Avoiding being overcharged by your smart meter by using AC line filters.

[Fflint]

I believe it is a well established fact many modern smart meters measure highly distorted waveforms inaccurately. For those unaware I'll mention results of one study found errors ranging from +500% to - 40% with a waveform produced by a string of 30 household lights on dimmers (half led, half cfl).
There was also a study done more recently that concluded the above results are "questionable", but they arrived at that conclusion by using waveforms "that meet maximum allowable THD"(according to local gov standard) . Of course if they used a lot less distorted waveforms they got a lot less error (up to 5% if I remember correctly). And they also added that "classic" resistive loads on the same AC line further improve on the measurement error and no one uses just non linear loads in their house right? (this was written in 2018 or so).

Now in 2022 in my house I pretty much only have non linear loads including 75 led lights, induction hob, all appliances with electric motors use inverters etc. I haven't checked current waveforms I have (I'm waiting for CTs to arrive from China), but I'm pretty sure I'm being overcharged by around 10%~15%. (I did all the usual checks in the past)

I imagine the error in smart meters measurements is due to not high enough sampling frequency plus the averaging method used. In my country (as well as in rest of EU I think) if you complain to the power company they will "recertify" your meter with pure sinusoidal waveform drawn by a resistive load.

So I became interested in various filtering technologies. I'm not talking about those "save thousands on electricity" scam boxes that contain a tiny capacitor, but real series and parallel filters.

Unfortunately based on what I read one needs both types of filters to cover two kinds of distortion. Voltage (distortion) sources (all devices that contain diode rectifiers and filtering capacitors) and current sources (all devices that switch on a very low impedance load for a very brief period of time such as many dimmers, led lights etc).

Techniques for design of those filters are known. Also one would probably need some sort of overcurrent bypass of series filters in case one has a high power resistive load (electric flow water heater) that is used rarely.

However, I'm starting this thread to ask if anyone already has designed and is using any filters like those I mentioned. Perhaps existing filters built to improve AC power supply quality could be used in some configuration.

Or perhaps there are some creative low(er) cost solutions to improve individual electrical loads. For example led lights or central heating water pumps (tiny inverter driven). Those electric pumps are running 24/7 in winter. They are supposed to draw 20W~50W  on their setting, but it is measured more like 200W (with two of them). Perhaps sticking an isolation transformer between them and AC line would help? I don't know. So what do you think?

[Siwastaja]

I believe it is a well established fact many modern smart meters measure highly distorted waveforms inaccurately.

Extraordinary claims require extraordinary evidence, so maybe give references to the studies you talk about? Most definitely it is not a well established fact.

This sounds like a typical case where you shouldn't waste time engineering something before you truly understand the problem - if it exists at all. Likely it doesn't.

[Karel]

Electronic energy meters’ false readings almost six times higher than actual energy consumption

https://www.utwente.nl/en/news/2017/3/313543/electronic-energy-meters-false-readings-almost-six-times-higher-than-actual-energy-consumption

"The study, entitled ‘Static Energy Meter Errors Caused by Conducted Electromagnetic Interference’, are
published in the scientific journal ‘IEEE Electromagnetic Compatibility Magazine’. The Van Swinden
Laboratory (the Dutch Metrology Institute), which carried out a countercheck, has confirmed these
results. The study was conducted by Prof. Frank Leferink (Professor of Electromagnetic Compatibility at
the UT), Cees Keyer (lecturer at the AUAS and PhD candidate at the UT), and Anton Melentjev (at that
time, a student at the AUAS)"

[uer166]

From the website you linked:

Here, the electricity being consumed no longer has a perfect waveform, instead it acquires an erratic pattern. The designers of modern energy meters have not made sufficient allowance for switching devices of this kind.

When they dismantled the energy meters tested, the researchers found that the ones associated with excessively high readings contained a ‘Rogowski Coil’ while those associated with excessively low readings contained a ‘Hall Sensor’. Frank Leferink (Professor of Electromagnetic Compatibility at the UT) points out that

As someone who designs and certifies MID meters, both 3-phase and 1-phase, the above smells of BS, the designers certainly have allowed sufficient allowance for high harmonic content. There are a few cases where a lot of the meters in the field read extremely low when presented with a half-wave rectified load, namely the current-transformer based ones. I don't see how anything can read 582% high unless you're trying to defeat the device in some non-real-world edge case. Also, hall and rogowski? Seriously? The author couldn't have chosen anything more exotic than that, household meters are CT or shunt-based.

I have found the actual paper scihub and going through it now, I suggest you do too to understand how the authors may have been mistaken (or not).

[uer166]

Really funny bit:

After opening meters the seal has to be broken, and we observed that all manufacturers use their own specific digital signal processor with proprietary software.

