XW Pro Primary/Secondary load imbalance with v2.04? [RESOLVED]

[JBertok]

Has anyone running a pair of XW Pro 6848 in Split-Phase Primary/Secondary noticed imbalance in load sharing between the two since installing Firmware v2.04? Under all circumstances in v1.xx firmware the loads would balance, on average, seamlessly between both inverters. Sometimes the bulk of the loads would be powered by one, but then it would flop over to the other so overall they both got basically the same duty. (I run zero-sell)
What I get now is random behavior. They almost have a personality. Some days the loads will balance exactly as before, other days the Secondary XW Pro just sits there idle, appearing normal and ready to go, but won't help with the loads unless Inverter 1's load goes over 9kW. In that scenario Inverter 2 will start contributing, drawing DC to make AC it is supposed, (top image blow) but Inverter 1 is still pulling from grid and maxed out, and Inverter 2's contribution is only temporary. Inverter 2 will begin to incrementally offload responsibility back to inverter 1 until it's back to doing nothing again. (second image below) Yesterday was a good day - the loads were equally shared between the two and grid consumption avoided while plenty of DC was available. Two days ago - antics. I am starting to wonder if it's a odd vs. even day bug. (For clarity, all time setpoints under Grid Energy Management are set to 12:00AM on both XW Pros which means disabled.)
I have to decide if this is actually a bug in v2.04 and consider going back to v1.11 rather than holding out for v2.10. Anybody on the inside have word on v2.10?

Other observations/test:
- Inverter 2 still shows the front panel AC1 LED good if I throw the breaker. That's not right.
- Putting inverter 2 into Standby and then back to Operating does not help.
- Rebooting Inverter 2 does not help.
- Reducing the XANBUS to just the two XW Pros, the Battery Monitor, and the Conext Gateway does not help.
- Removing all loads momentarily does not help.






If I downgrade I will post back with an update.

 

[JBertok]

UPDATE:
I've found a work-around for this problem. Based on what's worked, I do think this is a bug with v2.04 relevant to multi-XW configurations configured for zero-sell.

Both inverters:
Advanced Configuration --> Grid Energy Management (Grid Support) --> Maximum Export (Sell) Sell Amps [0]

Inverter 1:
Basic Configuration --> Sell Enable/Disable [Disabled]

Inverter 2:
Basic Configuration --> Sell Enable/Disable [Enabled]

I feel this is a bug because you should not have to enable Sell for self-consumption zero-sell. I certainly didn't have to with v1.11. I'm happy with each inverter maintaining their usual 30-50 watts margin of persistent consumption from the grid rather than running too close to unintentionally exporting. As usual, you'll always see a brief moment of export when a large load drops, and it seems like the inverters more quickly re-synchronize their production to loads with sell disabled. Almost like meeting the "Maximum Export" setpoint is a little lazier and takes a few more cycles. I will be watching daily metrics to see if the cumulative export exceeds a comfortable value - probably less than 50W considering the monthly total is the number to worry about.

I'm going to drop Schneider a line to let them know what's going on for the unlikely case they're actually not yet aware of it needing to be fixed for the next release.

 

[JBertok]

UPDATE #2:

Nope. It cannot correctly control export when sell is enabled despite the "Maximum Sell" value being 0.
I am giving up on v2.04 and will be going back to Firmware v1.11 as soon as it's feasible to shut down the entire house. Disappointing.

 

[JBertok]

I went back to v1.11 on the XW Pros, and even back to 1.17 on Insight. (Hilariously, the verbiage of Master/Slave changes to Primary/Secondary in 1.18) The random behavior of Inverter 2 persists.

I now suspect something is wrong besides settings or firmware. Inverter 2 doesn't pass power through AC1.

 

[JBertok]

I'm posting an explanation of my methodology and what I've done to finally resolve this problem.

With Inverter 2 behaving erratically, I figured getting a clear picture of exactly what functionality is affected would be a necessary prerequisite to opening a trouble ticket if I was to go that route. The house was put on bypass so I could work on the inverters. I wanted to put both of them into standby and then switch both of them to "Split-Phase Stand-Alone" and run each through the same functionality tests to see what the differences were. The two inverters were A/C isolated from each other in both input and output with the PDP breakers. This is where things got truly bizarre. Both Inverters responded to the settings independently in exactly the same way, but it's HOW they responded that surprised me.

