Smart Meter Installed

[coppice]

Just a note……When I decided to build my own centralized home energy monitoring thing I passed on CT’s (current clamps) and went for a proper meter with RS485/modbus output.
CT’s are sooooo easy to fit but meh!

https://owen-brothers.com/singlephase/direct-connected-mid-approved/ob418-multi-function-100a.html

Ian

You conflated two things there - CTs and clamps. Clamp on CTs are not the greatest thing. They aren't the most stable things, even if you keep the jaws clean. Ring CTs are a whole different matter. They are very stable, and the norm for high current AC measurement.

[Siwastaja]

Ring CTs are a whole different matter. They are very stable, and the norm for high current AC measurement.

Yeah. With non-clamp types, I would be more concerned about phase shifts and frequency response, than the 50/60Hz AC accuracy. And this of course includes the whole chain of processing.

Also coupling of 50/60Hz AC voltage into current measurement signals causes AC drift which is not easy to calibrate out because unlike random noise, it's in perfect sync with voltage measurement and therefore acts like real export or import power.

[coppice]

Ring CTs are a whole different matter. They are very stable, and the norm for high current AC measurement.

Yeah. With non-clamp types, I would be more concerned about phase shifts and frequency response, than the 50/60Hz AC accuracy. And this of course includes the whole chain of processing.

Also coupling of 50/60Hz AC voltage into current measurement signals causes AC drift which is not easy to calibrate out because unlike random noise, it's in perfect sync with voltage measurement and therefore acts like real export or import power.

CT phase shifts are an interesting issue in utility meters. When you have a small meter with CT sensors - say a 60A 3 phase meter for a small workshop application - the phase shift of the CTs gets properly calibrated out at the factory. When you have a bigger meter, like the numerous 6A 3 phase meters intended for use with large external CTs, those external CTs are mixed and matched, without individual calibration. Most meters only seem to allow the multiplier factor to be set. No tweaks for the small but often significant phase shift of the external CTs are usually provided. So, these meters cannot give very accurate results at poor power factors. I have a Voltech PM1000+ power analyser, which can measure 20A internally. For higher currents, or temporary attachment with a clamp on sensor, Voltech supplied optional CTs. They documented the phase shifts of these sensors, but the analyzer has no facility to tweak the phase correction. They only provided for the scaling factor of the sensor. People often do a poor job with these measurements.

As for the leakage of the voltage signal into the current signal, this is mostly a CMRR issue. If you use a modern metrology device with switch capacitor sigma-delta converters the CMRR can be huge, and any leakage is almost always a PCB or wiring issue, and solvable by anyone competent. When you use other kinds of ADC, like SAR or continuous time sigma-deltas, or put amps and other complexity in the signal chain, the CMRR can be a limiting factor.

[EEVblog]

I've changed the Grid Trickle Feed setting from 5W to 50W, I'll give it a few days and see what happens.

After 5 days is made an average difference of 0.35kWh/day
It it really was a 45W increase, over 24hrs that would equal 1.08kWh/day
So I'm not sure that's the problem, but there is an increase there, so 
I've changed it back to 5W today.

[EEVblog]

Just discovered I can download the data from the energ provider website!
Annoyingly with alternating fields for solar and usage, so I need to split out every 2nd row to analyse the data, but I get 30min interval data.

[EEVblog]

Two different days, showing it can actually go to zero.

[EEVblog]

[IanJ]

Two different days, showing it can actually go to zero.

When discharging my battery I notice I am importing between 5 and 10W. Seems to wander up & down every few seconds. Haven't noticed if it goes to zero at all though.

Am no expert, but here's my off the cuff theory:
Reactive Power Compensation (VAR): Import a small amount of power to manage reactive power in the system, maybe to maintain power quality and manage fluctuations in voltage.
Exporting real power, the inverter may still import reactive power to compensate for local voltage issues or grid stability.
Some equipment in the home may require reactive power, especially motors, HVAC systems, or anything with inductance or capacitance.

Via the Solis Inverter API I can read the Inverter Reactive Power value.....haven't actually tried it yet to see what sort of values it produces. There are also limits I can set........hmmmm!

Ian.

[coppice]

Two different days, showing it can actually go to zero.

