diy solar

diy solar

XW Pro off grid AC coupling nightmare

Consumerbot3418

Fitting square pegs into round holes... for fun?
Joined
Jan 17, 2021
Messages
188
Location
Midwest
I could write a novel, but I'm going to do my best to try to keep this first post as concise as possible. More details likely to follow later in this thread.

I have a pair of XW Pro units hanging on the wall in my garage, along with about 20kWh of LFP batteries, and I'm basically unable to use them at all.

I started with a DC Solar trailer (2x Sunny Islands with 2kW PV DC coupled via Midnite), and added a 7.7kW Sunny Boy. Been running that combination successfully off-grid for about 1.5 years. Ran into some flicker issues when I added an east array of 3kW of Enphase IQ7s, but that was resolved with a grid code update.

Right now, I've got the Sunny Islands' AC coming into the Schneider PDP as "grid", with the Sunny Boy and house loads on the load side. Everything works perfect in bypass when XW is off/standby. (pointing to something unique about the XW Pro or its associated AC/DC cabling)

If I try to make use of the XW Pro units, though, everything goes crazy, intermittently, usually triggered by a load switching on or off--very bad noises from the XW Pro, horrible waveform on the oscilloscope, and UPS switching.

I have tried:
  • different firmware versions on XW
  • running just a single XW in backup mode with only Sunny Island attached, and EV charger as test load
  • different grid codes (on XW and Sunny Island)
  • connecting Sunny Islands to AC2 (gen) input on XW
  • checking all AC and DC cable connectors for tightness, heat under heavy load
  • contacting Schneider
There may actually be two different issues, possibly unrelated, as I don't know of anyone trying to use Sunny Islands as a "grid" input. I'm not even really committed to setting up that way. I'd be satisfied if I could just get XW Pro to work well in backup mode with the AC-coupled Sunny Boy.

In bypass mode (with only Sunny Island AC on the XW input), I can still charge from "grid", or send power to it. When I try to send (ie contribute to house loads from the 20kWh battery), it sometimes seems like it loses sync with the AC waveform, yielding sounds of frequency sweeps and even "clunk" sounds from XW, as the lights in the house start blinking, UPS switches, and o-scope loses sync.

In backup mode, it's somewhat different. When the XW helps supplement AC PV (invert mode), things seem alright. But once there's excess PV, it gets dumped to battery, as expected. If I try to activate my EV charger at the same time PV is getting dumped to battery by XW, I end up with something like this:

bad waves.jpg

I couldn't be more disappointed with Schneider's support. After a phone conversation, I was asked to email specific information. After 9 days, I got a reply asking if I'd tried downgrading the firmware, how old the system is, and when it started to happen. After my reply, it took another 12 days for them to respond with a referral to a distributor.

I really like my AC coupled system (minus the XW), which stacks the 12kW capacity of the Sunny Islands with the 11kW combined AC PV, when the sun is shining... But I'm considering going DC coupled, or getting rid of these XW units. I initially wanted to go with Victron Quattro, but don't have a US grid code, and I had hoped to one day have an inspectable system, and be able to grid tie... It remains a lofty dream at this point.

I'd be really grateful to hear from anyone that's successfully running XW Pro off-grid with AC coupling, especially details of firmware, settings, and so on. I'd also be thankful for any advice or suggestions that anyone wishes to contribute.
 
Last edited:
At some point, I probably should have kept a journal of all the various configurations I've tried, as there are so many variables. Right now, for the sake of simplicity and troubleshooting, I'm planning to concentrate on trying to get a single XW Pro working in backup mode with just the Sunny Island, and a test load.

I had suspected an issue on the DC side of things, but I've seen the misbehavior happen with very few amperes flowing in or out of the battery, while I've been able to pull/push well over 100A without any symptoms. The erratic and intermittent nature of the problem is pretty frustrating. I really don't want to feel like I have to babysit these units while they're in operation, to avoid risking damage to my appliances and so on.
 
Have you tried telling XW its input (from Sunny Boy) is "Generator" rather than "Grid", so it won't do anti-islanding and won't backfeed?
 
Thanks @Hedges for the reply!

Yes. I should have mentioned that I did try using the Sunny Island on the generator input, and still had problems, although I’m not sure I remember what the exact behavior was when I tried that.

I’d really like to retain the ability to backfeed the Sunny islands and keep their lead acid battery charged, but it still seemed worth trying just to see if it helped. But it didn’t. I've revise the original post to reflect this.
 
Last edited:
OK, two other alternatives could be XW as generator into SI, and XW as grid into SI.

