LED driver earth leakage


I have a situation where LED panel lights appear to have a voltage of roughy 80V between the metalwork of the panel and the containment overhead (which is bonded to earth). 

The LED driver is double insulated, and SELV and a plastic case and remote from the panel. It does stipulate in the user manual that the driver does have earth leakage, but how does this happen when it has no earth conductor, and it’s output is SELV? Is it leaking through the panel despite the fact it’s technically isolated?

The cable between the driver and the panel light is a DC connection, plus and minus, with the panel light also being a class 2 product. 



  • Could it be your digital voltage indicators picking up stray voltages between isolated metal class 2 and earth bonded containment. Digital voltage indicators have a high input impedance,makes them sensitive to stray voltages, which can result in false readings.

  • It's incredibly difficult not to have a small amount of leakage - not to earth but to the incoming supply. Medical grade power supplies manage to it down to a possibly unmeasurable level, but yer average LED supply will have a very slight leakage path from the incoming supply, whether through the interwinding capacitance of the isolating transformer, or even semi-deliberately through EMC suppression capacitors. So anything connected to an "isolated" mains supply will typically float to somewhere around half the supply voltage.

    The question is, how much current is available / what is the source impedance. A meter may measure 80V, but that doesn't mean that anything more than a minute current can flow.

    I'm on my home computer rather than my work computer at the moment, so haven't got access to standards, but from a quick Google (so no promises this is correct!):

    Section 13 “Leakage current and electrical strength at operating temperature” of IEC 60335-1-2020 specifies the following maximum allowable leakage currents for major types of household electrical equipment:

    for class II appliances and for parts of class II construction – 0,35 mA peak;
    for class 0 and class III appliances – 0,7 mA peak;
    for class 0I appliances – 0,5 mA;
    for portable class I appliances – 0,75 mA;
    for stationary class I motor-operated appliances – 3,5 mA;
    for stationary class I heating appliances – 0,75 mA or 0,75 mA per kW rated power input of the appliance with a maximum of 5 mA, whichever is higher.

    Incidentally, when I am on my work laptop, which is a very reputable brand with a metal case, I'm able to feel a distinct "tingle" when I touch the lid when it's plugged in to its SELV supply! It's fine, it's meant to be like that, but probably does measure about 115V to earth. But can't source any real current.

    Of course, if switching your meter to mA shows currents above the above (hopefully followed by a trip tripping!) then the LED supply / transformer is not healthy...



  • It might be worth repeating the measurement with a multimeter (such as a Fluke 114) that has a low input impedance voltage measurment range.

    Most multimeters have in AC voltage input impedance of about 5 MΩ which will cause the problem that AMK describes above.

    The low input impedance range on the Fluke 114 is only about 3 kΩ and solves most problems of this nature.

    No connection with Fluke, other than a user of their instruments.

  • The Guide to Electrical Installations in Medical Locations elucidates the phenomenon of leakage currents, which result from capacitive, inductive, or resistive coupling of electrical devices. It expounds that no insulator is flawless, and thus every electrical device is prone to some leakage current. It also illustrates this with the example of a laptop chassis, which can exhibit up to half the main’s voltage between the accessible conductive parts and earth, when measured with a high impedance voltmeter. However, the guide wisely cautions against attempting this experiment.

  • The latest (5 th Edition) of the IET Code of Practice - In-Service Inspection & Testing of Electrical Equipment has a flat limit of 5 mA for AC leakage current.

    Previous editons of the Code of Practice had leakage current limits similar to, or the same as, the ones you have reproduced from IEC 60335.

  • The LED driver is double insulated, and SELV and a plastic case and remote from the panel. It does stipulate in the user manual that the driver does have earth leakage, but how does this happen when it has no earth conductor, and it’s output is SELV?

    It might be "double insulated" but like a laptop supply require a mains earth (cpc connection) for functional purposes.

    However, there are electronic converters that don't meet the "BS 7671 requirements" (IEC 61140) for SELV or PELV, due to the replacement of BS EN 60950-1 with BS EN 62368-1, and that is why BS 7671:2018+A2:2022 has requirements in some of the Part 7 sections, such as Sections 701, 702, 710, numbered '7xx.414.4.5' saying something along the lines of 'Where SELV or PELV is used <in stated conditions> a source described in Regulation 414.3 (iv) shall not be used.'

  • The driver has no earth connection, only line and neutral, hence the confusion. 

  • Hi Andy, 

    Your laptop PSU has an earth, whereas this LED driver has no earth, only live and neutral.

    Can the same leakage occur from the secondary side of the power supply? 

  • When you measure the 80-volt, is the fixture installed in the ceiling grid?

  • Yes indeed! Could well be even with the lights completely removed from the grid, a voltage is present between the grid and the containment.

    Once had a severe dig when my ear brushed the metal bars of a suspended ceiling grid at the same time as my hand made contact with the metal trunking in the void above. Fell off the step ladders as a result and pulled a fair section of the ceiling down with me. All in front of of a storey full of bemused office workers!