Load Power

[reef]

I'm running a H1-5.0-E and F3600 setup (4.8kW and 4.62kW arrays) and am finding an issue with inaccurate reporting of load power. As far as I can see, "Load Power" is not a parameter that is measured but is possibly the sum of MeterPower and InvPower. The issue I'm finding is that as the F3600 is connected to the H1 via a CT clamp, the "Meter2Power" value which should also used in this calculation is ignored by the app and energy stats, so as my generation from here increases, my load value becomes increasingly inaccurate.

Yesterday for example, the data dump from the v2.0 app shows we used 9.1kWh (LoadPower). That is correct as it comprises of 5.8kWh from the grid, 1.5kWh from the inverter and 1.8kWh from CT2. Unfortunately, the app and energy stats only show 7.6kWh as they totally ignore meter2power.

I'm guessing this isn't a wide issue as dual inverter setups are rare, but is this something that can be programmed into either app so that an inversion of the CT2 figure can be included in the calculation? Its a bit annoying having to download the data from the cloud if I want to know the load that day.

[Dave Foster]

Yes load power is a sum that takes into account what it measures at the CT (meterpower) and load power is the amount of power the inverter has to generate to get a zero sum at the CT, then after that anything negative is feed_in.

The problem with a separate generation source is the inverter sees an imbalance and can’t accurately work out the house load power - so the need for CT2, and when connected to the other generation source the inverter has all the metrics it needs to accurately work out your house load power.

I’ll be honest i’ve never looked too closely at the stats for a system with CT2 (I know CT2 should be shown as negative power) but I was expecting you would have the precise load power reported in real time and in usage.

Do you correctly see the real time house load power, or is that in-accurate ? - if that isn’t correct it sounds as if CT2 might not be positioned correctly and may also be seeing some of the house load.

Also as previously mentioned Fox cloud inverts CT2 so invert in Energy Stats is needed to make it correctly in the graphs, often if load power is wrong it’s because the CT2 has been installed the wrong way round and turning it round should correct it. (Fox app 1 is misleading as it always shows it as positive even when it’s the wrong way round)

I don’t know whether this will help but Al has said that Energy Stats doesn’t include CT2 in PVPower (yet) and he’s had a few requests to include it - he’s confirmed he will get to that later in February once he’s cleared all the openAPI work.

[reef]

Yes, the house load reports correctly in real-time and the CT2 value is negative in energy stats and on the cloud (as Meter2Power) but not in the Fox app as you say.

When downloading the raw data in the v2 app, "load power" is actually correct, but the apps appear to ignore CT2 in the calculation (possibly as its negative?) and just use InvPower and MeterPower.

The data is there as it should be and shows correctly in the graphing on the cloud. Its just the reported Load figure that is wrong, so perhaps it can be sorted in software?

[Dave Foster]

Hmmmm that’s a good one - i’ll need an expert for this one

[alpriest]

Energy Stats only uses the parameter loadsPower in the real-time display for house load.

meter2Power is optionally combined with pvPower in calculating the total "Yield today", based on the setting "combine ct2 with pvpower".

I could imagine that the position (and direction) of CT2 would make a difference as to how these values display. So I could add an option to include meter2Power in the real-time loads if you think that would help?

[reef]

Is it possible to calculate the "Used Today" using the LoadsPower? Here are today's figures:

Adding up the 5 minutes figures show "LoadsPower" at 5.9kWh for the day, but Energy Stats (and Fox's apps) both show 3.6kWh instead. I could be wrong, but it looks like the apps are adding InvPower and Meter2power but as meter2power is negative, its doing 7.14 minus the 3.59kWh to give the 3.6kWh figure, so its reporting 2.3kWh too low.

In Energy Stats, I have the "Invert CT2 values when detected" option ticked, as then it correctly shows the feed from inverter 2 as positive. I don't use the "Combine CT2 with PV power" option though as this essentially subtracts the PV yield of inverter 2 from the main inverter due to it being negative and I get an incorrect lower number.

Looking at the figures over a few days, it looks like the way the inverter is calcuating the "Loadspower" figure is:

InvPower (7.14kWh) + MeterPower (4.78kWh) + (the inverse of) Meter2power (3.59kWh) = 5.94kWh (loadspower)

I'm assuming the CT clamp from Inverter 2 to Inverter 1 is the correct way around as it is seeing it as a negative value (incoming power).

