Enecsys Monitor Reactivity - Experimental Evidence

[tc847]

For those of you not familiar with the Enecsys monitor, it is provided with Enecsys micro-inverters only and can display information about each individual panel: specifically Current Power and Power Generated (over various time frames, today, week, month, year, lifetime).

The process works like this:
- each individual micro-inverter sends (wirelessly) information to a device inside the house which is connected (via cable) to the home network;
- from there the information is relayed to Enecsys servers (probably situated in the USA but not confirmed);
- via a web interface a user can then monitor the information being relayed by the Enecsys database.
So, to be clear, the information does not come directly from the micro-inverter to the PC being used.
The monitor software also display the local time and weather but it is not clear where this information is being obtained as I do not believe the micro-inverters are measuring ambient temperature or other meterological data.
When a system is installed a diagram of the panels is sent by the installer to Enecsys such that they can configure the panel display on the monitor to match the physical layout of panels.  In this way when looking at the monitor the information displayed for each panel (which is numbered) can be visually reconciled with the physical layout.

Under normal circumstances the monitor will refresh its display (automatically) once a minute.  It appears that the 60 second cycle is started at various points as you move between views on the monitor.
What may be less obvious to many users is that if you re-click on the same view you can refresh the information more frequently than once a minute (without impacting the timing of the 60 second refresh cycle).

There has been much debate about the speed of response of the monitor to the change in conditions of the panel and this experiment was designed to establish the frequency of updates of information.

Equipment used
10 x Astronergy CHSM6610P 225W panels with 10 x Enecsys SMI-240W-60 micro-inverters 150 degrees 35 degrees
Sony S1 tablet using native browser to view the Enecsys monitor
Citizen wrist-watch with yellow second hand
Flattened 750g Shreddies cereal box to create shade sufficient to cover between 6 and 9 panels depending on angle and distance

Method
By creating and removing artificial shade from an individual panel using the cereal packet and establishing the time to reaction by repeatedly refreshing the monitor.

Findings
Initially when creating artificial shade the reading in the monitor took a random amount of time to be amended (always under a minute)
After a few attempts it became clear that the update for that panel was happening at exactly the same time each minute.
That is, for the first panel, the update was taking place at xx:45 each time it went round.
The impact was identical whether I had created/removed shade at xx:50 or at xx:40 (ie 55 seconds or 5 seconds earlier).  {Note: experimental conditions here did not allow for greater accuracy in timing).
I moved to a different panel and established similar results but at a different time (in this case xx:55).

Conclusions
Each panel is transmitting information about its state exactly every 60 seconds.
Each time information is sent it is the information as at that moment in time (not an average for the last 60 seconds).
That information is reaching the user's monitor in "near real-time" (within less than 5 seconds).
The monitor will show the most recent reading for each micro-inverter whenever the screen is refreshed.  Therefore the aggregate reading of the monitor will be combining figures from each micro-inverter that are between 0 and 60 seconds old (plus the transmission delay time which appears to be negligible).

I hope that someone will find this interesting and/or useful.  Otherwise my meerkat impersonations using our dormer windows will all have been in vain.

TC

[Wickham]

That is what I found, too.
Although the Enecsys website data is refreshed every minute, the panel micro-inverters send data in rotation out of sync, so some panels may just miss the website update and you don't see the change in their data until the next website refresh.

[Iain]

Hi
Just curious.
Does that explain why (as some have reported/commented on) there can be quite a difference between the logged total Enecsys output and the total generation meter output?
Iain

[HPSauce]

Thanks for doing that set of experiments tc. I call our neighbour a meerkat. They're so cute.

Did you get time to try the effect of shading various numbers of cells on the panel?
I am not familiar with your panels. It may be useful to give a short summary of their key specs if you do those tests - OK I'm being lazy.

Iain, pretty sure this has no bearing on the difference between the logged total Enecsys output and the total generation meter output question.
This only becomes really significant under low output conditions.

[clockmanFR]

tc847,
Very nicely documented and thanks.
I had been toying with this concept for some time now, but the initial failures and poor quality/workmanship put me off going down the Enecsys route.

