Accuracy of the Flukso solar measurement.
I start to wonder how accurate is the solar measurement with Flukso.
My setup for measurement and comparison:
A: kWh dinrail meter with S0 output to Flukso
B: kWh dinrail meter with S0 output to Qurrent ( dutch Site )
C: measurement to SolarEdge Portal
D: measurement to Pvoutput.org taken from Flukso
C:Monthly measurements from A and B manualy
When I compare A and B the difference is about 2,45 kWh per month, there fore
a assume that the dinrail meters are within the specs.
When I compare the output of B with the output of SolarEdge portal, B is of
by 350Wh at 12,5Kwh output.
So i can assume that the measurement of the portal is also with in the
limits.
Now for the tricky part
When I compare the Measurments from PVoutput.org ( input by A ----> Flukso
----> pvoutput.org) to the SolarEdge portal, PVoutput is approx 28% to high.
there for the measurement of Flukso is to high.
Not so strange when look at the graph of Flukso it show 3402 W from a 3060Wp
installation
At this moment I'm correcting the output from PVoutput.org so that it comes
more in line with the output of the SolarEdge portal.
My question is there away within Flukso to correct this.
| Attachment | Size |
|---|---|
| Output from Inverter Output from Flukso.jpg | 74.74 KB |
| Spike 30000.PNG | 56.12 KB |
If you are using an S0/impulse meter the FLM is not measuring, but counting pulses; thus, accuracy is directly given with the factor to valuate a pulse - what is your meter's specification? What is the pulse factor you configured in the FLM?
Be aware that there is (still) the issue with "broken factors"; see https://www.flukso.net/content/setting-correct-constant-s0
@Gebhardm
A: kWh dinrail meter with S0 output to Flukso --- Eltako WSZ12DE-32A Accuracy class +/- 1%
2000pulse/kWh
Flukso constant 0,5
B: kWh dinrail meter with S0 output to Qurrent --- Carlo Gavazi EM 10 Din MID Compliance ( annex MI- 003 )
1000pulse/kWh
Configuration looks reasonable.
You may check what the FLM actually counted via the built-in MQTT-broker; install mosquitto (http://mosquitto.org/) and subscribe to the counter topic of the respective sensor:
This shows what is currently stored in the FLM like what your meter displays (kWh); write this up together with what the display of your meter shows; do the same some time later and see if the accrual matches; this should provide you evidence where the error might be.
See https://www.flukso.net/content/mqtt-mq-telemetry-transport for more information on MQTT utilization.
Also check if the cable is connected correctly on the meter as well as at the FLM; certain users report that the wires were not tightened in the terminals what produces misbehaving pulses.
I have uploaded a screenshot it clearly show the difference, the shape of both graphs are the same but the peak of Flukso is about 1000W higher.
I checked the wiring as suggested by Gebhard no faults where found.
Any thoughts on where the fault can be found.
Is it possible that the new software might solve this problem
You should dig a bit deeper to try and find the root cause of the issue. As suggested by @gebhardm, compare the counter value on the Eltako's display to the one ouput on the FLM's MQTT bus. Take two snapshots with e.g. a day in between, and compare the result. The FLM counts the pulses output by the DIN-rail meter. So the kWh reported should be exactly the same, give or take a pulse.
Another action you could take is to hook up the EM10 to the FLM, change the meter constant to 1, and let it run for a day.
@ Icarus75
As suggested by you i connected the EM 10 to the FLM.
At the same time I got the minute tab working on all my devices.
Those two combined gave me the opportunity to see what's going on
To my surprise everything seems to be okay, I compared the output from my SolarEdge to the output of the FLM and found 400 watt to be 400 watt on the FLM minute tab.
After I observed the situation for a couple of minutes out of the blue a spike of 800 watt showed up on the minute tab being double the amount of the SolarEdge display.
When I compared the output of the SE and the FLM later that day I found that when the output reached about 1800 watt, the doubling occurs 2,3,4 times every minute with a high spike of 30.000 watt.
See attachmend.
