An Enphase Ensemble Installation

svetz

Works in theory! Practice? That's something else

Summer is Over!​

It's official, the curve is on the downward trend!

1628343596726.png
 

WildLeg

Solar Enthusiast
Yo Svetz, was thinking about the best way to add a 3rd party battery/inverter to Ensemble to power a completely off-grid setup. Hypothesized that a big battery and split phase inverter would be best added on the incoming grid side of the enpower smart switch. What are your thoughts? Grid battery would certainly need to be sized right/oversized. Currently ensemble require the grid to initialize the system…IQ8PV may change that requirement. Jack Rickard was using a similar technique to power his shop before Ensemble came along.
 

svetz

Works in theory! Practice? That's something else
,,,the best way to add a 3rd party battery/inverter to Ensemble to power a completely off-grid setup.
Not a fan of that, see the discussion in Incrementally Adding AC Batteries, same basic problem that current sources will bite you without a master controller.

Jack Rickard was using a similar technique to power his shop before Ensemble came along.
I saw that. If the consumption is always more than the microinverter output or the inverter is big enough like Jack's with an empty or battery with head-space it'll work, but the edge cases always get you (e.g., full battery). I still think MID (microgrid interconnection device) inverters will be generally available before too long, or the IQ8s. But, #274 still seems like the most economically viable solution at this point.
 

WildLeg

Solar Enthusiast
Thanks for the discussion. The inverter from the grid inverter would be the master inverter. It wouldn’t necessarily need a lot of kWh. Ensemble can be set to self consumption or whatever use case optimizes the system . When the grid inverter stops working the ensemble battery would provide the power to the remainder of the system. I see a bit of redundancy with a system like this. I agree as far as cost goes recharging with a generator is cheep, but may defeat the purpose. The main goal is to hack the system into not requiring the grid to initialize. At best payback would be the customer charge you pay for every month...and of course resiliency.
 

svetz

Works in theory! Practice? That's something else
...The main goal is to hack the system into not requiring the grid to initialize. ...
AC Coupling is only the reliable solution I know of for that.

...It wouldn’t necessarily need a lot of kWh....
The AC coupling solutions always seem to need to be big to control the microinverter current reversals; Outback talks about some of the engineering challenges in their app note.

...Ensemble can be set to self consumption or whatever use case optimizes the system...
The "no export" idea is interesting, I haven't heard of anyone doing that and I've no clue how well it might work as no export "tries" to keep the export to zero...but transients do get through.
 

WildLeg

Solar Enthusiast
AC Coupling is only the reliable solution I know of for that.


The AC coupling solutions always seem to need to be big to control the microinverter current reversals; Outback talks about some of the engineering challenges in their app note.


The "no export" idea is interesting, I haven't heard of anyone doing that and I've no clue how well it might work as no export "tries" to keep the export to zero...but transients do get through.
Thanks for your thoughts. This was mainly a thought experiment for me. Am currently getting set to submit a system design to Enphase for approval. Its an interesting process.
Cheers,
WL
 

GXMnow

Photon Sorcerer
I have not run off grid for very long yet, but since the last firmware update, it worked well.

My Schneider XW-Pro provided the local "grid" and the Enphase iQ7's had no problem syncing up and charging the battery. The main limitations are that you must have less than 6,500 watts of microinverters on the 6,800 watt rated XW-Pro inverter output, and the battery bank must be able to accept the entire wattage of the solar array. 6,500 watts of solar / 48 volts = 135 amps of charge current. I only have 4,800 watts of solar panels, and with just 16 iQ7's, my solar power clips at 3,900 watts, so it handles it with no problem at all. Th only thing I have noticed is a few of the cheaper LED light bulbs do flicker a little. Looking at the power waveform, I am very surprised there is flicker as the wave form looks very clean, actually less distortion on the wave than my grid here. I am not sure what artifact causes the flicker, and it is only on a few bulbs, most look dead solid.