When discharging my battery I notice I am importing between 5 and 10W. Seems to wander up & down every few seconds. Haven't noticed if it goes to zero at all though.

Am no expert, but here's my off the cuff theory:
Reactive Power Compensation (VAR): Import a small amount of power to manage reactive power in the system, maybe to maintain power quality and manage fluctuations in voltage.
Exporting real power, the inverter may still import reactive power to compensate for local voltage issues or grid stability.
Some equipment in the home may require reactive power, especially motors, HVAC systems, or anything with inductance or capacitance.

Via the Solis Inverter API I can read the Inverter Reactive Power value.....haven't actually tried it yet to see what sort of values it produces. There are also limits I can set........hmmmm!

Ian.

A grid connected inverter is playing a constant balancing act, pushing out just enough power to meet the immediate need. There is going to be some tolerance in this, and maybe that tolerance is 5 to 10W. Also, as the demand changes, the impulse response of the control loop is going to allow some momentary exchange of energy in one direction or the other, and again that may show up as your few watts of residual exchange.

[EEVblog]

A grid connected inverter is playing a constant balancing act, pushing out just enough power to meet the immediate need. There is going to be some tolerance in this, and maybe that tolerance is 5 to 10W. Also, as the demand changes, the impulse response of the control loop is going to allow some momentary exchange of energy in one direction or the other, and again that may show up as your few watts of residual exchange.

Yes, the only way around this is to switch off the grid connection at night when you are using the battery. Maybe a bigger (or multiple?) inverter would make a difference. In my case I've got one 5kW inverter that powers the house at night, load loop response time unknown.

[bsdphk]

Have you already made a sanity-check, to find out if your provider is reading the correct smart meter ?

It is not unheard of that meter numbers get mucked up one of the many ways it can happen.

For "typical households" that can go on for years before anybody notices, and the curves you show do not look totally unreasonable for a normal household where nobody is home during the day.

ALso: Consider talking directly to the meter via the optical port?

[Bryn]

I never bother with smart meters... one engineer from a gas company we were with years ago tried to convince us with one until he had to agree with my mother that it's no point (given that it's only two of us in the house and we don't burn a lot of energy on average).

Now, I have British Gas trying to persuade me with getting one but I hear they're not all that reliable and not many work. Plus, they're too much of a distraction and would make you feel more paranoid of how much energy you're using, just to "save the planet" 🙄

[Electrodynamic]

I suspect the error is in the smart meter time resolution.

I remembered that when I was data logging measurements the time resolution really came into play. So as an Engineer I always want max data down to the millisecond however the data file then become massive and unmanageable. One measurement every 0.1 seconds gives us 864,000 measurements in 24 hours or 26,000,000/month. So we compromise and say every few seconds is enough. However the last value is then held for a few seconds regardless of the current value giving us misleading measurements. We can average the values but this is still inaccurate and simply adds two values and divides by two.

Now add in dirty power, transients or inductive spikes where the higher value is held for a few seconds despite the current value and the errors starts accumulating over time.

An internet search claims smart meters measure power every 1 to 5 seconds for real-time data and 15 minutes for average consumption data. It's also claimed household power consumption profiles can have different recording intervals (1, 10, 15, 30 and 60 min) and different reporting periods.

I have a feeling the claimed "measurement accuracy" could relate to the sensor accuracy which has nothing to do with the time resolution. The sensors could be 99% accurate in one instance but if the time resolution is 60 seconds the total accuracy is still low. The more I read about smart meters the less I like them and the accuracy sounds like a crap shoot in my opinion.

[Monkeh]

An internet search claims smart meters measure power every 1 to 5 seconds for real-time data and 15 minutes for average consumption data. It's also claimed household power consumption profiles can have different recording intervals (1, 10, 15, 30 and 60 min) and different reporting periods.

I have a feeling the claimed "measurement accuracy" could relate to the sensor accuracy which has nothing to do with the time resolution. The sensors could be 99% accurate in one instance but if the time resolution is 60 seconds the total accuracy is still low. The more I read about smart meters the less I like them and the accuracy sounds like a crap shoot in my opinion.

I think perhaps rather than an internet search you should review the standards which set the requirements for metering. Digital meters have been around for decades and nobody is raising concerns about their billing accuracy (some probably did in the early days, of course.. but we didn't have an internet FUD machine to fan the fires of ignorance back then), so what makes you believe the same thing with a radio is worse?