Its all a bit confusing!

[Dave Foster]

If your real time load power value is showing correctly, your CT2 has to be the correct way round.

I’ve just heard from Al he has added an option in Android 2.10 and iOS 2.4 that will allow CT2 to be incorporated into loads, should be available in the next 24 hours - can you try that and see how it performs for you please.

[reef]

I've just installed the update so I'll keep an eye on the load figures and see how they come out. A big thanks to Al.

[reef]

Unfortunately, the issue still persists and the new option to include CT2 in the load power just negates the live value but does not affect the cumulative total for the day. In the screenshot here the main array connected to the H1-5.0-E is generating 1.01kW. The F3600 is generating 809w. The house load is around 700w, but because the new option to include CT2 in with the house load is ticked, then it actually deducts it from the true figure due to CT2 being negative and therefore shows nothing.

The main issue Im having are not the live values - they're fine, its the cumulative total which is shown in the red box. This is what is inaccurate. in almost 3 hours with a steady load of 700-800w, it has ticked up by only 1kWh. A download of the main data from the inverter in the v2 app correctly shows 2.2kWh was used in the "loadspower" column.

[Dave Foster]

ok thanks, i'll give Al a nudge, this is probably more complex than it looks - bear with us.

[Dave Foster]

I know you've swapped messages with Al, i've been discussing with him, and also Tony Mathews who has a lot more experience of installations like this, Tony confirmed that there is nothing unusual about your installation, what is unusual is that it's not working.

So I think we all agree that there are some very confusing results here, can we start at the beginning, as we could do with some additional information to work through the problem.

How did this system evolve - was it an F3600 solar inverter first and the Hybrid was added separately or were they both installed at the same time ?.

Does the F3600 itself have a CT installed or is it just the Hybrid that has CT1 and CT2 fitted.

Is it possible you could share a schematic but preferably a picture of where CT2 clamp is installed - it should be on the live tail from the F series inverter as near to the inverter as possible so that it doesn't detect anything other than the F series output.

Are both plant's (dataloggers) allocated to the same account/site ?, for example Foxess may aggregate the load data differently if they are in separate accounts/sites.

Are you on Facebook ?, it might be quicker for us to direct message rather than post here, and just use this thread to share data - if you are, can you private message me your facebook or whatsapp details.

[reef]

Hi Dave, this is the installation:

The H1-5.0-E inverter was installed first in 2023 and the F3600 last month. I'm not very knowledgeable and maybe you could confirm from the picture, but I don't think the installers have actually installed a CT clamp between the two as I requested. Looking at it, there appears to be nothing obvious connecting them both. I always assumed as I was getting a CT2 reading on the cloud/energy stats/Fox app that it was present, but now I'm not so sure. I've flicked through the inverter front panel settings and it isn't seeing it. I also have the added complication that the installers also didn't install a wifi module onto the F3600, meaning I had to install one myself. That added the problem that the inverters are on two different identically named sites in the cloud.

If this is the case then I'm not happy with the installer as they've already had to come back due to forgetting to install the generation meter on the second inverter!

[Dave Foster]

It's not visible on the picture but it could be fitted where the solar distribution box is, i've sometimes seen them clipped round the live inside the DB.

But it answers one question, there is definitely no CT fitted for the F series which is correct for this type of install.

The fact that you are getting a reading from CT2 even though the inverters are in 2 distinct accounts makes me think it must be fitted, so it's a question of where - i'm sure you're familiar with what the Foxess ones looks like it's a small white cable clamp with a black cable - the main CT(1) clamp will be fitted around the live tail of your smart meter.

I think we need to establish as a matter of fact where the meter2power reading is coming from - I would have thought if the Fox accounts are distinct then there is no way that there could be any data exchange in the cloud as it has nothing to tie them together.

On the H1 hybrid if you can go into Settings, Feature, Meter2/CT2 it will give you 2 options 'Set' & 'Check' - if you select 'Set', it should say [CT], if it does you have a CT2 clamp fitted (somewhere) - just click the right hand return button to get out of there. Obviously if it says [Disable] then there is no CT2 fitted but i'm guessing there is - can you confirm what you see there.