[series530]

Thanks TC - not easy doing those experiments. I'd love to do the same but I would break my neck in the process.

The rotational approach makes sense. I thought though that the inverter was accumulating information constantly. I just wonder if the data being sent is the average of the data over the minute interval rather than a snapshot at that point. On a cloudy day the amount of energy hitting the panel could be varying quite considerably even over a minute and a fair bit of data could be lost. Either that or the data is a snapshot and this may explain some of the error between the website and the PV meter.

Secondly, I have it on authority that the delay between data being transmitted to and being presented on the website is "up to 15 minutes but normally a lot less than that". I get the feeling that E are just covering themselves. I have, however, noticed that there can be a huge lag between my system shutting down (as seen on my optismarts) and the monitor site saying that there is no power being transmitted.

I don't wish to question your data. I am really appreciative that you have done the work. I just want to pose a few questions and seek feedback - let's hope though that we don't end up with a mark two version of the mega thread that we all know about from earlier in the week.

[tc847]

HP, yes I did a further set of tests on cell shading but I feel that the results (surprising) belong in a separate thread as they don't really relate to the monitor, they relate to the micro-invertors and/or panels (and I honestly don't know how to attribute my findings).

Series, I mentioned in my intro that the monitor displays 2 types of information: current power and power generated. My experiment related to Current Power only as detecting an impact on Power Generated would require a much longer time span.
The reason that I concluded that for Current Power the value is a snapshot is this. Having established that the update for a given panel was at ss seconds using my second hand, it made no difference whether I applied or removed shading 5 or 35 or 55 seconds ahead. If it were an average the length of time that shading had been applied would have impacted the reading.
As for Power Generated, my assumption is this. As well as sending Current Power, the micro-invertor is also tracking power generated *continuously* and sending a new total each time. Whilst I have no experimental evidence for this, I have read in another thread that people who have been away and who have switched off internet connection have found that they see a huge "catch-up" figure being applied once re-connected. I cannot think of any other explanation for this reported behaviour.
As for the shutdown time lag I have noticed a similar result. I am suspecting that the ability to detect the difference between 6 or 7W and nothing is harder for the software than the135W to 20W shift that I was achieving.

Off back to my turret where I have something on the slab.

TC

[Wickham]

The "catch up" has happened to me several times when my internet connection has been off or down. I once got 15.8kWh in the first hour recorded after the internet was restored (12pm to 1pm) for a 3.04kWp system!

However, the data sent in the "catch up" could be just the accumulated data from the 60 second snapshots or the continuous data. I suspect that it would make hardly any difference even if a dark cloud shadow moved over a panel for a few seconds so that the 60 second data missed recording that short reduction in generation.

[jez54]

I'd like to add my thanks to tc for the results of that useful experimentation.
For whatever reason I did not manage to do my own experiment here at the weekend.

Enecsys' people advised me verbally on Friday that: (a) their Monitor is not - and indeed was never intended to be - a measuring tool of comparable accuracy to an OFGEM approved generation meter; (b) this is so because the Monitor power measurement is dependent on the combined effect of the tolerances of the individual elements of the inverter and their "commercial" components (think what happens to the accuracy of a resistance bridge if the 3 reference resistors have significant and unknown tolerances), and (c) consequently Monitor data does become markedly less accurate once the output of the panel reduces below 40-50% full output, and (d) they also acknowledge the conversion efficiency of the inverter is reduced below 40-50% full output.  Without wishing to invite unhelpful discussion, I might hazard a guess that (d) may also be true of other inverters, only we've had neither the cause nor the means to discover?

I am certain that this is sufficient explanation of my original problem - the discrepancy between my Monitor output and my generation meter - and to that extent (only) I am satisfied.

The much wider questions which this investigation has raised in relation to my system remain open, although I now have a plethora of new information to process, but until I receive the final written report from Enecsys, which I expect on Wednesday or Thursday this week, I shall not be in a position to reach any firm conclusions as to what options I actually have or what steps I should take next.  This is still a very sensitive commercial matter, for me and others, and therefore I ask that any discussion remains appropriate.