I did not let my system run off grid long enough to get the battery bank fully charged. My real power failure that did last hours was mostly at night with no solar coming in. I ended up running the battery bank down below 30% to keep the power going. The grid came back before the sun came up.

From my experience, I would expect the iQ8's would also see the Schneider XW inverter output as a reliable grid. When running off grid, it does push all of the extra power to the battery bank without a problem. Once full, it should do the frequency shift, but that has not been tested.
 

svetz

Works in theory! Practice? That's something else
Haven't seen any problems at a high level, but based on @GXMnow's recent experiences thought I'd look into my logs for the last few days, found some interesting bits:

11:02 Cellular Envoy Reboot
11:02 AC disconnect active, reason Envoy dynamic control not running: Set (clear at 11:08) -- huh?
11:06 Watchdog Reset
11:07 PLC security feature enable set to true -- Security features? Don't recall seeing these messages before
11:07 PLC security mode active set to true
11:07 PLC security use 256-bit keys set to false
11:07 Modem Disconnect Set:Modem registration with the service provider failed, Is sim active!
11:08 Clock Adjusted -- It self updates the clock?
5:10 EnchargeSoCMaintenance : Clear -- It's set to 80% SoC, so this probably isn't balancing after all or if it is very impressive

Three of the IQ8s had this sequence, but the power didn't go out and wasn't on backup:
5:09 Islanding​
5:09 Grid-Tied
5:10 EnchargePCUStateGridFault​
5:10 Multi Mode On-Grid​
5:10 EnchargePCUStateThrottled​

Rev #'s are the same as previously:
Envoy: D6.0.102 (a24079)​
Profile: IEEE 1547:2015-WHB​
IQ7+s: 4.27.04 / 4.27.09 and 4.18.02 / 4.18.01 (the replacement microinverters still running the older version)​
Encharge: 20.46​
Enpower: 20.46​
The Encharges are interesting... this snapshot is from this morning... note the "installed date" is two days ago and the "updated date" occurs in the future. I wonder if that's the timestamp it checks to see if there's a new version? Just noticed the "Last Report" occurs in the future too. Clock is probably set correctly as the "detected" field is probably correct.
1631450870050.png
AFAIK it's not hurting anything, but I'll enter a bug report on it anyway.​
Update: Oops... guess you can't go by the release number, currently it's 1.2.4162/20.46 and previously was 1.2.3734/20.46. So it was updated.
 
Last edited:

GXMnow

Photon Sorcerer
Those are some odd log messages. I have not seen those either.

I think the date format is probably Day/Month/Year then it is not in the future. At my work, we have had a few issues when we get requests for a job and they use the Day/Month/Year format and we get all confused. Is this for 2 weeks ago? Using Day/Month/Year, your last report date would be this morning at 8:32 am and the running image updated date puts it at Friday, 2 days ago. The "Detected" time shows the same as the updated, so probably seconds after the update completed.

It looks like my iQ7s were last updated March 9th, but unlike yours, mine still show the detected date as back on July 9, 2019 when my system was first commissioned. The running image in my iQ7s is 520-00082-r01-v04.27.04

Since I have it open, my full grid profile is " (CA Rule21 201902 VV VW FW) "
I am pretty sure the FW means Frequency Watt control. The VV and VW appear to be about controlling reactive power. This makes sense. My system normally reports it is running at 0.95 or so power factor. When I was on backup power, most of the load was my refrigerator, so mostly reactive, and the power factor was down to 0.65
 

svetz

Works in theory! Practice? That's something else
Enpower Relays: https://www4.enphase.com/sites/defa.../Load_and_Solar_Circuit_Control_5_11_2021.pdf
The Encharge Storage System (ESS) must have sufficient power capability to sink all the power from the​
PV system during transitions to off-grid and must source power to support all loads in intentionally islanded​
(standalone) operation. This may not always be feasible. In these situations, auxiliary contacts in Enpower​
can be used to disconnect or shed excess solar or heavy loads​


