-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Beware SMETS2 meters
- Thread starter mikemillar
- Start date
Ulmo
Active Member
Maybe I'll find a better thread for this later, but I'm seeing frequent instant_reactive_power measurements between -2,081 and 5,156.5 in my sample period of August 1 til current (2020) in my records, where negative corresponds to times when there is generally inflow from grid and positive is when there is generally outflow to grid (I can tell looking at the voltage swing that happens because we're 11 houses down from the utility's last transformer (single home sized homes in a suburb, so something like 800 feet plus or minus 50% -- I have to go to work so can't measure right now)). (Outliars were filtered out, min -7,073.05 max 5353.53.) I don't know if I sign switch the values going into my database to align all the numbers to be the same for my records, and don't have time to look it up. I remember only seeing instant_reactive_power measurements in the low hundreds range, not thousands, so I find it odd. Little has changed in our home since then, but it hasn't stayed static.
I've just received an email from Tesla UK which includes:
"We have received reports that, under some circumstances, there can be an incompatibility issue between the Tesla Powerwall storage system and the Landis+Gyr E470 100mm smart meter. In some cases this may result in inaccurate utility meter readings; the impact varies but could be up to +/- 5kWh per day on some sites." The email includes a photo of what the meter looks like and recommending that any Powerwall owner who has this meter should contact their supplier.
"We have received reports that, under some circumstances, there can be an incompatibility issue between the Tesla Powerwall storage system and the Landis+Gyr E470 100mm smart meter. In some cases this may result in inaccurate utility meter readings; the impact varies but could be up to +/- 5kWh per day on some sites." The email includes a photo of what the meter looks like and recommending that any Powerwall owner who has this meter should contact their supplier.
sashton
Member
I received the email. I had already contacted Octopus about the timing issues with the L&G E470 but thought I should re-register this complaint. I was told, after they had a discussion with their 'metering dept.', that they had not received any similar complaints - which is surprising.
I've been following with this with interest as a few weeks ago I switched over to Octopus an they installed the L&G E470 smart meter only a few days later. My solar system + Powerwall 2 is currently being installed and should be up and running today/tomorrow. I have already sent an email to Octopus asking if they are aware of this issue and what steps they intent to take in order to remediate the problem. So that's at least one more 'complaint' they now have....
JohnTisbury
Member
I would take their comment with a pinch of salt. I have registered a formal complaint back in July 2019 and followed it up repeatedly for the past 18 months, I'm almost on the payroll now! In mid August Octopus confirmed the issue did exist and Tesla / L+G had replicated the issue in Tesla's test area.
JohnTisbury
Member
This is a screenshot of the email from Tesla received 15th October 2020 confirming the issue and what to do about it. No idea why they have marked it as Private and Confidential, I've removed that bit!
sashton
Member
Their reply this afternoon said they had had a sudden influx on this issue. They seem to have forgotten that I emailed them a year ago about these two issues - 1) this E470 was incorrectly recording use in the wrong time period i.e. I was drawing 20kW one minute after off-peak started and 4 minutes of that was recorded as peak use 2) the apparent leakage which was not recorded by a MID class B meter in series.
I'll give them another month.
I'll give them another month.
I have this exact same problem. I’ve contacted Octopus and am awaiting a reply. Hopefully a meter change will fix this too.
I don’t think so. The issue is supposedly with one particular model of meter when combined with a powerwall. Although both are said to be within spec. In my case I’m just comparing what the IHD says compared to the Tesla app (or a clamp meter on the outgoing cable). It’s the opposite to what most appear to be experiencing and the meter itself isn’t set to record export. So it’s very difficult to verify which is at fault but I’m leaning towards it being the Powerwall as the clamp meter confirms there is power being exported.
Glan gluaisne
Active Member
The bottom line here is that domestic electricity metering in kW has always (historically) assumed that all equipment/appliances used in a domestic environment would have a PF close to unity, or at least better than 0.9. Electricity metering for commercial/industrial use has always been in kVA, rather than kW, to force industrial users to fit PF correction. Fitting PF correction makes the reactive power reading on a kVA meter match more closely to the kW reading on a true power meter (obviously if the PF = 1 the two would always read the same).
As far as the electricity distribution network is concerned, what matters to them is reactive power, or more accurately the current flowing through their cables and transformers, rather than the true power. This is why relatively heavy industrial consumers have always been metered and charged in kVA. What's been happening with domestic equipment and appliances over the past decade or two has been a switch to using inverter controlled motors, switched mode power supplies and LED lighting with a very poor PF. As a consequence, the legislation was changed to allow energy suppliers to meter, and charge, for kVA for domestic supplies.
