diy solar

diy solar

An Enphase Ensemble Installation

OMG! You just opened a brain storm.

My cells (when they finally get here) and I have an Overkill BMS waiting... I read it has a programming API via Arduino... I could easily do what you're suggesting with fairly trivial programming on a ESP8266 device. As you pointed out above... could also do a counter of micro-cycles and do a clearing, full-charge every so often to wipe out any memory effect!

Oh! This could be fun!

You need to have the sailors lash you to the mast before you do anything rash.
 
My Enphase iQ7's got an update on Feb. 10, 2021 but it is just "Image 520-00082-r01- v04.20.03" which is still a bit lower than the numbers you posted for your system. I told them I have the Schneider XW-Pro inverter, and to think of my system the same as an Ensemble or Tesla Powerwall2 with the battery inverter for backup operation. Maybe the version they put on yours is specific to the EnCharge?
 
Could be I'm on a different branch of the software, but if there are parallel development paths to me that means bugs down the road when they try to merge features back to a single release.
 
Just realized I missed my panel's first birthday (1/27 was the PTO)! Probably as I didn't start reliably collecting all the data until March. What a crazy year.... guess I need to update the OP too for January....

1613743151029.png

I'll compare the February data of my first two years in March, not horribly scientific but a first pass at the 1st year panel degradation numbers.

Not a real exciting year although we did survive a very weak cat 1 hurricane (our top winds locally were only in the 60s). No major problems and the one minor problem (some microinverters occasionally stop producing until the next day) was patched so the maximum outage now on them is 5 min.
 
Last edited:
The Envoy can support up to 600 microinverters, so obviously all the power isn't going though it. But, it communicates to the inverters via ethernet over powerline, so a signal must be tapped to the power.

One of the things the Envoy can support is zero export. Since the power doesn't pass through it, the microinverters must have their power throttled via an API rather than frequency shifting. If so, with some reverse engineering, I don't see why the over-priced Envoy couldn't be replaced with a something akin to Raspberry Pi with an ethernet over powerline adapter. Might be important for Incrementally Adding AC Batteries too. First obvious problem is there are varying frequencies for powerline technology. Wonder if anyone has experimented with it?
 
Battery test today!
7:41Turned the grid off, Enlighten: Received event, but no indication on main status display, production 110 watts
7:46Foghorn: 58%, consuming 1883 watts, expected runtime 4:08 (4 hrs, 8 min)
7:52Enlighten switches to show off-grid alert, array at 150 watts
7:53Foghorn SoC 58%, consuming 1763, 4:25
8:00email from Enphase saying I'm on battery power, turned off the dehumidifier (forgot it was on)
8:071613829435968.png
8:191613829621746.png
9:00Foghorn: 63% SoC, charging at 2761, fully charged in 1:48
1613829579607.png

So, the system transitioned without a blip (no PC reboot), initially there wasn't enough sunlight so the battery started discharging. When there was enough sunlight, the battery started charging towards 100% (rather than 60%, it's normal SoC for winter). So, all in all no problems.

Extraordinary events
Java:
            "Grid Instability: Set/Clear"  // Interesting these came up while off grid
            "AC Frequency High: Set/Clear",
            "AC Frequency Out Of Range: Set/Clear",
            "EnchargeOnGrid",
            "Multi Mode On-Grid",
            "EnchargeGridModeFromEnpowerMMOnGrid",
            "EnpowerControllerStopRelayPulses",
            "EnpowerGridOperatingFull",
            "EnpowerControllerStartOpeningNFTRelay",
            "EnpowerGridConnectedMode",
            "EnpowerControllerStartClosingMIDRelay",
            "EnpowerControllerStopRelayPulses",
            "EnpowerGridIsGood",
            "EnpowerControllerStopRelayPulses",
            "EnpowerVLoTrip",
            "EnpowerControllerStartOpeningMIDRelay",
            "EnpowerMicrogridMode",
            "EnpowerAITrip",
            "EnpowerControllerStartClosingNFTRelay",
            "EnchargeOffGrid",
            "Multi Mode Off-Grid",
            "EnchargeSoCMaintenance : Clear",
            "Islanding",
            "EnchargeGridModeFromEnpowerMMOffGrid",
 
Last edited:
The Envoy can support up to 600 microinverters, so obviously all the power isn't going though it. But, it communicates to the inverters via ethernet over powerline, so a signal must be tapped to the power.

One of the things the Envoy can support is zero export. Since the power doesn't pass through it, the microinverters must have their power throttled via an API rather than frequency shifting. If so, with some reverse engineering, I don't see why the over-priced Envoy couldn't be replaced with a something akin to Raspberry Pi with an ethernet over powerline adapter. Might be important for Incrementally Adding AC Batteries too. First obvious problem is there are varying frequencies for powerline technology. Wonder if anyone has experimented with it?