Generator hookup: https://www4.enphase.com/sites/default/files/downloads/support/Generator Tech brief.pdf
Enphase Enpower smart switch includes a built-in independent generator port which eliminates the need for generator Microgrid Interconnection Device (MID) . It also supports intelligent remote start of generators without requiring any intervention from the homeowner. Enpower takes care of turning on and controlling the 2-wire and utility-sense-based remote start generators when the grid is down.​


Storage Guide: https://www4.enphase.com/sites/default/files/downloads/support/201124_Storage_User_Guide.pdf
The Encharge storage system performs two critical functions in your system. The batteries, internal to Encharge, store energy for later use or for use during a power outage. The IQ 8X-BAT microinverters in the Encharge units provide the voltage and frequency necessary for the operation of your solar array and the electrical circuits in your home during an outage. The IQ 8X-BAT microinverters converts your harvested energy into usable AC electricity for your house. Encharge communicates with the Envoy through a mesh network of wireless signals​
Power Control Systems: https://www4.enphase.com/sites/default/files/downloads/support/PCS-Tech-Brief_1.5.pdf
Power Control Systems (PCS), as defined in NFPA 70, National Electrical Code 2020 Edition, control the output of one or more power production sources, energy storage systems (ESS), and other equipment. PCS systems limit current and loading on the busbars and conductors supplied by the power production sources and/or energy storage systems​
 

svetz

Works in theory! Practice? That's something else
2 Wierd issues
I finally got some time to start looking over some of the updates in the new Enlighten program (their monitoring solution), one of them is a battery report. Oddly, it reports one of them going to ~40% SoC on the 13th. The power wasn't out so no reason to see one battery dip:
1632225016952.png

I've only been self-logging daylight hours to keep watch on the panels, so don't have data around midnight, but the "before" looks about correct (around 80%) and the next morning is a bit low (78.8%). But, the first entry on the 12th shows a soc of 43%. Previously I was treating all the batteries as a single SoC, given the past problem (#226) from the time change and odd messages from the 12th reported above (#289), I'm starting to think I should report each battery independently. Sadly the log messages in 289 don't capture what happened at 2 AM, and that's what started the problem. Either I missed it in review or thought it normal.

So what about the battery graph for the 12th?

1632227954400.png

In the display above, green is battery activity. It started at 80% where it should be at, then at 2 AM discharged to 40%. Then as the sun came up it recharged back to 80% and has stayed there happy as a clam ever since. Any one battery is only 25%, so this affected multiple batteries, since they don't drain to zero it was probably all of them.

One peculiarity is the bottom graph shows a discharge starting at 2 A.M., but the graph above doesn't show the power going to the house or grid.

See the green at the top (discharging) while charging from noon to until 2 PM? As it's on-grid, you'd think it should not be using battery power. But I think it hints at a complicated relationship between the batteries and the envoy. I suspect some of the batteries must have been charged over 80% in the Envoy's desire to get the system as a whole over 80%, and then it had to discharge those over 80% while charging those under.

The messages in post #289 do cover that period, so that's probably what those SoC maintenance messages are about.

Issue 2
One of the new reports (System->Reports->Monthly Net Energy) allows you to export data as a spreadsheet (which is really nice). The power last went out on the 6th, it was just a momentary blip, but reports in the outage log as 5 minutes which you'd expect. But, when I sort on discharging from the batteries and toss all the rows less than 100 Wh and the data from the 12th I see:

Date/TimeEnergy Produced (Wh)Energy Consumed (Wh)Exported to Grid (Wh)Imported from Grid (Wh)Stored in AC Batteries (Wh)Discharged from AC Batteries (Wh)
9/1/2021 4:00​
0​
504​
0​
286​
6​
224​
9/2/2021 4:00​
0​
565​
0​
444​
4​
125​
9/3/2021 4:00​
0​
400​
0​
257​
4​
147​
9/4/2021 4:00​
0​
659​
0​
508​
4​
155​
9/5/2021 4:00​
0​
800​
0​
731​
4​
73​
9/7/2021 4:00​
0​
626​
0​
416​
5​
215​
9/7/2021 17:00​
601​
1306​
0​
743​
145​
107​

So, at 4 AM for the first 7 days (the sixth being an exception), the batteries went into discharge. The Enlighten battery graphs do not show this "blip". After the 7th it starts acting normally (no discharge) until the 12th when it went nutso per above, and then since the 13th all has been what I expect. Possibly something that was fixed in the update and I just never noticed it before.