Therefore it's no wonder that the output on Flukso ( cumulative ) and Pvoutput are higher than my readings on my SolarEdge Portal
What kind of cable are you using to interface your e-meters with the FLM? Make sure that they are drilled wires and as far away from any high-voltage/AC-lines as possible; this rather seems like you get a spike-in from somewhere else (or the terminals do not get required contact to ripple in disturbance)
@Gebhardm
Cables that I have been using:
Telephone wire, Shielded UTP, 2x0,5mm2 , and now 2x1,0mm2 shielded, shield connected to ground on 1 side.
Length 1,80 meter
Disconnect all other measurements water, gas, mains.
No result on solving my "problem".
If possible can you send me by personal mail, a howto on the " built-in MQTT-broker".
I'm always willing to learn but need some guidance
Mail is on its way - an easy instruction can also be found in https://www.flukso.net/content/mqtt-mq-telemetry-transport - based on these experiences the gauge display came into this world utilizing some pre-work of others... See Pietro Spinas attempt to describe in detail what to do on a Ubuntu machine https://www.flukso.net/content/how-mqtt-nicely-designed-gauges and my explanations on github - for a dedicated panel view you may use https://github.com/gebhardm/flmdisplay/tree/master/panel - this works on Mac, Linux, Raspberry Pi, and similar machines more or less "out of the box" (and brings some fun learning to set up an own app environment)
@Gebhardm
Please give your reaction on my cabeling.
Cables that I have been using:
Telephone wire, Shielded UTP, 2x0,5mm2 , and now 2x1,0mm2 shielded, shield connected to ground on 1 side.
Length 1,80 meter
Disconnect all other measurements water, gas, mains.
@wizzopa - the cabling seems O.K., I actually run my solar inverter pulse-emeter with an unshielded wire of around the same length than yours.
You state that the peaks occur with both meters, so also at the EM10; do the peaks occur randomly or do they show a certain pattern? Can you try to look at the e-meter and the minute tab at the same time? As the e-meter may show an LED flash you might bring this in sync with the display on the minute tab - this is to sort out where the error occurs - on the e-meter (rather unlikely when the peaks occur on both meters), through the cabling (does a loose cable produce peaks?), or by the FLM itself (do peaks occur on no sensor connected, but pulse input enabled?)
This is really a strange behavior...
In my quest to find the problem of the spike/peak/pulse in the minute tab, I like to sum up things i have done to find it:
* Comparing measurement ( See topic opening )
* Giving specs from the Din Rail pulse counters.
* Changed the input to the FLM from Eltako to Em 10 Carlo Gavazzi.
* Bought a new EM10 Carlo Gavazzi
* Changed for the third time the cabling and check if all screws where tight
* Checked if there is any relation between the blinking LED and the spike/peak/pulse.
This afternoon will the sun was shining:
* Replaced the Eltako for the New EM 10.
* Disconnected the old EM10 and ran the new one alone.
* Replaced the new one with the old one.
* Connected the the Eltako and the old EM 10 in Series.
Non of the above made any difference the spike/peak/pulse keep on coming.
* Disconnected the screw terminal no measurments no spike/peak/pulse
The only difference i found that when the Eltako is connected to the FLM the spike/peak/pulse are double the amount of the EM 10, that is not so strange,
Eltako being a 2000 puls/kWh and the EM 10 1000 puls/kWh counter.
I still think it has something to do with the FLM it self or .........
You should dig a bit deeper to try and find the root cause of the issue.
How deep must i go.
If any of you Fluksonians wants to look inside my FLM remotely with Mosquitto, let me now and we will setup a Teamviewer connection to do so.
I humbly think you digged deep enough - maybe as a last guess before Bart has to take over...
Did the spikes occur just recently or from the very beginning of your installation?
There were reported not correctly plugged sensor boards; so, just as a guess and if you feel safe: Open the FLM and check the correct fitting of the sensor board or if you may spot any indication of malfunction (the tiny parts may prevent to really see if there is a cold lead on a transistor or resistor) - ground yourself before doing so to unload any potential static charges...
That's it with my Latin from remote; Bart, do you copy (refreshed from your vacation)?
@Gebhardm
It all started some months ago. with the comparison of other sites on Pvoutput,
I noticed that my pv installation produced more energy than reasonably was possible.