As opposed to uh, an Arduino and reusable seal?? 

In any case, the paper has a lot of truth to it in a technical sense, but also has a lot of problems and misleading info IMO. Some of the issues are:

• They used a mechanical meter as their "reference" meter. Who says that that thing is able to read the high-bandwidth signals they put through it at all? I mean maybe it's not the smart meter that is +582%, but the mechanical meter is -582%
• It seems like they cherry-picked old crappy units that may not actually have had anti-aliasing filters, in which case, yes, you may get aliasing and an over-estimate of energy in very specific conditions.
• The condition was a extremely high distortion/harmonic load, that was relatively low power in total, without any resistive base load. Unless your entire house behaves like a 10kW CFL lamp on a dimmer, you can sleep easy. There is a reason why high-power electronics are required to have PFC, that is part of the mitigation for stuff like this
• They made no attempt to use a real reference meter/transfer standard as a reference. Even a simple digital power meter would have been more credible.

In any case, none of the issues identified would be solved by AC line filters, most of them seem phase dimmer-related in some extraordinary conditions, so in context, this is a mostly made-up problem.

[T3sl4co1l]

Big brain play: run your house off an inverter, supplied from an old fashioned SCR phase control.  Now your whole house is as disgusting as a 10kW CFL on a dimmer! 

Tim

[moffy]

Found this other article based upon the one referenced already:http://www.enmanreg.org/wp-content/uploads/2016/07/reported_meter_errors.pdf
but by:

Kris Szajdzicki is the founder (and now a director) of ND Metering Solutions. He has been designing & manufacturing kWh
Meters for about 40 years, specifically for Energy Management applications. He has also represented the Energy Management
Industry on numerous standards committees such as IEC TC/13 dealing with electricity metering, IEC TC/66 on product safety,
ISO TC/301 on Energy Management, etc.

I can understand that:

A Rogowski device.
This is an air cored coil which surrounds the current carrying cable.
These devices can be used to accurately measure high frequency transients, but the
associated circuitry must be suitably designed. As the output from a Rogowski sensor is
proportional to the rate of change of the current, namely di/dt, this signal has to be integrated
to provide a signal proportional to the current.
Where this integrator is designed for power frequency signals and the lower harmonics, fast
transients saturate the integrator causing errors. If the integrator is designed for such
transients, the output will be a much reduced signal level. This can be amplified, but the
increased noise levels will affect accuracy, particularly where the measured current is small.

If the integrator saturates, which they are prone to, then you could well get very high readings. A high crest factor can be a real issue, looks more like the qualifying waveforms used to certify meters do not cover a broad enough range of currents. Not saying that this is what is normal, but on the outer edges of possible.

[Simon]

I sounds like apparent power being measured. And maybe it is time to switch the way electricity is charged to consumers and have the pay the same way as industry by the apparent power they use not the lesser real power and the apparent power had to be generated as well..... If we are to stick with sine wave supply from mechanical generators then we need to stop pissing away power on poor power factor. If people were charged for the impact of their cheap junk they buy they will soon start to choose better products and the cheap shit will disappear.

Dave did a video on the power wasted by smoke detectors, the apparent power of a nations worth of smoke detectors was a significant number in the MW. As far as I am aware all power supplies over 1kW have to be power factor corrected.

If more generators that use inverters appear on the grid I guess it will be less of a problem, although current delivered in short large peaks is hardly efficient anyway and will still require thicker wiring.

[Marco]

You could try to get one of those big industrial PFCs (the passive ones) and test your home. You're not in Oz so it probably wouldn't be illegal.

[madires]

As far as I am aware all power supplies over 1kW have to be power factor corrected.

EN61000-3-2 requires SMPSUs with more than 75W to include a PFC (may be passive).

[Simon]

so what are the PF limits? the meanwell supply on my 3D printer is 0.5 PF according to some cheap power meter, it draws 350W

[TheMG]

Even ASSUMING you're actually being slightly overcharged, the amount of money you're going to have to spend to filter your AC power back to a nice unity power-factor, sinusoidal current and voltage waveforms, is likely going to far exceed any money you save (if any at all).

Also, no filter is perfectly lossless. You could actually end up increasing your power consumption and the amount you pay, just from the losses of the filter components alone!

[TheMG]

so what are the PF limits? the meanwell supply on my 3D printer is 0.5 PF according to some cheap power meter, it draws 350W

It's actually defined as a set of limits on the amount of each harmonic, rather than simply the power factor. But in order to meet the harmonic current limits you would very likely end up with a PF of 0.9 or better.