I guess you're trying to have two cascaded battery systems for some reason. I've thought about how that could be done with all SI; one guy wanted multiple cabins but also some power sharing.

And then there is the "Chargeverter" approach.

I've been thinking about rectified AC fed into a Sunny Boy (would want soft start or precharge to avoid caps being a dead short on AC input.) Without a PF correction front end, this would be a poor PF load.
 
I run 6kW of Solar Edge AC coupled to my Schneider XW Pro. I've run off grid for about 5 days in total with no issues.

Given our sample size of two, plus about a 1/2 from @GXMnow (but he may weigh in otherwise), I'd say the issue may be with the Sunny Boy.

However, now that I've read your post again, I'm not sure I'd run the output of one battery inverter to the grid input of another and expect perfect operation. That's probably outside of the design parameters. Try switching the Sunny Island for the XW (with that XW acting as grid) and see how the system responds. I'll bet it still doesn't like it but in a new and exciting way.

What is your goal on this set up? Seems odd or maybe useless.
 
I guess you're trying to have two cascaded battery systems for some reason
What is your goal on this set up? Seems odd or maybe useless.
Well, the idea was to have redundancy, so I'd no longer be dependent on that trailer, giving me the ability to lend it out for relief efforts after tornadoes or storms. Or perhaps sell it, eventually... Meanwhile, having the ability to fall back on its lead acid battery if the LFP gets emptied is another bonus.

I'm not sure I'd run the output of one battery inverter to the grid input of another and expect perfect operation. That's probably outside of the design parameters.
I had the same thought. And I'd settle for the XW Pro working in backup mode (off grid), AC coupled with Sunny Boy. But even that very simple config is extremely problematic, much to my chagrin.

OK, two other alternatives could be XW as generator into SI, and XW as grid into SI.
This might work, but I'd have to run another set of conductors out to the trailer (right now, it's about 50' of 4x#6), and figure out a bypass mechanism when I want/need to take it offline.

And then there is the "Chargeverter" approach.
This might be my next angle. I omitted this detail from the initial post, but I've got a pair of Eltek Flatpack2 HEs (2x2000W) controlled with arduino charging the LFP battery (adjustable voltage and current, controlled over wifi--I need to make a thread about this!). For whatever reason, the signal on the scope looks much cleaner when they're in operation, compared to the XW Pro's charger. The XW Pro charger takes a bite out of the sine wave every 2nd zero crossing (once per full cycle), and the EV charger does the same. Combined, it seems to be enough to cause XW to lose sync. Something to do with power factor?

Here's a fun 15 second video. Maybe best ignore my commentary in the audio, and just know that when I switch off loads in the house, it frees up PV that XW tries to sink into battery, ie it switches from inverting to charging. If I let it go longer, it would eventually settle and look like the photo in my initial post.
 
Last edited:
I've been thinking about rectified AC fed into a Sunny Boy (would want soft start or precharge to avoid caps being a dead short on AC input.) Without a PF correction front end, this would be a poor PF load.
I'm curious to know what the application of this would be?

I think my plan, moving forward, will be to run the XW Pro and Sunny Boy, primarily, and use my homemade "chargeverter" to grab power off the FLA batteries on the trailer, as needed. It doesn't give me ability to send power back to it, though, without another set of conductors. And first, I'd need to get XW and Sunny Boy to play nicely with one another. Based on what @Cmiller said in another thread, though, it should at least be possible.
 
“There may actually be two different issues, possibly unrelated, as I don't know of anyone trying to use Sunny Islands as a "grid" input.”

Fairly certain that @toms waterfalls SMA into Victron.
 
“There may actually be two different issues, possibly unrelated, as I don't know of anyone trying to use Sunny Islands as a "grid" input.”

Fairly certain that @toms waterfalls SMA into Victron.
Another detail I omitted is that I did precisely this with a pair Victron Multis, and no problems whatsoever.

Really would've preferred to stick with Victron, but radio silence from them about releasing a US grid code, or anything I might be able to use to pass inspection and get net metering on grid made me opt for what I thought was the next best option.

And, in fairness, it may be the "grid support"/grid code that's causing the flakiness. I can imagine the inverters are both pushing back and forth, trying to correct power factor/voltage, but overdoing it, causing oscillation. Or something like that... But this kind of stuff starts going beyond my limited comprehension. I feel like the image in the first post would probably mean something to someone much smarter than me... like @RCinFLA ? ;)
 
Last edited:
“There may actually be two different issues, possibly unrelated, as I don't know of anyone trying to use Sunny Islands as a "grid" input.”