Do you have Envoy?
Does it linearly adjust output of each microinverter? Or individually enable/disable them to vary production for zero export?
 
From this document,
the zero export is controlled by the Envoy sending commands to the micros. And it appears to just give them all a percentage factor. It sounds like they are normally told 100% until it starts to export, then it will ramp down, and ask for say 95%, or then 90% as the sun comes up and starts making more power than you are using. It just depends on how much power is being produced and exporting. Since it uses a CT on all of the micros, and on the grid main, it knows exactly how much it needs to adjust. It makes the most sense to have all of the micros throttle together. So if the micros are making 10 amps total, but it sees 1 amp exporting, it just commands 90%, so the micros drop to 9 amps, and zero export. Easy! If more loads turn on and/or clouds block the panels, it will see power coming in from the grid, and then command the micros to raise their output again. The document about it is pretty good, but I did read between the lines and a little assumption.

Check it out for yourself
On page 4 it talks about how it works. It samples every half second (500ms) and makes adjustments every 1.5 seconds. It will react and adjust in less than 4 seconds, Hawaii allows 30 seconds of "inadvertent export", so it is plenty quick.

My power meter appears to store a reading every 15 minutes. So this adjustment rate is surely not needed here. This is much faster than needed to ensure total export at zero for each 15 minute update. But that 30 second Hawaii limit does need some speed. Adjusting every few seconds is easy for a computer, but probably overkill. On my PLC program I was thinking of just 1 minutes samples and 2 to 5 minute updates. No need to go faster.
 
Good morning everybody, happy Tuesday!

Actual power outage this morning, started at 4:13 A.M. according to the event history.

No idea why, naturally the first thought was the start of the zombie apocalypse ( ? ) as it's somewhat eerie to look out the predawn window and see nothing. Which means it's probably cloudy too (no moon/starlight). Then a jogger runs by, lit up like a Christmas tree (zombie bait if ever there was one). Anyway, just the upper half of the keys are out, about 40 miles. They have a nice outage map, but they never say what the problem is or when it'll be restored. Good to know it's not the apocalypse though.

It wasn't completely dark, I could see light coming from someone's dock waayyyy down.

Had the power blip at 6:03, not sure why... lasted long enough I had to reset the clocks. I was looking at the live status when it happened and what was unexpected was the power consumption suddenly jumped up after that for a couple of minutes. Probably stuff like the water heater kicking on, the UPSes going into charge mode, refrigerators turning on the compressors, etc. as they reset.

Wife got up, I can tell as the power consumption jumped up again... yep.... coffee pot on, lights on, hot water tank on... fortunately, I can see dawn starting to break on the horizon... ;)
 
Last edited:
Internet went out, with it lost Enlighten and the at-a-glance view of productions/consumption/SoC. Fortunately have Foghorn right? Nope... Not sure why it didn't work...guess I need to figure out what the certificate error is so I can debug it in Android studio. Hopefully I can duplicate the conditions.

Update: Got Android studio working, apparently I needed to update sdk 29 (although it's using lollipop 21). But, can't duplicate the problem. <sigh>
 
Last edited:
Internet went out, with it lost Enlighten and the at-a-glance view of productions/consumption/SoC. Fortunately have Foghorn right? Nope... Not sure why it didn't work...guess I need to figure out what the certificate error is so I can debug it in Android studio. Hopefully I can duplicate the conditions.
I wish I could just view the same enlighten data straight from the Envoy. It is a pain that it must come from the internet. When my power was down, I lost internet after about 5 hours and Enphase installer Tool Kit also refused to let me look at the inverters. If I could have talked to them, I could have done the grip code change myself to get them back running. I had to connect the Envoy to my Cellular WiFi Hot Spot and have Emphase do it remotely. By the time we got it up like that, The sun was falling fast, and I only got about 1 KWH of charge. Even after the internet came back up, I still could not use the Enphase tool kit for about a day, then it just started working again. It makes no sense.
 
I try to avoid internet and windows based control (at least as the only available method). It is too complex and constantly changing. That ecosystem goes though many generations during the lifetime of lab instruments, photovoltaic systems, automobiles.

If is fine to use software on a windows based PC over WiFi/BT/Internet to monitor and make adjustments, so long as equipment also has pushbutton/display to do the same. And simple data ports allowing communication with any controller we want to implement.
 
I wish I could just view the same enlighten data straight from the Envoy. It is a pain that it must come from the internet.
You don't have to go through the internet; the important stuff at least seems to be available through the API... but yeah.... you need to DIY like foghorn.