- In both inverters, enabling Grid Support immediately ramped up export to 6kW even with Sell disabled and Max Export set to 0
- In both inverters, enabling the Battery Charger (even with Grid Support and Backup Mode disabled) immediately ramped up export to the grid to 6kW actually DRAINING the batteries, not charging them.
- Meters in/out were each verified to be working on both inverters - these were not false readings.

It's like everything was backwards.

So, I flashed both inverters back to v2.04 (Since v1.11 didn't resolve the issue mentioned above) and then wiped all settings back to factory defaults for both inverters. Next, brought them up independently up from Standby still configured as Stand-Alone, plugged in all my settings specific to this configuration, and functionality-tested everything. Both inverters now worked perfectly and exactly as expected. After that, switch the Multi-Unit configuration over to Master and Slave and brought them up in the correct sequence - no surprises. Everything seemed "normal", but actually better than I am used to in one particular way - somehow both inverters were now regulating their production more precisely and I am seeing Grid input bouncing between 0 and 0.1kW instead of between 0.1 and 0.2 as it used to. Anyone familiar with Insight is familiar with how inputs or outputs go from green gray when no power is flowing to or from them. I am not used to seeing that, so this is cool.



"Sell" is still enabled on Inverter 2. It's not fluctuating wildly all over the place like things did before, and the loads stay more balanced now between the two. I don't understand how clearing the configurations back to factory default solved my problems. Maybe some things left over in the configuration from several firmware upgrades in their lifetimes. Since the issues were never present before upgrading to v2.04, perhaps there's something we can't see being retained in the 1.XX configurations running on 2.XX that we can't see. Interestingly, I had flashed both back to v1.11 a few nights ago with no improvement either. I'll probably never know what exactly happened.

 

[JBertok]

Final update: It's resolved, but the HOW is amazing...

I have been watching my daily operations with diligent scrutiny, while resisting the urge to intervene and change setpoints. After day's worth of observation, this problem of imbalance wasn't actually resolved, and crept back into operation. Seemed especially predominant during the day when grid voltage sags and PV production is high, although this would prove unrelated as I'll explain.

First, the problem with Enabling Sell in a self-consumption zero-sell scenario:
The "Controls\Sell Enable" workaround with 0A Export doesn't work right. The "Grid Energy Management (Grid Support)\Maximum Export (sell) Sell Amps" setting is seemingly low in priority programmatically. The condition that seem to lead to power flow creeping unchecked into the grid is when DC bus voltage rises, like in the morning when solar comes up, or following a large load shutting down. Despite the reverse flow to the grid being clearly visible in Insight, the XW's are not recognizing it and responding to it in the load-balancing routine. Under these circumstances, 2-3kW per day is going to make its way into the grid. In my opinion, "Sell Amps" need to be much higher in functional priority.

I like the way the XWs supports the internal loads in "Sell Enable" mode. Running this way the XWs will ramp up, in steps, inverting to offset grid input. It has a more gentle hysteresis meeting demand. When "Sell Enable" is disabled, the XWs will immediately try to supplement all loads fully with inverting DC. Normally this works fine, as it minimizes pull from the grid in times of DC abundance. The problem is when the batteries are depleted and current availability is low. It is fair so say this is only an issue with Lead Acid users since these experience voltage droop at greater discharge levels. For those of us depleting and recharging FLA daily to maximize PV harvest, toward the end of the battery discharge cycle when the voltage setpoint is reached, the inverters will switch from "Grid Support" to "Pass-through" with an associated switch of the input relays. The battery pack, now with no load, will begin to rise in voltage, triggering the inverters to switch back into "Grid Support". When they switch back though they start by trying to offset to full amount of AC Load with Inverting, and pull an obviously massive amount of amperage from the battery, which with FLA will cause a huge voltage drop when the batteries are this low, and that starts the whole scenario cycling. There are many way to prevent this - bigger FLA battery pack, or use LiFePO4, or use a higher Grid Support voltage to they don't discharge so far, or even use State of Charge mode. That's fine, but it the "Sell Enable" would abide by the 0 Amp Export limit I could do exactly what I want without caveat.