Fairly certain that @toms waterfalls SMA into Victron.
I don’t think I can offer anything useful here i’m afraid. I’m completely off grid, and I go the other way by providing the Sunny Island generator input from a Victron system.

I’ve seen plenty of issues with getting systems to play nicely with the grid, kind of glad I don’t have to worry about it.
 
Well, the idea was to have redundancy, so I'd no longer be dependent on that trailer, giving me the ability to lend it out for relief efforts after tornadoes or storms. Or perhaps sell it, eventually... Meanwhile, having the ability to fall back on its lead acid battery if the LFP gets emptied is another bonus.

I think SMA Sunny Island + Sunny Boy will be the best grid-backup system you can have.
What it won't provide is peak/valley shaving aiming for zero export (something we in California will need for NEM 3.0), nor backfeed from battery.

So I'd put the XW on the trailer or sell it.

But then, I'm an SMA Slut and hoarder.
 
So sunny boy is ac coupled to xw pro, which is in turn ac coupled to sunny island? Sounds like fun.

I suspect the sunny island is sending some signals the xw pro can't handle. Have you measured the sunny island frequency when you get these issues? Your image showed 69.9hz? 😲

Have your tried connecting a true grid input to the xw pro and if so do you still have these problems?
 
So sunny boy is ac coupled to xw pro, which is in turn ac coupled to sunny island? Sounds like fun.

I suspect the sunny island is sending some signals the xw pro can't handle. Have you measured the sunny island frequency when you get these issues? Your image showed 69.9hz? 😲

Have your tried connecting a true grid input to the xw pro and if so do you still have these problems?

Maybe I wasn’t really clear, but I’ve operated XW Pro with nothing but Sunny Boy attached, and they didn’t get along. Dirty AC signal oscillating, as if they were fighting each other. Adding my EV charger at 2700W was enough to cause it to “lock” into that distorted signal seen in the image in the first post. By the way, it was transitioning between 59.9 and 60Hz.

I’ve not connected to “true” grid as my utility won’t permit me to do so without going through net metering process, including getting an electrical inspection from the state. Hope to do that one day, but for now, I just want XW to do what it’s advertised to do—AC coupling with Sunny Boy. Being able to charge from and backfeed the Sunny Islands would just be gravy.
 
So I'd put the XW on the trailer or sell it
At this point, I really wish I’d picked up a pair of SI6048s just like on the trailer (especially when they were so cheap!) and hung those in the garage. I’d read about difficulty with pairing them with lithium, though, and got excited about the low idle consumption and surge capability of the XW. And the PDP, which I figured would simplify things for inspection.

Wish I had a time machine!
 
My guess is the ul 1741-sa is not playing well in off grid mode. The XW Pro is frequency shifting off, but the sudden off is playing havoc with the sunnyisland. The loosing waveform may be the Sunnyisland disconnecting, and reconnecting the XW Pro.

I didn't get a clear picture of how you have it wired, but the only way I have envisioned a two inverter/battery system to work off-grid is:

Grid forming inverter load -> grid input of inverter 2
inverter 2 load -> main panel.

In your setup, the PV sunny boy should be on the Grid forming inverter.

Regarding chargeverter option: victron orion tr 48/48 to charge battery to battery and avoid ac conversion loss.
 
The one time I have had issues with my SI5048's is when I was using a Solax inverter where their support had given me the wrong settings for setting the frequency shift to match the SI's. Their setting meant the Solax was still making power above 52htz (my local grid is 50htz) so the SI's continued to increase the frequency above 52htz until the Solax shut down completely. Once I had the right settings the SI's never go above 51.8htz as the inverters output then typically balances the house requirements.

Not aware of any Sunny Boy that is capable of creating a grid, even when set as off grid that is only setting the Grid-tie settings to match an SMA SI frequency shift. My SB's are set to off-grid 50 and only provide power when they see a grid from an SI.
 
Last edited:
So our system at the shop (that I had mentioned in that other thread) has the older XW+ 6848s, so we don't have access to all the grid code settings like the new XW Pros do.

The main thing that we changed was to set the Sunny Boy's Country standard to "Island mode 60 Hz". (Under: Grid monitoring> Grid monitoring> Country standard) Operating mode is also set to "Active power limitation". (Under: System and device control> Power control module> Operating mode) There are options to adjust the Start and Stop for the frequency shift curtailment as well, but I don't think that is related to what you are seeing, as the frequency shift is not where your issues arise. Your issue seems to be related to the interaction between the Sunny Boy and the XWs under "shared load" type of scenarios.