I'd really like to eliminate any possibility that the F3600 datalogger is somehow sharing it's load power in the cloud - so if you wouldn't mind, could you unscrew and remove it - and leave it out for a couple of hours and then post the graphs from the foxess website of your meterpower, meter2power and loadpower for today so it shows the before and after.

The next thing to do is assuming it is fitted we must find CT2, the live cable from the inverter will be going from your loft into a distribution box usually where the main consumer unit is, it'll have the breakers to isolate the AC feed from the inverters - wherever that is, is there any obvious CT2 clamp fitted? - as I said before installers often clip it to the live wire inside the distribution box before it goes into the breaker but unless you have any electrical training I wouldn't take the lid off to check.

Once we know for sure you have a CT2 fitted to the H1, that there is no data flow happening in the cloud, and hopefully where CT2 is fitted we can move on.

With those eliminated, it's very possible that CT2 is fitted in the wrong place or the wrong way but we have to know where it is, and be able to flip it round to test that - I can quite understand you wanting to get the installer back to look at this but lets do those tests first and see where we get to.

Stupid question - your not located in the North West (or North Wales), in which case i'm close enough to take a look myself.

[reef]

Hi Dave,

I've been through the inverter settings and CT2 appears to be there. If I go to meter2/CT2 and set, when selecting "CT" it shows "success".

I actually had 10 days after installation without a dongle on the F3600 (From 11th - early on the 21st Jan) and the figures showed exactly the same discrepancy as now. On the 20th January for example the apps showed a total house load of 9.6kWh, but the "loadspower" 11.6kWh, so I'm confident it is not datalogger related.

Here's the distribution box cabling that is visible. Nothing immediately obvious to me, but there's not really much to see.

I'm guessing in a normal situation meter2power would be a positive number as all other power in (mains or PV) is usually positive and Feed-in (power out) negative? I'm wondering if the clamp (wherever it is) is the wrong way around as you suggest.

I'm in East Yorkshire. I had already emailed the installer yesterday as they actually forgot to fit the generation meter for the F3600 on 10th January, so returned on Friday just gone to do that but when I asked, the guy just wasn't sure. I wouldn't like to go rummaging around in the box as I am not competent in this area so I'll see what they say.

[Dave Foster]

I think that small black cable peeking out of the trunking looks a lot like the CT clamp cable, so i'm guessing it is inside that white distribution box on the right hand side of the photo.

I couldn't see how having 2 distinct accounts for the inverters could affect each others load reading but needed to be sure.

The meter2power feed has always shown negative on the Foxess cloud traces - it seems strange at first, but as it is external generation and is seen as part of the house load by CT1 it has to be deducted from what CT1 is measuring - the problem is that sometimes it is an absolute value, sometimes a signed value - it all depends where you look at it.

Ok, see what your installer can find - at this point i'm thinking it may be in the wrong place, but it could be the wrong way round - it's a simple test to find out, and then check to see what affect it has in the graphs.

Only your installer can add the 2 dataloggers into the same account so all your inverters will be listed in the same site (power station), it might be worth getting them to look at that as well.

Let us know how you are getting on, happy to help.

[Dave Foster]

Can you post a couple of graphs from the fox website, using the graph at the bottom of the energy flow page where you can select sensors.

It would be helpful to see meterpower, meter2power, outputpower, loadpower and feedin power all together on the same graph for each day,

you've mentioned the 20th January above so that days trace plus a few other days where there was reasonable solar generation would be useful.

[reef]

Here are the traces. The first is 20th January, before the wifi dongle was added the next day, Ive zoomed in to the daylight hours as the F3600 turns off and the Output power and Load Power then match exactly as only the H1 is online:

The next two are from the 12th and 16th February. I cant really show any generation from the F3600 better than this as its a NW-facing array and so isnt generating much over 3-4kWh per day at present.

What can be seen is as meter2power increases, output power reduces. When load is low, output power goes negative, which is what will happen when generation is high on the F3600 during the summer I imagine. Load Power is unaffected on the graph, it shows the house load correctly.

I'm wondering if its a cloud issue? The data is there, correct and going to Fox, but despite Load Power data being correct on the graph here, the issue persists?