The 4 AM time just also happens to be the end time for my fake time-of-use tariff. That's a bit tricky to explain, but I set the high rate as 4 AM to 10 PM. So from 4 AM to 10 PM the battery discharges to its minimum reserve (currently 80%). Since the battery is set to charge from solar during off-peak and the sun isn't shining from 10 PM to 4 AM, the battery stays at the reserve SoC all the time unless the power goes out.

Guess I'll call Enphase and see what they think.
 

svetz

Works in theory! Practice? That's something else
Enphase didn't see anything currently wrong and isn't taking action at this time, they're hoping it's a one-off that the patch fixed.
But it looks like two batteries (3194, 3230 (E3)) dropped to under 10%.

Despite the graph showing a sudden drop, my guess is they've been slowly discharging with that 100-200 Wh drop at 4 AM (for who knows how long) and not reporting the true SoC to the Envoy. I suspect when they hit 10% it generating a critical alert to the Envoy where it realized there was a problem.

So, in addition to logging individual battery data I'll add an alert if any of them go below 40%. I doubt it's useful though as the information I get is from the Envoy, what I really need to get is the cell voltages or direct information from the Encharge. The tech support guy could see them, so I know they're out there.
 

GXMnow

Photon Sorcerer
That is very odd to see that behavior, but I am trying to think of why and how this might be happening.
Having your fake high rate set from 4 am to 10 pm might be part of the problem. It can't charge from grid, and it can't charge any time the sun is shining. Could that be confusing the logic?

State of charge maintenance..... Is it possible it was pulling power from one battery and pushing it into one of the others to even out the state of charge? This would explain a state of charge change without pushing power to your loads or the grid. Maybe the system also does a cycle like that once in a while as a test of the batteries? I have no idea if they are actually doing this, but it would make some sense. Run one battery low, check it's performance, then use power from another battery to charge it back up, and test the next battery that was run low. Since it just move energy form battery to battery, there is not must lost/wasted and it could run a decent depth of discharge cycle on one unit at a time while still holding a good reserve in the other batteries.

Have you taken the cover off and looked inside one of the Encharge battery units? I think the spec says that each 3.3 KWH is using a stack of 4 iQ8 inverters, but I am curious if they all are running in parallel, or does each one have a number of cell connected? How many cells are they running in series? Since it is based on a solar panel inverter, I would expect it to be around a 40 volt system. Is there a separate BMS module in there, or does each iQ8 have extra wires to monitor a few cells? How hard would it be to extend out some wires to externally monitor all of the cell voltages?

My second battery bank here has only been online about 2 weeks, and I already miss not having the blue tooth. I trust the cells are doing well, and the BMS will act properly, but on my old battery, it was nice to be able to just pop up and app and see all the cell voltages and the current flow. The new battery bank with the basic dumb Daly BMS units tells me nothing. I have to get in there with a meter to see what the cells are doing. They cost a bunch more, but I may end up changing out the BMS units for smart ones. I have never been good with "faith" that something is working just fine. I like to look at data from time to time and see for myself.
 

svetz

Works in theory! Practice? That's something else
... it can't charge any time the sun is shining. Could that be confusing the logic?
It does charge from solar during the day if below the reserve, you can see that in the graph above. Seems to usually work just fine. You could be right though; when I explained to the Enphase tech why I had it set that way I could hear his eyeballs crossing over into double-vision. He said he hadn't heard of it before but there was no reason it shouldn't work.