The Envoy API commands still seemed to work (ran some in a browser), but it's not as easy as looking at a GUI. I'm sure what's wrong is my crappy programming (writing code is easy, handling all the possible exceptions is real work and not done). Because it's android on a cell phone there's also no log file and the Android GUI was a pain to work with...so I didn't really handle exceptions well. Something must have happened (e.g., a use-case I don't see in normal testing) such that it was throwing an exception causing it to hang/loop.

The code's in java, not quite the same code as what I used for logging (used J8 with lambda expressions all over, had to convert them all for Android). Perhaps I'll go back to the original and rewrite it as a PC app. Possibly recompile for an android tablet, wouldn't feel bad about a log file then. Have to think about it.

...It is too complex and constantly changing...
It is certainly subject to change at any time and without warning. The Enphase API responses are in JSON, so that makes it somewhat easier/protected. Haven't seen any of the stuff I use change in the first year that I can recall. What makes it fun/difficult is there's no guidebook to explain any of it. Some of what's in the API post I haven't seen documented anywhere else. If I could have debugged in Android Studio when the problem occurred probably could have found/fixed it.... my bad for not fixing android studio problems earlier (or their bad for messing it up in the first place depending on how you look at it ;-).
 
Last edited:
Update the OP with the February numbers. No reason to do a degradation calculation since I made more energy than last year (1% gain). I was hoping weather difference would average out over a month, but this is just proving its too variable. Given how wet the summer was even a yearly comparison probably won't be good enough.

The data logger stopped, hasn't been collecting for a week now.

Turns out the kind folks I've been mooching solar data from went off the air and lazy coder that I am didn't put an exception catch in (they seemed so reliable!). Fortunately, two more weather stations appeared (Thank you neighbors!). So now I have a primary, a backup, and a log message if they're not reachable. Also, it won't block the program if the web is out. I suppose I should really get my own setup so the internet out isn't a problem.

According to the program, my panels are running at 106% compared to their solar irradiance meters and we're about an hour past solar noon. From panel data I can see none of the panels are in the shade now. So, at least I can be fairly sure their sensors are not in the shade. Actually the panels should all be out of the morning shade too now, I'll have to check that tomorrow morning.
 
Last edited:
Probably one of the most frequently asked questions on the Enphase forums is how to reconnect the Envoy, but this one was a bit different and thought I'd repost it as I learned something:

When will cellular as backup to WiFi be fixed?
I always thought that the cellular with its 6 hour reporting was a backup to the wifi (with 15 min reporting).

But in looking through the forums it seems you have to pick one or the other and that cellular takes priority. Is that correct? If so, any timeframe on when it'll be fixed so one can work as a backup to the other?

Enphase response:
Hi, Yes Cellular takes preference. As far as I know we had not had a request to have it as a back up with both connected. I will pass on the idea.

I doubt there's that much profit in it for the installers, so possibly a training issue? I have both, but it was my understanding it was a backup.

Here's mine... looks like WiFi can be set to be dominate.

1615211326396.png

Here's another interesting post asking about Grid Instability errors, they have an image from a Sense monitor showing a half dozen dropping out at a time. Have to keep a watch on this one to see what they say since I see those errors too.
 
Last edited:
Here's another interesting post asking about Grid Instability errors, they have an image from a Sense monitor showing a half dozen dropping out at a time. Have to keep a watch on this one to see what they say since I see those errors too.
How long ago was the last post on that?
When I complained about the inverters locking out, they did end up doing a firmware update about 2 weeks later. I have not had a drop out since, but I also have not tried a power failure simulation since the update. They did tell me I should be able to switch them back to the "CA Rule 21" grid profile. I will try a power fail test in the current mode and see how it acts, and then change the profile and try it again. I just need to find a time I can shut down the PC's incase the test goes bad.
 
How long ago was the last post on that?

I usually skim through once a week, sorting by most recent. Let's see... that was March 6th, but that doesn't mean they have the update. In checking my event history, I had 10 Grid Instability: Set events yesterday. So for me the problem still occurs, but they reset after 5 minutes with a grid profile reset (that's what they're seeing in their graph too, although it looks like they have more going out for longer). It would be interesting to know what's going on (for example, if it was from the water heater turning on perhaps I could do something about it).

Out of the last 12 events over two days it was from 6 inverters. There might be some patterns, but the sample set is too small to be sure.

Inverter# OccurrencesTime
A33:19 pm, 9:19 am, and 10:27
B112:47
C212:41, 11:15
D211:17, 12:02
E211:11, 10:05
F29:20, 9:32

I'll check again in a few days. If it's the same inverters over and over perhaps they're just faulty?
 
Last edited:
Back
Top