Electromagnetic Interference For Fun And Profit

There was an urban legend back in the days of mechanical electricity meters, that there were “lucky” appliances that once plugged in would make the meter go backwards. It probably has its origin in the interaction between a strongly capacitive load and the inductance of the coils in the meter but remains largely apocryphal for the average home user. That’s not to say that a meter can’t be fooled into doing strange things though, as a team at the University of Twente have demonstrated by sending some more modern meters running backwards. How have they performed this miracle? Electromagnetic interference from a dimmer switch.

Reading the paper (PDF link) it becomes apparent that this behavior is the result of the dimmer switch having the ability to move the phase of the current pulse with respect to the voltage cycle. AC dimmers are old hat in 2021, but for those unfamiliar with their operation they work by switching themselves on only for a portion of the mains cycle. The cycle time is varied by the dimming control. Thus the time between the mains zero-crossing point and their turn-on point is equivalent to a phase shift of the current waveform. Since electricity meters depend heavily upon this phase relationship, their performance can be tuned. Perhaps European stores will now brace themselves for a run on dimmer switches.

If you’re curious about these old-style dimmers, take a look at some of their basic functionality.

Thanks [Dorus] for the tip.

24 thoughts on “Electromagnetic Interference For Fun And Profit

  1. The paper does not say much about the nature of the EM fields affecting the power meters. The ‘researchers’ did not measure conducted and radiated emissions per the methods in the EMC standards scoped for this equipment.

    I am not aware of any certified ‘static’ power meter being currently manufactured that does not have reverse current flow detection. I have seen some cheap (and inaccurate) stuff out of China that can been screwed with, but otherwise, the power meters I have tested know when something goes awry at the consumer’s site.

    The paper talks like non-linear power consumption by appliances is a new thing. For most of this century, most stuff has had to meet IEC61000-3-2, and other such standards, which means the power supply will employ PFC, which means these types of current waveforms should not be seen, which means they were testing non-conforming equipment.

    1. IEC61000-3-2 doesn’t outline any requirements for PFC in power supplies to be fair.

      All the IEC61000-3-2 standard outlines is the mains harmonic noise of the appliance, and what levels it isn’t allowed to exceed. A bit of mains filtering is in a lot of cases sufficient for the appliance to pass. All though, there is also some types of appliances that are except from the IEC61000-3-2 standard as well.

      So you jumped a bit too far ahead with: “which means these types of current waveforms should not be seen, which means they were testing non-conforming equipment.”

      I am however not stating that the test did use only conforming equipment, but rather that the IEC61000-3-2 standard has exceptions, and also only covers the low frequency part of EMI. (up to the 40th harmonic, and for 60Hz mains, that is only 2.4 kHz, a fairly low frequency to be fair, considering how for an example most SMPS units works well above 10x that frequency. But that is obviously a totally different category of noise that is covered by other standards and regulations.)

      Then there is the statement about Power Factor Correction, it is actually not that common for most appliances. In a typical home, there might only be a few appliances with active PFC. The most typical is a desktop computer with a “quality” power supply. Even power hungry appliances like an induction stove top doesn’t have PFC, these typically just ride the mains sine wave instead, same for a lot of LED lights (creating flickering as a side effect). Not to mention phone chargers and other “small” power supplies and appliances, these rarely if ever have any PFC features to speak of. And some IoT stuff just uses a capacitive dropper instead, with a nice power factor that is typically bellow 0.2

      But yet again, an induction stove top riding the mains can produce minimal harmonics, and its switching noise can “easily” be filtered away. Same for a capacitive dropper power some IoT device, the gulps of current through its bridge rectifier can be filtered away as well (or one can add a fair bit of series resistance, works somewhat “decently”), and this is also applicable to switch mode and linear supplies as well.

      But a triac based dimmer might have a harder time to just filter away its issues. And these dimmers have been common place for decades, so it wouldn’t surprise me in the slightest if they have some exception from these sorts of regulations. For better or worse.

    2. PFC is not required for powers under 100 watts I think. Certainly, both my grandparents 32 inch LG LED LCDs have power supplies without PFC. I had to take both apart, the older one had busted backlight LEDs and the newer one blew a high voltage capacitor on the primary side just out of warranty.

    3. The article states that the power meters are all IEC compliant. The load contains a common dimmer, which typically does not come with PFC.