Is it possible it was pulling power from one battery and pushing it into one of the others to even out the state of charge?
That's what was happening from noon to two p.m. for sure, but not the drop at 2 AM. I buy that's a mis-report/stuck queue since the batteries suddenly show 5 kwh gone, but from the graph above you can see it didn't get consumed by the house or get exported to the grid... it was just gone without discharging (it's the basis for my theory above).

Maybe the system also does a cycle like that once in a while as a test of the batteries?
I asked about that, they said when it does a SoCMaintenance you can see it drop... but 50% of the battery power is way beyond abnormal for that. He also said the system doesn't need to be set to 100% to do a cell balance, it's supposed to do it once a day. Sorta sounds like an active balancer, which after watching Will's videos makes me wonder (of course, could just be Will keeps getting crap active balancers; he does have an amazing tendency to get broken stuff).

Run one battery low, check it's performance, then use power from another battery to charge it back up, and test the next battery that was run low.
It might be worth running a capacity test on all four units separately. They can't be easily isolated and I'll have to wait until I don't need the Air Conditioning, one unit can't run it. Something for this winter.

Have you taken the cover off and looked inside one of the Encharge battery units? I think the spec says that each 3.3 KWH is using a stack of 4 iQ8 inverters, but I am curious if they all are running in parallel, or does each one have a number of cell connected?
It's really more than 3.3 kWh per unit (I suspect 4 to 5), they limit it so it has the power for the warranted period. Since each unit is an AC battery (defined as a dc battery + Inverter + charger for those unfamiliar with the term) they all run independently and the SoC in one could go to 10% while another is at 100%, but that would be horrid management of course. The "brain" should be looking at the shunt SoC and getting them to share equally. In my previous tests it worked flawlessly (e.g., at the end of the test all units had the same SoC).

There's a "naked" (e.g., cover off) IR picture of the 4 IQ8s in an E3 in post #117.

How many cells are they running in series? Since it is based on a solar panel inverter, I would expect it to be around a 40 volt system. Is there a separate BMS module in there, or does each iQ8 have extra wires to monitor a few cells? How hard would it be to extend out some wires to externally monitor all of the cell voltages?
Taking the cover off is okay, but you have to do some disassembly for the rest, and that voids the warranty so haven't gone there.
The IQ8D specs might not be too different, it's max DC V is 119V.

... I already miss not having the blue tooth....
That's the first thing I learned, you can never have enough diagnostics. At least you can whip out the meter for a double-check when you need to. But yeah, being able to get remote diagnostics and log them is priceless. Love my RTL-SDR when it doesn't hang or windows decides to reset the comm port. Hopefully, the upcoming W11 won't be a nightmare.
 

svetz

Works in theory! Practice? That's something else
This was interesting. I switched the battery profile to full backup this morning, figure
based on the data from the 12th (#294) might be "healthier" to let it run that way
for a few days and watch the logs.

The batteries are still set to only charge from solar, but I expected them to charge
from excess solar (which it's cloudy and I don't have currently).

Instead, it's what you see to the right. The batteries are charging from solar, but the
house is consuming grid, that is it's not waiting for excess solar to charge the
batteries which seems like a flaw.

I'm pretty sure the ability to manually switch battery charging from solar to grid is
new, previously the system decided for you.

Previously when StormGuard came on it charged the battery from the grid. That's
what I want it to do, a lot of times we don't get much warning so you want rapid
charging. Curious if that will automatically switch.
1632313576697.png

Update: Left a comment on the Enphase community page to see if there was a way to only use excess and they came back and said it did only use excess power. But, they couldn't explain the screen snapshot.

Update 2: While in the comments section ran across something useful:
Each inverter consumes around 60mW during the night and an Encharge 3 battery will consume around 20W to maintain SoC(State of Charge) when it is idle.

