliances. They use full mains onto the element for full power, and mains via a silicon power diode onto the element for half power.

    Older valve TV sets draw DC current from AC mains. The valve heaters are wired in a series chain and connected to the mains via a silicon power diode.

    Valve radios usualy obtain the HT supply from half wave rectification of the mains.

    Older traffic lights and Belisha beacons draw DC current from AC mains at night. During the day the lamps use full mains, at night they are dimmed to reduce glare and prolong lamp life. The dimming is achieved by switching a diode in series with the lamps, a slight line frequency flicker is observable on the lamps at night.
  • The saving grace for the RCD and ADS generally  is that during any fault of significant current fault the diode or triac, switch more power supply chip  or whatever usually stops diodeing and becomes a near short, as the semiconductor gets hot and all the carefully placed doping chemicals diffuse and mix together so 'P' and 'N' regions are no longer discernible.

    So, rather like all small children's paint mixing experiments result in brown, and all Scout's cooking looks burnt, all overheated semiconductors, even expensive computing ones, tend towards a rather mediocre low value  resistor. Of course in a high current fault the metallisation gets blasted off the chip surface or eventually  the bonding leads just evaporate, so above a certain I2t it does its own ADS.


    Also unless there is a smoothing capacitor, any fault  waveform will trip  the RCD type A kind, and many type ACs not anything more expensive.

    (and if the smoothing cap is electrolytic, and it nearly always is, then once the diode has gone and it is being subjected to the full AC, you only have a short time to duck before the content of the capacitor follows it. Teenage years of TV repairs and similar leave memories that are too deeply etched to forget.)


    As an aside

    Given that a smoothing cap will hold the voltage up you can make a 3 level heater with a diode and a C, at least for low wattages like soldering irons,

    1) pre--heat,single  diode in series, half power, soldering iron on stand, I have gone for a break

    2)run full waveform (either mains direct or  bridge rectified, heater takes either) indoor normal soldering of small parts

    3)boost full wave rectifier plus smoothing, DC of ~ 330V - ideal for soldering large parts or outdoors...
  • Broadgage, none of those things are "DC mains current", they are distorted mains AC waveforms. Above I gave a definition of DC, and none of these meet it. A half-wave rectified waveform will pass through a transformer and the only thing which does not is a waveform where dV/dt (the rate of change of voltage/current) is zero. An RCD differences two dI/dt waveforms and the resultant is the leakage current. The DC we are talking about here is a real DC current (dI/dt = 0) which is present on only one of the windings, usually the N-E bond and DC-fault in an appliance between some version of the supply and Earth.


    The overall problem with all this is that the "what if" brigade has tried to find a reason to fit more complex and expensive protection for less and less likely faults. A proper analysis says that many of these faults are not dangerous to persons, need at least two faults to occur simultaneously (such as an undetected DC fault and an accidental touch of a phase conductor), and the likelihood is probably similar to being hit by a falling piece of an aeroplane flying overhead. It is much more likely that such faults disconnect for other reasons, and result in appliance failure, which is a nuisance but not dangerous.  If we apply a similar criterion to roads, nothing except slow-walking would be allowed, and certainly no 2 wheeled devices, or use at all if the road was frosty or covered in snow. Just like so many other things, and Covid is one of them, "Authority" has gone too far, with extremely dire consequences for everyone. I for one have not enjoyed my 1-year largely solitary confinement prison sentence, without any visits, and I have to do all my own catering as well! You will notice that the only way to make any of these things "serious" is to exaggerate the risk and use false statistics to make a case. Just how many people are killed on RCD protected circuits by contact with a live conductor and no other co-morbidities? I suggest that the number is probably zero, or possibly 1 from bad luck in the circumstances.
  • A half-wave rectified waveform will pass through a transformer and the only thing which does not is a waveform where dV/dt (the rate of change of voltage/current) is zero.

    It's not just about the transformer in the RCD though. As I understand it some AC designs held the contacts closed with a permanent magnet and released the contacts by using an electromagnet energised from the sense coil to create a cancelling magnetic field. Such designs are naturally more sensitive on one half wave than the other - possibly only ever tripping on one half of the waveform - and can give quite different operating times on 0 and 180 degree tests - but with a full a.c. residual waveform they'll still trip eventually (and within the required times). Give it a half-wave rectified residual waveform though and it might be 50:50 whether it'll trip at all.


      - Andy.
  • You are making another error Andy. The trip signal is not DC referred, it is an alternating current, although the mean depends on the waveform, which is the magnetic field reversal point. A true RMS measurement of this signal will give the trip current, and the waveform is definitely AC, although not a sine wave. It also appears to be thought that a small DC current prevents transformer action completely. This is incorrect, although it will change the exact trip current. A current difference that is larger than the DC present will still give a difference signal, just not as big and possibly not as well shaped as may be expected. I refer you to any basic textbook on magnetism that deals with flux mechanisms. A strong magnet may be made to stop any attraction completely by winding a coil around it and applying sufficient current to cancel the magnetic field. That is the mechanism we are discussing, NOT some magic saturation property, which does not exist.


    I have done a considerable amount of work on RCDs, and strangely enough, they show the properties I have outlined above. This is why I am very skeptical about this DC current alleged problem. There is an extremely low number of scenarios where an appliance can apply just the correct current to cause problems and have no other effect, either on itself or the CPD. Most of the cases shown in BS7671 table A53.1 are seriously flawed, to cause a problem they need a critical resistance in the alleged fault path, and the If shown is not anything like a short circuit. I have attempted to draw out this from several sources and they cannot define any realistic fault which is shown in A53.1 which will not trip a correctly rated CPD, either in the appliance or in the supply circuit. It is true that a fault in the correct value range can increase the trip time or protection current, but how it can happen is never defined. The danger area, if any, is very low power devices, which are inevitably class 2, so they cannot cause this variety of single fault. Something is seriously wrong in the State of Denmark! (Shakespeare 1564-1616). He obviously anticipated BS7671.
  • Only if the actuating magnet is in the field of the L and N directly.. Any design that has a ring core and then autotransformer couples the fault waveform to a multi turn coil around the permanent magnet, will see the ripple of the rectified AC, as the transformer action removes the mean DC component.
  • Former Community Member
    0 Former Community Member
    nicemark:

    I have seen a number of videos which demonstrate the problems caused by DC currents on AC supplies with respect to RCD's. An issue which is apparently increasing with increased technology applications.  Please could someone explain how DC gets back into the neutral.


    Almost all PSU's in all equipment begin with an isolating transformer. So getting a significant pulsed "dc" signal back that way is not impossible but tricky. (Pulses only occurring on one side of the cycle). 


    My thoughts are that the DC is most often "created" by an unbalanced waveform, with a muted positive cycle, giving an overall DC flow.


    Please could you direct me to, or give examples of how DC gets into the neutral, and/or just how much of an issue it can be in some homes.


    Welcome to the forum! It is a good first post :)


    What I have wondered with regards this situation is if an installation has equipment that can apparently disable commonly found RCDs, would this disablement also apply to other installations that share a supply, for example with looped supplies?