I don't know the specifics of the regulation but I'm assuming it only applies to finished consumer goods and not to off-the-shelf SMPS intended to be integrated into a product or for industrial use? Maybe that Meanwell power supply was not intended for the European market at all? At 0.5 PF I would say there is zero power factor correction of any kind in that power supply and it definitely wouldn't be compliant on its own.

Anyways, especially here in 120V land, large SMPS without PFC are rather inconvenient. 350W at 120VAC with a PF of 0.5 is 5.8A of current, which is quite significant when you consider standard circuit breaker size in homes and offices here is 15A. In other words, the better the PF, the more actual loads you can put on the same circuit without tripping the breaker. I have a Sorensen bench SMPS max 60V 20A  that does not have power factor correction, at near-maximum load it will quite happily trip a 15A breaker on its own even though it's only drawing about 1400W input, which is a real pain in the butt!

[TimFox]

On the regulatory front, what is the legal definition of the billable energy (in W-hr) when the load is non-linear?
The scientific definition is the integral over time of the product of voltage and current, where both of them are bipolar real-valued functions of time, not necessarily proportional to each other or in phase.
(With a simple resistive linear load, that results in the usual measurement of mean power as the square of the rms voltage divided by the resistance, integrated over time to get energy.)

[ejeffrey]

the apparent power had to be generated as well.

Apparently power doesn't need to be generated.  It doesn't consume fuel.  However the generators, transmission lines, and transformers need to be *sized* based on apparent power.  There are losses associated with apparent power but it is around the ~10% transmission and distribution losses.  The main reason that utilities care about apparent power is not really the losses -- although they do care about that -- but the fact that reducing apparent power can avoid the need to build new or upgrade old infrastructure.  It allows them to support rising demand by simply increasing generation rather than adding new power lines and transformers.

Dave did a video on the power wasted by smoke detectors, the apparent power of a nations worth of smoke detectors was a significant number in the MW.

Those smoke detectors are completely embarrassing and very lame design, but a few MW of excess apparent power in a nationwide distribution network is not really significant.  That's no excuse for such poor power supply design, but it isn't really causing a major burden on the utilities.  Especially since, while bad power factor, the load from smoke detectors is constant over the day.  Poor power factor is basically the same problem as bursty demand.  In both cases infrastructure cost and losses are set by the peak usage, rather than average.

If more generators that use inverters appear on the grid I guess it will be less of a problem, although current delivered in short large peaks is hardly efficient anyway and will still require thicker wiring.

I'm not sure what you mean here, but inverters vs. mechanical generators is almost no difference for power factor.  Power factor losses come from the additional I^2 losses in distribution and somewhat in the generator itself, but inverters also have conduction losses that depend on power factor.  The one thing related to "inverters" that would improve losses from bad power factor is to have long range power distribution done via HVDC with AC inverters closer to the point of consumption.  This reduces the distance that the out of phase / harmonic current is transmitted and thus the associated losses.  This isn't anything magical about inverters vs. mechanical generators, it is just that the HVDC link is performing power factor correction.  You could do the same thing with a motor/generator coupling.

[madires]

so what are the PF limits? the meanwell supply on my 3D printer is 0.5 PF according to some cheap power meter, it draws 350W

https://en.wikipedia.org/wiki/IEC_61000-3-2

[Jeroen3]

And this is only loads. Do you have a transformerless solar panel inverter?

I have 8 enphase units on my roof, it's night now, the little DIN rail meter shows: 0.61A, 225.6 V, 5.5 W, 138.9 var, PF 0.03, 49.98 Hz.
Imagine roofs and fields with hundreds of these things...

[bdunham7]

I have 8 enphase units on my roof, it's night now, the little DIN rail meter shows: 0.61A, 225.6 V, 5.5 W, 138.9 var, PF 0.03, 49.98 Hz.
Imagine roofs and fields with hundreds of these things...

Dormant Enphase units have a capacitive load which actually helps cancel out the reactive current from inductive loads.

[Simon]

If more generators that use inverters appear on the grid I guess it will be less of a problem, although current delivered in short large peaks is hardly efficient anyway and will still require thicker wiring.

I'm not sure what you mean here, but inverters vs. mechanical generators is almost no difference for power factor.  Power factor losses come from the additional I^2 losses in distribution and somewhat in the generator itself, but inverters also have conduction losses that depend on power factor.  The one thing related to "inverters" that would improve losses from bad power factor is to have long range power distribution done via HVDC with AC inverters closer to the point of consumption.  This reduces the distance that the out of phase / harmonic current is transmitted and thus the associated losses.  This isn't anything magical about inverters vs. mechanical generators, it is just that the HVDC link is performing power factor correction.  You could do the same thing with a motor/generator coupling.

i should know all this having just proof read my university module on the modern energy system but, oh yea, I proof read it and the tutor was insulted that I know more than they do.... that's how little is in the module as I am far from the expert

what effect does poor power factor have on a generator? if it produces one AC cycle per rotation and the power factor is 1 then the load is constant as the generator rotates. If there are enormous peaks of current draw at the peak voltages then that is 6 positions in each rotation that will have extreme torque demands compared to the rest.