[Dave Foster]

Thanks for those, yes myself and Tony are having that exact same conversation about the cloud being the issue - we're going back to basics to work out what happens all the way from the inverter to the cloud - you'll also be pleased to know that H3 handles this differently than H1 (but still wrong).

Again please bear with us as we work through it

[Dave Foster]

As you say the load_power is calculating correctly, but not the cumulative - what I would really like to do is to gather raw data from the cloud and look at the report data (totals) and cross reference it, and/or establish a better method for getting to the cumulative loads.

It's a big ask but would you be willing to share your api_key and inverter serial number with me, and I will use it on my test openAPI system to find out where the error is creeping in - if you would, can you please email me on b5djf@yahoo.com and I will get this setup but only during the day when there is solar generation.

My test system uses around 24api calls per hour, so around 40% of your daily limit which should leave you plenty for the ES / V2 apps.

[reef]

Thanks Dave, email sent!

[Dave Foster]

ok, my logging is running - hopefully i'll get back to you shortly

[Dave Foster]

Just a quick update, i’ve spent most of the day running test scripts against your data, and have written some riemann sums that will convert your reported load power 5 minutes samples (kW) into a time based increasing total (kWh).

I’m going to need a full day tomorrow to record a daily set of results as it will show the differences between reported load cumulative and actual load cumulative when measured directly off the load sensors.

At the moment I see exactly what you are seeing, namely the real time load_power is correct but the reported totals are not, as of around 20 minutes ago this was the difference

The first number (3.7kWh) comes from a report fetch by day, which provides hourly sums and when totalled it gives your cumulative load power that you see in ES and Fox apps.

The second number is taken by feeding all of your 5 minute load power samples for today and converting them into hourly usage and when totalled it is 5.295kWh which is I think your ‘real’ cumulative load usage - it’s a heck of a lot of maths but necessary to make sure the data is sound.

From your typical use, would you say that 5.295kWh sounds more realistic than the reported 3.7kWh ? (up to 6.45pm)

[reef]

Hi Dave, it's a similar sort of difference as what I have been seeing. 3.7kWh is definitely not correct. Our usage today was pretty low as no-one was here most of the day, but it's actually a very good example of what is happening as the load totted up in the apps by just 0.1kWh between 8am and 3pm despite a low, but steady baseload. In this case, what was coming in from the F3600 exceeded the load, so output power was negative during this whole time and no load was counted in the total.

[Dave Foster]

Hi Dave, it's a similar sort of difference as what I have been seeing. 3.7kWh is definitely not correct. Our usage today was pretty low as no-one was here most of the day, but it's actually a very good example of what is happening as the load totted up in the apps by just 0.1kWh between 8am and 3pm despite a low, but steady baseload. In this case, what was coming in from the F3600 exceeded the load, so output power was negative during this whole time and no load was counted in the total.

Yes agreed, that is what I am seeing, they are correcting the load power to show the power coming in from meter2, but they are not correcting the load cumulative totals.

Just an FYI if you ever decide to move to a home assistant all of these can be corrected by formulas and integral sums (you don't need to go to the trouble of RS485, you can access their cloud data).

This is your cloud data for today, showing the reported load total 'Fox1 - Load', the integral sum of your real load_power (load_daily) along with an integral sum of meter2 (ct2_daily).

As you can see the 'Fox1-Load' tracks load_daily until meter2 starts to generate power at which point the difference is related to (but not precisely) ct2_daily (the cumulative sum of meter2 today)

But it needs some maths and a home assistant do this.

I am going to email Foxess r&d and explain what I am seeing (incidentally there are different issues with 3 phase hybrids, which I have to cover to them), this might be a lot of backend work for them so I don't know how they will respond, but i'll let you know the response.

I've already messaged Al to bring him upto date with this, it's a difficult fix for the app, a lot of data and maths - but there may be something he can do to correct your todays cumulative load.

[reef]

The F3600 array has generated around 4.8kWh today, so the result is as extreme as ever so should give a good example of it.

As of 16:50, the apps were showing a cumulative house load of 3.4kWh, but by my calculations from Loadspower, we've actually used 7.0kWh.

This is using a download of today's inverter data from the H1 and dividing the 5 minute Loadspower by 12 and adding all of the values up.

Thanks for yours and Al's efforts. It'll be interesting how Fox respond.