[nali]

Dave's meter probably contains an AD metering IC, something like AD7757. The metering is continously done on-chip, the MCU gets an interrupt pulse for either active or reactive power and just does the counting. No idea where this 15 second malarkey comes from.

https://www.analog.com/en/products/ade7757a.html#part-details

[IanJ]

I never bother with smart meters... one engineer from a gas company we were with years ago tried to convince us with one until he had to agree with my mother that it's no point (given that it's only two of us in the house and we don't burn a lot of energy on average).

Now, I have British Gas trying to persuade me with getting one but I hear they're not all that reliable and not many work. Plus, they're too much of a distraction and would make you feel more paranoid of how much energy you're using, just to "save the planet" 🙄

When you are on that side of the fence that’s exactly how they appear, I’ve been there, but when you switch and especially for electricity the benefits are big……like dynamic tariffs throughout the day allowing you to capitalize and even earn out of it.

Ian.

[Electrodynamic]

Monkeh

I think perhaps rather than an internet search you should review the standards which set the requirements for metering. Digital meters have been around for decades and nobody is raising concerns about their billing accuracy (some probably did in the early days, of course.. but we didn't have an internet FUD machine to fan the fires of ignorance back then), so what makes you believe the same thing with a radio is worse?

My experience comes from my place of work where we have a wide area secure LAN with over 10,000 points and digital metering up the ying yang. In theory once calibrated everything should work as expected but in reality it seldom does. Reality is messy and at any given time maybe ten percent is out of calibration or out of service. You could take ten pieces of identical supposedly calibrated metering equipment and get ten different readings after 6 months.

The problem is that many people not in the automation/controls industry or operating this kind of equipment daily have false expectations about reliability and accuracy. Like the appliance guy repairing equipment daily versus the buyer who thinks his new appliance is actually going to make it's warranty. So I would like to think the manufactures and most tech's know what there doing but all the evidence I have seen suggests this isn't the case. I mean, if everything should work as claimed in theory then why does everything tend to turn into a train wreck?.

[Someone]

Monkeh

I think perhaps rather than an internet search you should review the standards which set the requirements for metering. Digital meters have been around for decades and nobody is raising concerns about their billing accuracy (some probably did in the early days, of course.. but we didn't have an internet FUD machine to fan the fires of ignorance back then), so what makes you believe the same thing with a radio is worse?

My experience comes from my place of work where we have a wide area secure LAN with over 10,000 points and digital metering up the ying yang. In theory....

You presented a theory:

I suspect the error is in the smart meter time resolution.

I remembered that when I was data logging measurements the time resolution really came into play. So as an Engineer I always want max data down to the millisecond however the data file then become massive and unmanageable. One measurement every 0.1 seconds gives us 864,000 measurements in 24 hours or 26,000,000/month. So we compromise and say every few seconds is enough. However the last value is then held for a few seconds regardless of the current value giving us misleading measurements. We can average the values but this is still inaccurate and simply adds two values and divides by two.

Which is pure imagination at this point. Smart meters are billing from energy accumulators, not some periodic sample and hold. Otherwise there'd be a box to grab power on the non-metered cycles and avoid the billing.

[coppice]

Monkeh

I think perhaps rather than an internet search you should review the standards which set the requirements for metering. Digital meters have been around for decades and nobody is raising concerns about their billing accuracy (some probably did in the early days, of course.. but we didn't have an internet FUD machine to fan the fires of ignorance back then), so what makes you believe the same thing with a radio is worse?

My experience comes from my place of work where we have a wide area secure LAN with over 10,000 points and digital metering up the ying yang. In theory....

You presented a theory:

I suspect the error is in the smart meter time resolution.

I remembered that when I was data logging measurements the time resolution really came into play. So as an Engineer I always want max data down to the millisecond however the data file then become massive and unmanageable. One measurement every 0.1 seconds gives us 864,000 measurements in 24 hours or 26,000,000/month. So we compromise and say every few seconds is enough. However the last value is then held for a few seconds regardless of the current value giving us misleading measurements. We can average the values but this is still inaccurate and simply adds two values and divides by two.