In this case it seems that these meters were set to default to kVA, rather than kW, in error, although sooner or later I think it's inevitable that domestic metering will be switched over to kVA, anyway. The intention is to provide motivation for the designers of equipment and appliances to build in effective PF correction, so that the load looks less reactive as far as the distribution network is concerned. This would bring domestic metering in line with industrial metering, and make PF correction a priority, as it will reduce bills.
I know very little about the Powerwall, but having read this thread it seems that it has a very poor PF when it's running at a low level, perhaps just maintaining freq sync with the grid. If this was fixed, so that it looked more like a pure resistive load (i.e PF = 1) then the problem would go away, even after domestic metering switches over to kVA (assuming it does at some time in the future). The same applies to the manufacturers of other domestic equipment and appliances. It's not at all uncommon for many LED lights to run with a PF of around 0.5, for example, and many switched mode supplies in appliances from TVs to phone chargers are much the same. Years ago, pretty much every domestic appliance had a PF of better than 0.9, even inductive loads, like some motors, were usually PF corrected to some extent, usually to a PF of better than 0.85.
As far as the electricity distribution network is concerned, what matters to them is reactive power, or more accurately the current flowing through their cables and transformers, rather than the true power. This is why relatively heavy industrial consumers have always been metered and charged in kVA. What's been happening with domestic equipment and appliances over the past decade or two has been a switch to using inverter controlled motors, switched mode power supplies and LED lighting with a very poor PF. As a consequence, the legislation was changed to allow energy suppliers to meter, and charge, for kVA for domestic supplies.
In this case it seems that these meters were set to default to kVA, rather than kW, in error, although sooner or later I think it's inevitable that domestic metering will be switched over to kVA, anyway. The intention is to provide motivation for the designers of equipment and appliances to build in effective PF correction, so that the load looks less reactive as far as the distribution network is concerned. This would bring domestic metering in line with industrial metering, and make PF correction a priority, as it will reduce bills.
I know very little about the Powerwall, but having read this thread it seems that it has a very poor PF when it's running at a low level, perhaps just maintaining freq sync with the grid. If this was fixed, so that it looked more like a pure resistive load (i.e PF = 1) then the problem would go away, even after domestic metering switches over to kVA (assuming it does at some time in the future). The same applies to the manufacturers of other domestic equipment and appliances. It's not at all uncommon for many LED lights to run with a PF of around 0.5, for example, and many switched mode supplies in appliances from TVs to phone chargers are much the same. Years ago, pretty much every domestic appliance had a PF of better than 0.9, even inductive loads, like some motors, were usually PF corrected to some extent, usually to a PF of better than 0.85.
I have a date from octopus to remove the E470 100mm meter and replace with (I hope) the E470 120mm one on 12th November 2020. Once it is done I will raise the matter of compensation with them.
Well done to everybody for their persistence in managing to get Tesla and Octopus to recognise this issue and finding a solution.
I received an email from Octopus (after enquiring what was happening ref meter replacements. A link to their blog is included:
We are aware about Tesla sending the email out informing all of their Powerwall Customers that the L+G E470 100mm Meter is not compatible with their powerwall. This is true and is an issue we are aware of.
I've added your account on a spreadsheet for now and we will be in contact soon to fix this for you.
Meantime please have a read on our blog about the current issue with this type pf meter and the Powerwall incompatibility, it is explained on the bottom of the page here: The perils of being a tech innovator, working with a tech innovator.
Thank you for your cooperation and I do apologise for any inconvenience caused.
I really cannot fault Octopus for their customer service!
We are aware about Tesla sending the email out informing all of their Powerwall Customers that the L+G E470 100mm Meter is not compatible with their powerwall. This is true and is an issue we are aware of.
I've added your account on a spreadsheet for now and we will be in contact soon to fix this for you.
Meantime please have a read on our blog about the current issue with this type pf meter and the Powerwall incompatibility, it is explained on the bottom of the page here: The perils of being a tech innovator, working with a tech innovator.
Thank you for your cooperation and I do apologise for any inconvenience caused.
I really cannot fault Octopus for their customer service!
Last edited by a moderator:
Glan gluaisne
Active Member
Interesting article by Octopus. If I had to guess, then I'd say that the problem may well relate to an aliasing error, related to the sampling frequency used by the meter A/D converter(s). I built an excess PV generation diverter and import/export monitoring system, years ago (still works fine), and used an Analog Devices electricity meter chip to do all the heavy lifting. This has a couple of high resolution A/D converters, one to measure the current waveform and the other to measure the voltage waveform. It's a common chip used in many electricity meters, and samples at 900 kHz, with 16 bit resolution.