Our sinewave does sometimes go a little bit goofy, but that happens under specific load conditions, such as when our AC is running, and that is because of using a VFD/Freq. Drive and 3 phase compressor. I feel it is because of the AC being rectified to DC in the VFD that the wave form gets skewed. But we just see a "flat-top" in the sinewave. Not the jagged effects that you are getting.

A few questions that come to my mind:
  1. If you try running with only the XWs and the SB, and with no loads, do you see this happen?
  2. Can you connect the issue to any specific loads? (I.e. is there a specific load that is causing the issue to start?)
  3. Does it happen at a point where there is nearly a "break-even" between SB feeding the full load vs needing a "little bit" of battery/XW power to feed the loads? (At home I see a little bit of flickering under conditions where my Fronius Grid-tie inverter is nearly playing even on the loads. It is AC coupled to my Victrons.)
I have attached 2 screenshots showing some of our settings on our SB at the shop. I don't have all the settings open, but tried to have the ones open that may be relevant for you.
 

Attachments

  • Screenshot 2024-06-11 084258.png
    Screenshot 2024-06-11 084258.png
    45.7 KB · Views: 3
  • Screenshot 2024-06-11 084411.png
    Screenshot 2024-06-11 084411.png
    27.8 KB · Views: 3
At this point, I really wish I’d picked up a pair of SI6048s just like on the trailer (especially when they were so cheap!) and hung those in the garage. I’d read about difficulty with pairing them with lithium, though, and got excited about the low idle consumption and surge capability of the XW. And the PDP, which I figured would simplify things for inspection.

Wish I had a time machine!


$3800 delivered for two. Last one.
 
A few questions that come to my mind:
  1. If you try running with only the XWs and the SB, and with no loads, do you see this happen?
  2. Can you connect the issue to any specific loads? (I.e. is there a specific load that is causing the issue to start?)
  3. Does it happen at a point where there is nearly a "break-even" between SB feeding the full load vs needing a "little bit" of battery/XW power to feed the loads? (At home I see a little bit of flickering under conditions where my Fronius Grid-tie inverter is nearly playing even on the loads. It is AC coupled to my Victrons.)
First, thanks very much for taking the time to respond and share the screenshots of settings.

1. Yes, I have seen that behavior occur with literally nothing but XW and SB, but very sporadically.
2. The EV charger seems to trigger it fairly reliably. Other large loads sometimes do, as they switch on and off. (oven, hot water heater, space heaters)
3. I've definitely noticed that it's flakiest when it's close to break-even, but it really seems to only happen when the XW charger is active.

In your setup, the PV sunny boy should be on the Grid forming inverter.
I have had Sunny Island feeding grid input of XW, and Sunny Boy on the XW load. In that config, XW connects and passes through Sunny Island, and Sunny Island handles frequency shifting. The idea was to be able to switch off Sunny Island and let XW take over. I think I'll give up on that idea.

Regarding chargeverter option: victron orion tr 48/48 to charge battery to battery and avoid ac conversion loss.
Nice idea, but the problem is the batteries aren't proximal--probably around 75' apart. To get meaningful amps, I'd imagine I'd need some fat conductors. I'm okay with a little AC conversion loss. My Eltek FP2 "diy chargeverter" is 96+% efficient, .99 power factor, and was incredibly cheap. I might use that idea to send power to/from my Ryobi zero turn mower, but that thing is awful expensive. For the same price, I could buy and mount an inverter to my mower, making it a portable power station, and then use it to power another chargeverter. Loss of efficiency, but major utility gain.

$3800 delivered for two. Last one.
Don't tempt me! I actually found some even cheaper, and seriously considering grabbing them. I'd still need a time machine to avoid the painful loss of value on the XW/PDP, and all the time I've wasted trying to get the bloody thing to work right.
 
Last edited:
While writing the thread starting post, it occurred to me that although I'd performed factory reset on the XW multiple times, I'd never done so on the Sunny Boy. After an embarrassing amount of time and research, (I think) I eventually found it under "Device Parameters", "Device", "Operation", "Load parameter", "Load preset". I'm not even 100% sure that was actually it, but I did get offered the installation assistant the next time I logged in, so I guess that was it?

Anyway, it didn't seem to help, but was worth a try.

I also re-re-checked the AC wires between the Sunny Boy and XW.

Running out of ideas.
 
First, thanks very much for taking the time to respond and share the screenshots of settings.

1. Yes, I have seen that behavior occur with literally nothing but XW and SB, but very sporadically.
2. The EV charger seems to trigger it fairly reliably. Other large loads sometimes do, as they switch on and off. (oven, hot water heater, space heaters)
3. I've definitely noticed that it's flakiest when it's close to break-even, but it really seems to only happen when the XW charger is active.