Update 3: Tried to duplicate the problem, but it worked as expected (excess solar was used to charge). I suspect it's because it's software-driven, the brain is deciding what to do and I probably caught it at the perfect time such that it was a little out of date.
 
Last edited:

svetz

Works in theory! Practice? That's something else
The low SoC battery alerts work! Had a few of them pop up this morning. The Encharge 3 (32988) lost comms for an hour today, so the system reported 0% SoC. Learned from the tech support guy that the encharge reports via wifi, but is controlled via Zigbee.

Losing comms wasn't a big deal, but the battery charged 300 wh while it lost comms which seemed dangerous since I'd left the system at 100%. They said the IQ8s had override to prevent overcharging, what was happening is the "headspace" was being charged. They suggested moving the dongle in the envoy, but that doesn't make sense to me as Zigbee is a mesh network (e.g., the 3 is next to the 10 and the 10 was communicating). Finally, just to be safe, they rebooted everything. I'll leave an update if I learn anything new.

Here's the relevant events:
Code:
"Mon Sep 27, 2021 08:17 AM EDT" "EnchargeSoCMaintenance : Clear"
"Mon Sep 27, 2021 08:08 AM EDT" "EnchargeAGFPropagationDone"
"Mon Sep 27, 2021 08:08 AM EDT" "EnsembleModGone : Clear"
"Mon Sep 27, 2021 08:05 AM EDT" "Device is excluded from the aggregate SOC: Clear"
"Mon Sep 27, 2021 08:04 AM EDT" "EnchargeOnGrid"
"Mon Sep 27, 2021 08:04 AM EDT" "EnchargeZigbeeCommFailure : Clear"
"Mon Sep 27, 2021 08:03 AM EDT" "EnchargeSwitchedToPrivateKey"
"Mon Sep 27, 2021 07:48 AM EDT" "EnchargeZigbeeModuleError"
"Mon Sep 27, 2021 07:48 AM EDT" "EnchargeSwitchedToGlobalKey"
"Mon Sep 27, 2021 07:48 AM EDT" "EnchargeGridTied"
"Mon Sep 27, 2021 07:48 AM EDT" "EnchargeLostCommsWithGWSwitchTOGridTied"
"Mon Sep 27, 2021 06:48 AM EDT" "EnchargeZigbeeModuleError"
"Mon Sep 27, 2021 06:48 AM EDT" "EnchargeZigbeeCommFailure : Set"
"Mon Sep 27, 2021 06:59 AM EDT" "Device is excluded from the aggregate SOC: Set"
"Mon Sep 27, 2021 06:58 AM EDT" "EnsembleModGone : Set"
"Sun Sep 26, 2021 09:14 AM EDT" "EnchargeSoCMaintenance : Set"

This is interesting, the 8:17 EnchargeSoCMaintenance clear is from a set at 9:14 AM from the day before. Seems like a long time, I'll leave an update if I find anything on that too.

Update: So zigbee networks are considered highly reliable in white papers. But in practice, I do see people saying to put the zigbee dongle on a USB extension well away from the microcontroller. Possibly RF from the microcontroller is impacting the dongle?
 
Last edited:

svetz

Works in theory! Practice? That's something else
Was just looking at the local utility power costs, looks it's gone up quite a bit to $0.14496/kWh, an increase of about 9% from July last year. To their credit, the local utility has managed to keep their costs the same, the increases are FPL increases. The increase seems to be fuel prices, even though consumption stayed about the same from last year (up 0.7%) the total $ for fuel went up 60.8% (mostly natural gas which went up >65.6%). Nuclear costs went down 1% (yes, the fuel costs are higher, but so is the production). Similarly, we burned nearly twice as much coal than last year and its $/MWH was down a bit.
1633694954473.png1633695017829.png

FPL's solar was up 50%! More than enough to offset the increase in power consumption, so without solar, the costs would have been even higher!
Embarrassing though the "sunshine" state only gets ~4% of it's power from solar.
 
Last edited:
Top