Fronius bug/discrepancy Donator vs Free

[saliya]

Hi folks,

I have now got a few months worth of data collected from my system using two different upload methods:

1. PVOutput Donator-based service. More or less correlates with data from www.solarweb.com.

Data: https://pvoutput.org/intraday.jsp?id=73555&sid=65253
Settings: Fronius Daily Log
SolarAPI v1 - Logdata - Data by HTTP POST every 1h / pvoutput.org:80 /
/service/r2/froniuspost.jsp?sid=65253&key=blahblah

2. PVOutput free service (same inverter, different settings, does not correlate with (1) or www.solarweb.com).

Data: https://pvoutput.org/intraday.jsp?id=74088&sid=65717
Settings: Inverter to PVOUTPUT
SolarAPI v1 - CurrentData - Inverter by HTTP POST every 5m / pvoutput.org:80 /
/service/r2/froniuspost.jsp?sid=65717&key=blahblah&dayenergy=1
Settings: Meter to PVOUTPUT
SolarAPI v1 - CurrentData - Meter by HTTP POST every 5m / pvoutput.org /
/service/r2/froniuspost.jsp?sid=65717&key=blahblah&n=1

There are discrepancies between the datasets - more or less, once the system starts generating power
the power used figure in the free system (2) is wrong. I have experimented with various settings on (2) to try to bring it into line with (1) without success.

Am I doing something wrong, do I need to update firmware, or is this a bug (‘feature’)?
The meter is the generic modbus “Fronius Smart Meter” I think the model is a 63A-3.

Is it possible to get this meter and inverter combination to show solar data correctly on pvoutput.org using the onboard settings and a free PVOutput account (if so, how?), or is it never going to work properly unless my status is ‘Donator’?

Thanks,

Saliya

[jimdcollie]

Hi Saliya ,jim here.i see in the Detailed year view each system has the same numbers for the 6 stats it records ,small difference’s will be seen in the live views because 5 min data is averaged over 5 min and the hour log push is averaged over 1 hr, so because one is 5 min and one over a hour the curves will slightly different, however as the year view says both systems are doing exactly the same in all 6 recored parameters, hope this helps.

[saliya]

Hi Jim,

really appreciate the response, thanks. I don’t think it’s to do with
the logging interval but happy to try different settings out (e.g. hourly
pushes).

It seems to me that the Energy and Power Used figures that don’t agree;
and they only disagree while the system is making power.

For example, if you look at yesterday’s data in both views:
(1): https://pvoutput.org/intraday.jsp?id=73555&sid=65253&dt=20190612&gs=0&m=0

(2): https://pvoutput.org/intraday.jsp?id=74088&sid=65717&dt=20190612&gs=0&m=0

between say 7am and 4pm, you’ll see what I mean.

Is this expected behaviour?

Thanks,

Saliya

[willisave]

So, you’re basically concerned about discrepancies in the red consumption lines in the two systems, is this correct?
Consumption doesn’t come from the inverter - the inverter system has no idea what you’re using, it only reports the green generation line.
Consumption comes from the meter system - so best to concentrate on the meter upload commands in each system.

System 1 red consumption line looks fine.
The pattern in the second system, where during sunlight hours consumption drops to zero and has gaps with no data, usually means you have a ‘net’ setting wrong - and the system is subtracting the green generation line from the uploaded red consumption data, and getting negative numbers (which plot as 0 or blank).

Check, for both systems, the settings page in the ‘Advanced Settings’ section that you have consistently the same settings for ‘Net Calculations’, ‘Net Power Tracking’, and ‘Net Method’.

Also, in System 2 your Meter upload line has ‘&n=1’ (which indicates the data should have the net calculation done) while the System 1 Logdata line does not have a ‘net’ setting in it. Try setting ‘&n=0’ in the System 2 ‘Current Data Meter’ upload line instead to turn off the ‘net flag’ and see what happens.

(see https://pvoutput.org/help.html#push-froniuspost)

[bankstownbloke]

Donation mode allows the additional hourly ‘Log Push’ which usually matches the data sent to solarweb. The 5-minute push feed between the inverter and the net meter can sometimes be unreliable depending on the firmware / setup etc.

[saliya]

Thanks @willisave for the input and suggestions. Yes, exactly, concerned about the consumption discrepancies between the two systems. Non-Donator system seems inaccurate.

Tried switching n=1 to n=0 in the meter upload (on 12th June); power used then follows generated power very closely in system (2) (as in: it’s now completely wrong). n=1 is closer to correct but is still wrong when the system is generating power. Switched back to n=1 today.