      The problem is not that there is no reverse current detection, but rather that a current peak with a particular phase angle causes false measurements. This works both ways btw, the researchers could also have the meter show too much usage.

      The conclusion drawn is that meters compliant with the relevant IEC standard are not guaranteed to measure correctly.

      1. At least one of the meter model manufactures shown does not indicate conformance to the EMC directive. Another has a D of C, but no reference to any harmonized standards in the EMC or LVD directives. Some may be, but am not able to determine for all imaged devices.

        Anyone else have any regulatory information on this stuff?

      1. Load does not make it “appear to run faster” but actually makes it run faster. He’s talking about things that might make the load look like it is bigger than it actually is.

        1. I believe my inductive hot playe may have nade my power neter triple its actual use. Im checking on it just now with Twente uni email. I could be wrong but its definitely possible according to some information there..

  2. There are a fair number of bits of weak science in the paper – the situation where the meter is recording negative consumption corresponds to where the THD of the supply is well in excess of 5% – given that they talk of crest factors between 48 and 42, not a surprise. They also fail to consider the differences between a live supply and the output of an AC power source – especially in terms of transient behaviour and line impedance which will have significant impact particularly on higher harmonics.

    It also fails to consider the differences in behaviour between a traditional meter and a modern smart meter designed to meter in four quadrants – in which case the highly non-linear load could appear to be back-feeding into the supply – I note they have failed to calculate the effective impedance of the load. A dominant negative impedance will appear to be supplying energy to the network.

    The other thing that is not considered is the EMC compliance of the switch-dimmer in question and the impact of a non-compliant unit when it’s load represents a significant proportion of the load – particularly in light of the fact it draws 50A spikes on every halfcycle.

  3. The scary thing about this is that although under some conditions the meter shows far lower than actual consumption (or generation rather than consumption) but under other conditions shows far higher than actual consumption.

  4. Here in the US, the old style meters could be plugged in backwards, and reduce the amount used.
    I had a sis-in-law that did that, and she received a “Credit” on her next bill.
    Needless to say, my brother raised hell, and switched it back.
    The new meters, not so much..

    1. I still have an eddy disc meter and it won’t run backwards, as it has a ratchet to stop this type of shenanigans.
      However this causes me a problem because my solar (grid tied) generation experiments quite often generate more than I can use, and the excess goes off to the grid, so the generating company gets it for free.

      I’m wondering if the Dutch meters specifically can run backwards to encourage people to install solar, so on a good day they power their neighbours and get a reduced bill. Do we have any Dutch readers who can comment?

        1. You should change electricity supplier. Right now by law they are required to refund you the full amount including tax for supplied electricity, until you put in more than you take net over a year. so unless you make significantly more than you use, you can use the grid as an infinite free battery.

          the new “smart” meters usually have 4 or 6 separate counters for 2 different rates, and supply and use. the old meter actually turned backwards, and could be locked by strategically placed magnets.

          the pollical talk for a few years has been to eliminate the law that requires supply to be compensated for at least the same price as use (called “salderen” ) because cries of duck graphs and harder load balancing, and even some song of unsafe conditions due to subgrids being powered when they are supposed to be off. all of which are bullshit, and the once doing the crying are not the ones who do the load balancing. load balancing itself is an issue, but that is mostly because the politicians like pandering and posturing more than spending money where its needed. which to my knowledge is an issue everywhere.

  5. years ago when I was in college majoring in physics, I got involved with the theater department for fun (and better dating prospects). I built a four-channel SCR-based lamp dimmer for some productions we did in the round in the student union. Found out accidentally that certain dimmer settings would cause the class scheduling bells to ring all over the campus (the central timekeeper sent audio tone commands to the bells over the power line). It was fun to watch folks suddenly stop playing bridge and rush off to class only to find that they were way early. Over the next three years, the folks who installed the bell system came out any number of times in attempts to eliminate the erratic operation. Evidently nobody noticed that the bells never acted up again after I graduated …

  6. I heard somewhere that those electronic power meters where sometimes false whith PWM dimmed LED (only the peacks are mesured).
    And for the old times meters, a friend of mine had a very low consomption for years : they used to put a strip of photographic film between the glass and the frame of the meter witch could stop the wheel, of course they had to let it run for a few days a month, so they could have a bill…

Leave a Reply to Alexander WikströmCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.