The original problem and the real fix:
With Inverter 1 observing 100% power passing through from AC1 to Load1, Inverter 2 has no observable load to offset with inverting. If Inverter 1 then starts inverting at 100% capacity (6kW), with this condition still present Inverter 2 continues to see nothing and any loads beyond 6kW simple pass through Inverter 1, from the grid, while Inverter 2 just sits there.
Real Cause:
It turns out that my original wiring (over a year old now) was 100% the culprit. I cabled each of the two inverter input breakers with individual 4AWG conductors back to my main meter panel. Each individually fed by a dual-pole 60A circuit breaker. I did this for reasons I thought made sense to me: each inverter's entire feed protected by a separate 60A breaker, each one individually disconnectable at the main panel if needed, and greater capacity over a single shared 100A circuit, although the main supply is only 100A, and obviously 4AWG THHN is so much easier than 2AWG to work with in conduit and a tiny 30 year-old meter panel. (Incidentally, I did use 2AWG to feed the PDP AC Load breakers busbar to the whole-house transfer switch.)
Real Fix:
The input breakers for both inverters apparently must be bridged on the input sides using the bridge bar that comes with the PDP, to be fed from a common source. The inverters depend on seeing no voltage difference between their inputs, and it is a feedback-loop situation - the more voltage difference the more the Primary (Master) will push until it carries the full burden of AC load. This is why Inverter 2 would slowly trickle away from helping the load. I found this out on a hunch by jumper the same phases on the two inverter input breakers. INSTANTLY Inverter 2 began supplementing load. I measured only 0.1V difference between the two feeds but there is obviously reactive load difference, whatever the case it is the confirmed cause of my problem.
1. Installed the Schneider PDP input breaker 2-terminal busbars.
2. Connected both common phase leads from the 60A breakers in the main panel to the busbar terminal.
3. Installed a 4p tie handle (QL4HT) onto the 60A breakers outside.

Before (don't do this!), and After:


I finally got what I wanted to see. Look at those AC1 grid inputs.... ?

 

[SniperX]

GREAT problem-solving. Thanks for documenting this.

 

[JBertok]

Up above I mentioned a "tie handle" for the two XW Pro circuits. With them now tied into a common busbar at the PDP, I can't have two un-linked double-pole breakers supplying the two pairs of 4AWG cable. Since the main panel bus bars (30+ years old) are rated to only 100 Amps, this pair of breakers needs to behave like a single 100A breaker, so I changed the 60A units I had out to 50 and derated AC1 in the XWs. The tie bar kit (Eaton QL4HT) has a steel rod that goes through the holes and then the handle snaps over everything. Here is what that looks like:

The two circuits are no longer discrete, so it needs a re-label too at this point. ?

 

[wheisenburg]

So, I flashed both inverters back to v2.04 (Since v1.11 didn't resolve the issue mentioned above) and then wiped all settings back to factory defaults for both inverters.

Apparently there are "Hidden Settings" that get cleared when you run this Reset to Factory Defaults.

I was having weird issues with my inverters. The reset to factory seemed to fix it.

 

[wheisenburg]

It turns out that my original wiring (over a year old now) was 100% the culprit. I cabled each of the two inverter input breakers with individual 4AWG conductors back to my main meter panel. Each individually fed by a dual-pole 60A circuit breaker.

When my solar was installed I got them to put a 200 amp Disconnect breaker just after the power meter. This was then run into my main panel.

So when adding my Schneider Battery backup system, I was able to turn off this main breaker to allow me to rework things as follows:

1. Wired the Disconnect breaker to a new 200 amp "Grid Panel" with 2/0 copper. The wire goes into the main breaker.
2. Wired the "Grid Panel" to the PDP Grid input with 2/0 copper. The wire starts in lugs on the bottom of the grid panel and runs to the fingers for grid in on the PDP.
3. Wired the Load Out lugs on the PDP panel to the Main breaker on the Main Panel which has not become the Battery Protected Panel. This was also 2/0 copper.

Yes, I bought a boat load of 2/0 copper. Yes, I ran it through giant 2" EMT. Yes it was a real pain in the neck. Over kill? Probably.

I guess it does show that following the instructions and wiring everything up EXACTLY as they have shown, could be important for reasons that we may not understand.