Which is pure imagination at this point. Smart meters are billing from energy accumulators, not some periodic sample and hold. Otherwise there'd be a box to grab power on the non-metered cycles and avoid the billing.

Simple meters that merely accumulate net exchanged energy have no time issues. Ones that accumulate in separate bins for import and export, 4 quadrants, or other multi-bin schemes can have a sampling time element in their behaviour with dynamic loads. However, I have never seen one where time resolution below 1 second will make any difference.

[EEVblog]

For those playing along at home.
Still getting some very large solar feed-ins even with the 25kWh battery.
Expect this to rise even more come summer time.
I guess I really need a plan that pays the best feed-in tarrif.

EDIT:
https://www.ipart.nsw.gov.au/solar-feed-tariffs-nsw-1

[Siwastaja]

Simple meters that merely accumulate net exchanged energy have no time issues. Ones that accumulate in separate bins for import and export, 4 quadrants, or other multi-bin schemes can have a sampling time element in their behaviour with dynamic loads. However, I have never seen one where time resolution below 1 second will make any difference.

Yeah. Any modern meter would meter to separate "bins" or "registers" for import and export, real and reactive power.

At any single point in time, there is only "power", a 1-dimensional scalar value, the physical quantity: P = E/dt = U*I.

But a modern meter can then choose to accumulate the value in a separate counter based on the sign. As meters would sample at least thousands of times per second, during any longer time period (even just 1 second) there can be both import and export. Now the key question is, how is this billed? Now this stops being physics and engineering, and becomes an arbitrary choice and varies by country, region, company, sometimes there is legislation involved, sometimes just convention. And it seems that knowledge and discussion always stops here. Many are well able to explain how the energy registers are accumulated, but given place X on Earth it seems no one knows how that data is used to calculate the billing. Which is weird because usually people are quite interested in their contract terms and reasoning for their bills, like you wouldn't accept buying bananas and getting a random bill and then just speculate over the Internet about potential accuracy issues in the amplifiers used in digital scales.

I'm nearly able to answer how it works in Finland: on some customers, net (import + export) is calculated for each hour, resulting in one scalar for that hour. For others, the net is calculated for each 15 minutes, resulting one scalar for 15 minute period, or, if looked at hourly data, both import and export on same hour is again possible. This means you can export excess PV to grid at 1kW for 7.5 minutes and then import 1kW for another 7.5 minutes and be billed for zero. Then again, even I don't know if I'm being billed on the 1-hour or 15-minute net period.

[Siwastaja]

My experience comes from my place of work where we have a wide area secure LAN with over 10,000 points and digital metering up the ying yang. In theory once calibrated everything should work as expected but in reality it seldom does. Reality is messy and at any given time maybe ten percent is out of calibration or out of service. You could take ten pieces of identical supposedly calibrated metering equipment and get ten different readings after 6 months.

You are just being delusional. But worry not - energy meters are fairly simple devices which measure both voltages and currents within the single box, directly interfaced to a specific measurement IC or a microcontroller which performs the power calculation. The performance is also tightly regulated by legislation (e.g. in EU, MID directive). Any internal timing is definitely a non-issue, I can assure you, at least with any even remotely "normal" household load.

[Jeroen3]

You mean there is some error in the accumulator choosing which counter is active?
I don't think that is that much of a problem. The integration time should be fixed, and the totals at the end of this time are added to the active tariff depending on the direction of energy. Should not have error. (or at least, it's within the class of the meter)

[Someone]

Many are well able to explain how the energy registers are accumulated, but given place X on Earth it seems no one knows how that data is used to calculate the billing.

Given most of the tariffs in Australia price incoming and outgoing energy at significantly different rates I'm confident they use the import and export registers directly. This is evidenced by Daves data where import and export are explicitly called out in each 30min period and can both be > 0 in the same period down to tiny accumulations.

Example meter that aggregates a 3 phase supply to the simple data ready for billing:
https://mercs.com.au/wp-content/uploads/2019/12/20181707_ME-PM-M4-5-PROTOCOL.pdf

Code: [Select]

Reg 38 Import Wh since last reset
Reg 39 Export Wh since last reset

[Jeroen3]

I wonder how fast the power oscillations need to be for the meter to start missing them.