I've just re-read the spec for the A/D conversion parts of this chip, and one thing jumps out of that, and that is that the phase error between the current and voltage channels is only guaranteed to be less than 0.1° over a fairly small incoming frequency range, and that the phase error will increase with input frequency. There's a note in the data sheet to the effect that additional filtering may need to be added before the sampling inputs in order to maintain an acceptable phase error between channels if the incoming frequency is outside the 45 Hz to 65 Hz range.
The Octopus article mentions noise being within the EMC Directive limit (I used to be head of Type Approval for some UK kit, so was a Notified Body under both the EMC and LV Directives). That's not really a very tough limit, particularly with regard to conducted emissions, and there have been other cases where equipment that complied with the EMC Directive has still caused significant interference (switched mode power supplies being a very common cause of problems). If this particular smart meter hasn't got adequate filtering to suppress high frequency conducted emission components from the Tesla Powerwall, then it seems possible that the erroneous true power readings being logged by the meter are aliasing artefacts, arising from the meter A/D sampling frequency being too low to accurately measure the high frequency noise signal.
Interestingly, the clue to this might be that the meter appears to be recording kVA, rather than kW, when a wide band check meter was fitted in series. If that accurately recorded both the current and voltage waveforms within the low level noise from the Powerwall, then it seems quite probable that it might seem, at first glance, as if the smart meter was erroneously set to record kVA.
Full marks to Octopus for being so tenacious in investigating this, and in being so open about reporting it.
I've just re-read the spec for the A/D conversion parts of this chip, and one thing jumps out of that, and that is that the phase error between the current and voltage channels is only guaranteed to be less than 0.1° over a fairly small incoming frequency range, and that the phase error will increase with input frequency. There's a note in the data sheet to the effect that additional filtering may need to be added before the sampling inputs in order to maintain an acceptable phase error between channels if the incoming frequency is outside the 45 Hz to 65 Hz range.
The Octopus article mentions noise being within the EMC Directive limit (I used to be head of Type Approval for some UK kit, so was a Notified Body under both the EMC and LV Directives). That's not really a very tough limit, particularly with regard to conducted emissions, and there have been other cases where equipment that complied with the EMC Directive has still caused significant interference (switched mode power supplies being a very common cause of problems). If this particular smart meter hasn't got adequate filtering to suppress high frequency conducted emission components from the Tesla Powerwall, then it seems possible that the erroneous true power readings being logged by the meter are aliasing artefacts, arising from the meter A/D sampling frequency being too low to accurately measure the high frequency noise signal.
Interestingly, the clue to this might be that the meter appears to be recording kVA, rather than kW, when a wide band check meter was fitted in series. If that accurately recorded both the current and voltage waveforms within the low level noise from the Powerwall, then it seems quite probable that it might seem, at first glance, as if the smart meter was erroneously set to record kVA.
Full marks to Octopus for being so tenacious in investigating this, and in being so open about reporting it.
I have only had a PW2 for a couple of months and I do not appear to have the issues outlined in this detailed thread and in the Octopus' blog. I do have the L&G E470 SMETS2 100mm meter. What posters haven't specified here is their set up. I have 7kWp of solar on the house and on an adjacent detached garage. The PW2 and the Solaredge inverter are located in the garage about 3Ms away from the Gateway2 (located next to the external meter box). The Tesla Meter is located in the garage, as is a Harvi for the MyEnergi products (Zappi and EDDI). A further Harvi is located in the Gateway 2. They all seem to play nicely. What I am asking is - as has been suggested in the Agile ToU Forum - could device proximity be a factor? I am also told that as the L&G 100mm meter is the successor to the 120mm meter, there are not many 120mm meters around. Personally, given the issue that I had late last year with a botched gas meter firmware update, I would prefer a non L&G meter replacement should Octopus decide that a meter replacement is needed.
Octopus engineer is swapping my L&G meter out right now for a new Kaifa MA120. Not much info on these online, apart from type approval being granted in April last year. Apparently there are 320 Octopus customers with Powerwall + L&G E470 and every one of them is getting a new meter. They have 8 engineers dedicated to this task, so should be all done in the next month.
JohnTisbury
Member
Octopus Energy called me today to schedule a meter swap on 30th November, swapping out both meters (gas and electric) hopefully an end to our 18 months of phantom drain.