If noticed that the power was higher than 3060Wp which is is impossible for my location.
At first thought it had something to do with cabling so i changed it no result, found a way to offset the power on pvoutput up to 28% that worked for energy/power.
Still i was not satisfied with that solution, than the topic on the minute tab and the solution for that , implemented it and found spike/peak/pulse in this tabe.
@icarus75 @gebhardm
I think i have done my part in trying to solve this mystery, IMHO it is clearly coming from the FLM
At this moment my installation produced about 13,276 Kwh, data from Qurrent and Solaredge Portal,
reading from Pvoutput 16,329 kWh
reading from Flukso 16,866 kWh
almost 3 kWh more thanks to spike/peak/pulse.
( Time difference between readings about 5 minutes. )
Bart is this a challenge for you?
These spikes are most likely due to false triggers on the pulse port. It looks like you've got the PV meter hooked up to port 2. Have you tried connecting it to a different pulse port on the FLM? Port 5 does not seem to be in use. Make sure to set the meter constant correctly on that port as well.
@Icarus75
As I thought all along "IMHO it is clearly coming from the FLM"
You beat me to it, this action was the last one on my list, I tried to avoid this the reason "data loss"
It's working.
Me being stuck with a FLM without a spare port that was intended for my new solar panels 2750 Wp that will be installed next friday.
Is there any kind of warranty?
@gebhardm
Thank For....... I humbly think you digged deep enough.....
and sticking by me during my quest.
@ Wizzopa Let it run for a while to make sure the glitches are really gone. You can then return the FLM to FluksoHQ and we'll either repair the FLM or provide you a new one.
Please note that it was crucial to find the root cause of the issue first. This is not a matter of trying put the blame somewhere else. It's just that in order to take appropriate action, we have to know what exactly is going wrong.
@Icarus75,
Thank for your positive reaction to my warranty request
If possible please send a new FLM, so I can start collecting data from my new system on friday september 5 without de loss of data when changing on a later date.
I am conducting a experiment with an East West configuration total 5750Wp on a 3000W inverter.
So It's important to me to collect data from the beginning..
You are absolutely right when it comes to finding the root cause of the problem I'am doing that on a daily basis being a maintenance engineer in a pharmaceutical company.
Yeah I have the same problem! Only my output is 14% more!
It is so weird... I thought I'm crazy... Tried almost everything, at latest another port, and voila... Fixed. I'm guessing more people have the same problem, only never checked it...
I saw you have a constant of 0,5 and 28% difference in output.
I have constant of 1 and 14% difference... It seems there is something wrong inside our flukso's....
Wizzopa, is your problem solved and how?
Hi Hans,
Comment by Icarus/Bart
These spikes are most likely due to false triggers on the pulse port. It looks like you've got the PV meter hooked up to port 2. Have you tried connecting it to a different pulse port on the FLM? Port 5 does not seem to be in use. Make sure to set the meter constant correctly on that port as well.
and its still working.
But the FLM needs to be send to Flukso for repair.
Comment by Icarus/Bart
Wizzopa Let it run for a while to make sure the glitches are really gone. You can then return the FLM to FluksoHQ and we'll either repair the FLM or provide you a new one.
It started with my current clamp from my PV on port2. The grafic started at 200 watt! It didn't came under the 200 watt. And it stopped at 200watt. Okay maybe the clamp is not a good option for a PV. Then the din rail meter with s0 pulses gives 14% too much at port2. That should be the wiring. At last my wood stove water didn't have a good counting at port2. See the pictures on my topic. I'm waiting for a resolve now...
I'm not waiting for a new flukso...
I want it just resolved.
Maybe with port6, but I'm awaiting an answer...
Port 6 is just for Dutch a smart meter connection.
Ordinary pulses shall not work on this port.
Hi Hans,
please send a reply to the mail i sent you
I never got any email wizzopa. I have send to someone else by the contact form and he never respond either. When I send it, there were 2 warnings. You're mail been send successful. And your mail didn't got send, contact the administrator. You can send it to hanskraayeveld at hotmail.com
I also never get the mail service work here ...
Perhaps Bart could tell us what the problem is ?