Examples of n=0:

(1) https://pvoutput.org/intraday.jsp?id=73555&sid=65253&dt=20190617&gs=0&m=1
(2) https://pvoutput.org/intraday.jsp?id=74088&sid=65717&dt=20190617&gs=0&m=1

‘Net Calculations’, ‘Net Power Tracking’, and ‘Net Method’ settings for both systems are the same; but on the ‘non-Donator’ system I cannot alter ‘Net Power Tracking’ or ‘Net Method’ (greyed out / not selectable).

Any other suggestions? Any non-Donator users with the ‘Fronius Smart Meter’ who get proper consumption data?

Thanks,

Saliya

[willisave]

Sorry - I don’t have a Fronius meter in the system, I get my consumption line from battery. I’ve run out of knowledge.
I’m guessing the meter is wired in at the grid connect point, which means it ‘sees’ a combination of the actual house load and the inverter output. When the inverter is generating more than the house is sucking, the result is power being pushed out to the grid, which is ‘negative grid consumption’ - and PVO won’t plot negative numbers for consumption.

[saliya]

@bankstownbloke and @jimdcollie

Changing the report interval to 1h highlights the problem.

I changed the reporting interval to 1h on 2019-06-25; you can see an example here:
https://pvoutput.org/intraday.jsp?id=74088&sid=65717&dt=20190626&gs=0&m=0

When the system is generating power greater than usage; pvoutput.org shows usage as 0.

For my system, this is wrong (there is always usage). Further, it does not seem related to
either lost/missing data or the reporting interval.

So it must be related to the calculations applied to the data by pvoutput.org.
I am happy to try to help fixing this, if pvoutput.org is interested in a fix.

If pvoutput.org is not interested in fixing this bug for non-Donators, please just say so
(or say nothing).

Cheers,

Saliya

[darrinm81]

Interesting.

Looking at:
https://pvoutput.org/intraday.jsp?id=74088&sid=65717&dt=20190626&gs=0&m=0

The five hourly entries 09:00 to 13:00 show Power Use as 0. This seems common on other days - most typically during the day.

So somewhere the Power Usage value is either zero or null. Zero feels very unlikely unless both solar panels and meter box have gone off line.

Could you check SolarWeb for the same periods (chart and report) - are there power usage values there? I assume there is a value because the 5 minute data shows a value?

The table has white and green shading from 00:00 to 08:00. What does this mean?

Can you try and remove the Hourly push service and re-add it?

I think we need to rule out the Fronius Push Service. However it does seem like it could be a problem with PVO reading in the data.

[saliya]

Hi @darrinm81. Thanks for your response. I think that particular graph is misleading; because at around 6:58 on that day I changed the system from reporting every 5 minutes to reporting every hour; and I think there’s another bug in pvoutput.org that incorrectly handles interval changes on the day of the change.

If you look at the next day:

https://pvoutput.org/intraday.jsp?id=74088&sid=65717&dt=20190625&gs=0&m=1

you can see the smoothed 1h graph that clearly shows zero usage between 8am and 12pm; and this is not the case in my household. You can see the actual (or fairly close) usage for this time at

https://pvoutput.org/intraday.jsp?id=73555&sid=65253&dt=20190625&gs=0&m=1

showing that there is no time during generation when the usage is zero (which is more or less correct).

So far nobody seems interested in fixing this; so there may be the opportunity for another player to enter the game…

[bankstownbloke]

So far nobody seems interested in fixing this;

It is not possible to get both to match as the 5-minute data meter push to PVOutput is instantaneous net data which may not be accurate depending on when the push was made. It also relies on being in-sync with the separate inverter push to calculate consumption.

This is why the auto=1 and Log Push methods have been added to address these issues by using average energy data.

[darrinm81]

Thanks for your response

I think it would be worth experimenting a little more with the n=0 parameter. Your n=0 results avoided the 0 power usage results (due to gross v net). Also the shape of the line also reflects the 5 min v 1 hourly updates.

You may want to also look at the ‘mode’ parameter.

[saliya]

Hi folks,

Attached are some graphs from yesterday. PVOutput can show the proper data on a
non-Donator account if the data from the meter is modified before sending to pvoutput.
It would be good if this in-flight modification could be done at the pvoutput end (see #4, pvoproxy data).

1. Solarweb. There seems to be data missing between 23:55 and 00:00 but this is close to authoritative.
   

   

2. Donator service. Hourly logfile upload via Fronius Push. Roughly corresponds to solarweb
   

   

3. Free service. 5-minute upload of inverter and meter data. Consumption (while the sun is shining) seems incorrect to me on this one.
   

   

4. PVOProxy service. 5-minute upload of inverter data as per (3) but the data is modified before being uploaded to PVOutput.org. Consumption data (while the sun is shining) seems correct. I don’t have the exact details of what is modified yet.
   

   

[giampi_crl]

Hello, at the end what was the way to have: “data from the meter is modified before sending to pvoutput.”
What exactly was modified to get the correct data as shown in the last graph?
I have the same issue but the final solution is not posted…