I have had Sunny Island feeding grid input of XW, and Sunny Boy on the XW load. In that config, XW connects and passes through Sunny Island, and Sunny Island handles frequency shifting. The idea was to be able to switch off Sunny Island and let XW take over. I think I'll give up on that idea.
When you have(/had) the Sunny Islands feeding into the XWs and SB downstream, basically pushing back through the XW, did you see the same type of behavior? And if so, what was the state of the XWs at the time. E.g. passthrough, charging, "grid supporting"...?

An additional test you could try, just for more data, would be to disable AC charge in the XWs, then see what happens with the SIs passing through the XWs to the SB. This would allow SB to push back into SIs through the XWs without the XW being in a "charging" state.
 
When you have(/had) the Sunny Islands feeding into the XWs and SB downstream, basically pushing back through the XW, did you see the same type of behavior? And if so, what was the state of the XWs at the time. E.g. passthrough, charging, "grid supporting"...?
Yes, but I get the same behavior even at night, or with Sunny boy disconnected. So it really seems like overly aggressive “grid support” (via grid code, not “grid support” in the sense of backfeeding). I think it’s worst when I try to get XW to backfeed. It would be good for me to try what you’ve described, but I should get serious about carefully annotating all the different variables while experimenting.

An additional test you could try, just for more data, would be to disable AC charge in the XWs, then see what happens with the SIs passing through the XWs to the SB. This would allow SB to push back into SIs through the XWs without the XW being in a "charging" state.
That’s actually the way I’ve got it “working” right now. Pass through, letting Sunny island control everything. Neither charging or discharging with XW. Charging with diy chargeverter, and at night, disqualifying ac input so it’s in backup mode with no ac coupling. This is fairly stable, but doesn’t really give me any of the features I wanted, especially independence from the trailer ie ability to control Sunny Boy.

If it’s overly aggressive grid code (trying to stabilize grid, confusing since this seems like grid “support” to me, and happens even if “grid support” is disabled), then it might explain both of my issues. I’ve tried tweaking and disabling individual grid code parameters on XW, though, and it only seemed to make matters worse.

Still don’t think I’m any closer to narrowing down if I have a problem with hardware, firmware, settings, or wiring.
 
Last edited:
Yes, but I get the same behavior even at night, or with Sunny boy disconnected. So it really seems like overly aggressive “grid support” (via grid code, not “grid support” in the sense of backfeeding). I think it’s worst when I try to get XW to backfeed. It would be good for me to try what you’ve described, but I should get serious about carefully annotating all the different variables while experimenting.
That does sound like it has something to do with how the XWs are handling their synchronization to "grid". I wonder if it has anything to do with grid being a lot more "stiff" vs. the SIs. I don't know exactly how the XWs work to "watch" the grid frequency and voltage.
That’s actually the way I’ve got it “working” right now. Pass through, letting Sunny island control everything. Neither charging or discharging with XW. Charging with diy chargeverter, and at night, disqualifying ac input so it’s in backup mode with no ac coupling. This is fairly stable, but doesn’t really give me any of the features I wanted, especially independence from the trailer ie ability to control Sunny Boy.
Yea that's a bummer!
If it’s overly aggressive grid code (trying to stabilize grid, confusing since this seems like grid “support” to me, and happens even if “grid support” is disabled), then it might explain both of my issues. I’ve tried tweaking and disabling individual grid code parameters on XW, though, and it only seemed to make matters worse.

Still don’t think I’m any closer to narrowing down if I have a problem with hardware, firmware, settings, or wiring.
I can't think of anything else off the top of my head, but I'll keep this in the back of my mind and if I think of something else that you could try, I'll let you know!
 
Maybe I wasn’t really clear, but I’ve operated XW Pro with nothing but Sunny Boy attached, and they didn’t get along. Dirty AC signal oscillating, as if they were fighting each other. Adding my EV charger at 2700W was enough to cause it to “lock” into that distorted signal seen in the image in the first post. By the way, it was transitioning between 59.9 and 60Hz.

I’ve not connected to “true” grid as my utility won’t permit me to do so without going through net metering process, including getting an electrical inspection from the state. Hope to do that one day, but for now, I just want XW to do what it’s advertised to do—AC coupling with Sunny Boy. Being able to charge from and backfeed the Sunny Islands would just be gravy.
So, removing the sunny islands completely out of the equation and simplifying the issue, the xw pro does not want to ac couple correctly with the sunny boys?
 

diy solar

diy solar
Back
Top