Now if this is an inverter it just deals with it providing it can do the peaks, so my solar system is 4kW at most but has an 8kW inverter and at times when i am only generating say 1kW with 8kW capability it can deal with the poor power factor oaf a load, so if something in my house or a house nearby has poor powerfactor my inverter has more chance of dealing with it than a generator that may just end up producing a not so sinusoidal output as 6 times a cycle it hit a bit bump.

[Kleinstein]

A poor power factor will usually lead to extra losses in the transformers and distribution cables. There is a difference between the pure phase shift part, e.g. from capacitors and motors and the nonlinear parts from SMPS without PFC correnction. The inductive and capacitive loads can cancel out before the transformer (at least before the 2nd layer) and when there is usually a inductive part (the transformer it self add some inductive part) there may be already capacitors at the transformer station, either fixed and sometimes switched.

The nonlinear part does generally no compensate that well. Most of those bad supply are similar and take there peak current when the voltage is highest.
Before PCs had PFC the mains in an office building was quite distorted with clipped peaks.

The smart meters did show wrong reading in the test with a current that was way beyound acceptable limits. In addition this were mainly old meters that failed. The readings where relative high, but the absolute reading still moderate. The usually main consumers (e.g. heating, coocking, AC) have a much better PF and would not be effected.
Ideally the meter should show a warning that the load is exceeding permitted limits on the harmonics. Quite often the users don't know when they have too much crappy equipment to exceed the limits and they should do something about it.

[uer166]

I sounds like apparent power being measured. And maybe it is time to switch the way electricity is charged to consumers and have the pay the same way as industry by the apparent power they use not the lesser real power and the apparent power had to be generated as well.....

That is absolutely not what the paper claims. All the meters are designed to measure real power, which is actually easier than measuring apparent power since you don't need to do any RMS calculation math. There is no reason to measure apparent power unless you need to, since it adds complexity. The issues they found are some edge cases related to aliasing, meter bandwidth, and seemingly EMI.

[Alti]

Related topic on harmonics and smart meters.

[Simon]

The thing is power factor seems to cover a multitude of things the 3 main ones I can think of are

Chopped waveform (dimmers)
extreme peaks ( "smoothing" capacitors)
just plain shift

Either way I don't see how the generators cope, if there is a high demand for current that the generator cannot take but is trying to generate then it's going to be a problem. Why else do industrial loads get charged by power factor as well?

[Kleinstein]

The industrial customers get an extra charge for apperant power too, but with a considerably lower rate. This is to encorage taking measures to do the power factor correction already locally.  Things can also vary locally. They are often also charges for the peak power (as this determines the needed size of the transformers and lines).

The phase shift part is the easiest to deal with. There are usually some capacitors at the transformers to reduce the out of phase current, some a fixed and some are connected only when needed.

For the waveform distortions part of this is compensated from other devices, like old incandencant lamps or heaters, that consume current proportional to the votlage. Induction motors also have a tendency to restore the waveform to a more sine shape. So while they add to the out of phase current, they also help with the harmonics and also a little with the peaks.

The capacitors for EMI suppression and to compensate the phase shifted current also deal with much if the sharp peaks.

So not all the crap on the mains has to go back to the generators. Other parts also participate.

For measuring there is no big difference in the effort to measure the real power and the apparent power - especially not for the smart meters. The smart meters sample the voltage and current waveform and than do the rest numerically.  At least some of the chips (e.g. ADE7753) provide both the active power and RMS values and even more info on the grid quality. Would be nice if the meter would also record those to show drop outs and other failures - the information seems to be there.

[jonpaul]

We have smart meters in USA and France, and they have closely followed our expected power consumption.

The bulk of load is either resistive (haters), inductive (motors) or PFC corrected if over 75W.

 The Power transmission and distribution industry and ERPI, have researched these issues and coordinated the metering and power quality standards long ago.

The smart and dumb meters are specifically designed to measure the true real power consumed very accurately regardless of PF or current wave.

Government regulations require  the metering accuracy and compliance testing.

Thus,  I am skeptical of the OP premise that smart meters can be overcharging or that a line filter will change the KwH readings.

Does the OP have any specific evidence (other than weblinks) that support his hypothesis?

With Kind Regards,

Jon