Adding storage to my Enphase system

[GXMnow]

My last contact with Schneider tech support was actually helpful. When I got the PLC making adjustments every 5 seconds, I was concerned if I was beating up EEPROM or Flash memory. I asked if this was an issue. They had to send it up to a higher level tech, and a day later I got a reply. They explained that any settings changes are done in ram, and it periodically copies the ram settings into the EEPROM. That copy to the non volatile memory is always happening, so it makes no difference if you change the settings constantly or not.

The other questions I asked them about AC coupling, they were no help at all. Basically told me to change my solar to DC coupled with a Schneider MPPT controller. They never admit that their software has holes in it.

 

[GXMnow]

I got home from work at 5:15 and all seemed well, but at about 5:30 the lights flickered. I didn't think much of it. I was walking to the bathroom at the time. All my normal lights turned on, but the LED light had a noticeable pulsing to it. Hmmm. I wen back to my main home office PC and sure enough, the light in the corner of the kitchen is off. Yup, I am off grid. XW-Pro log shows it as AI Over frequency. But I went out and check, there is nothing at the main panel, even the power meter LCD display is blank.

Too bad it's kinda late. My Enphase solar is only making 350 watts total. It was probably a bit higher when it first went off grid, but my PLC still had the charge current dialed down to just 5%. So it was frequency shifting which was shutting a few inverters off again. I manually set the charge rate back to 50% and not it is a solid 60 Hz. Looking at the event log, I see at least 4 inverters threw the grid frequency high error, but they all cleared now and are trying to produce. It just ramped up a bit and is now making over 600 watts. Then Enphase saw another frequency bump and 5 shut down again. Very odd. I have charge block se in the XW, I jus moved that time to 6:30 pm, so it should be allowing charge.

I put my Fluke 76 meter on the XW output. One leg is at 121 volts and the frequency is bobbing from 59.9 Hz to 60.1Hz, constantly moving a bit. It should be steadier. Maybe I will need to do a firmware update. A few of my iQ7's are not happy with the power. As the sun is dropping, I am now down to 5 still making power, only 160 watts now, my house is using 800 watts. Let's see how long the battery will go. The JK BMS is still showing 92% remaining. In reality, I can only use 60% unless I change settings. So I really have about 50% left. Let's call it 300 amp hours to go. I am using 12 amps. 300 / 12 = 25 hours. So it should make it until sun up tomorrow. I bet my internet access won't stay working though. It dies after 4 hours last time.

Looking at the Enphase app, I should still be getting double this power, but it's not happy.

 

[400bird]

I wen back to my main home office PC and sure enough, the light in the corner of the kitchen is off. Yup, I am off grid. XW-Pro log shows it as AI Over frequency. But I went out and check, there is nothing at the main panel, even the power meter LCD display is blank.

Hmm, that's concerning that the XW display is blank.

Too bad it's kinda late. My Enphase solar is only making 350 watts total. It was probably a bit higher when it first went off grid, but my PLC still had the charge current dialed down to just 5%. So it was frequency shifting which was shutting a few inverters off again. I manually set the charge rate back to 50% and not it is a solid 60 Hz. Looking at the event log, I see at least 4 inverters threw the grid frequency high error, but they all cleared now and are trying to produce. It just ramped up a bit and is now making over 600 watts. Then Enphase saw another frequency bump and 5 shut down again. Very odd. I have charge block se in the XW, I jus moved that time to 6:30 pm, so it should be allowing charge.

I had assumed the charge block time schedule was for on-grid only. If the charge block schedule is still active off grid, I should probably figure out an alternate solution in my PI energy controller.

I put my Fluke 76 meter on the XW output. One leg is at 121 volts and the frequency is bobbing from 59.9 Hz to 60.1Hz, constantly moving a bit. It should be steadier. Maybe I will need to do a firmware update

If you get a chance to check after the sun sets, I wonder if the enphase micros are affecting the frequency?

I bet my internet access won't stay working though. It dies after 4 hours last time.

This 4 hour limit sounds like it's the ISP not using a back up generator, just 4 hours of battery. Frustrating

 

[GXMnow]

With only an hour of daylight left, I figured it couldn't hurt. So I tried pushing the iQ7's to the IEEE default 2015 grid profile. They accepted it and it say profile deployed, or something like that, but I have no production at all now, and no entries inthe event log since the profile update. There is still enough sun to be making 500 watts, so I expected them to at least try to come online, and maybe throw errors if the grid is unstable, but NOTHING.

The sun is now getting too low, so I am not going to try and change the profile again, the electronics in the iQ7's are powered from the solar panel, not the grid side.

400bird,
The So Cal Edison power meter outside is dark. My XW-Pro is lit up and running. It shows it is making 0.8 KW right now.

I don't know if it is the bobbing frequency that caused them to keep shutting off, or if it was really frequency shifting. In any case, my solar is done for the day, and I am on battery until the grid comes back. I need to add to my code to make the settings changes when the grid goes down. I never got around to writing that part.

SCE just texted, they expect to have the power back up around 8 pm tonight. So just one more hour.

 

[GXMnow]

Okay, I just went and checked, since the Enphase output is at zero watts. The output of the XW is still solid at 120.1 volts, but the frequency dropped to just 57.6 Hz, but it's no longer bobbing like it was with the Enphase pushing power. And, the inverter status page in InsightLocal is also showing 57.6 Hz, dead on match to my Fluke meter. Several of the inverters did show grid frequency low, and it is low.

 

[GXMnow]

I do remember others reporting an issue with the frequency sticking at the wrong frequency after it was tracking a generator, I never heard of it happening while off grid. like this. I would like to reset it, but then my network and this PC will crash for a while.

 

[GXMnow]

After 9:15 and the power is still out. Internet is still working, but it might not last much longer.
My battery bank is holding up nicely, still 54.8 volts. I reduced loads in my house, but I now have a cord over to my neighbor running his refrigerator. So I am now pulling just over 1,000 watts, about 19 amps at the battery. I only dropped 6% state of charge on the JK BMS in 2.5 hours. At that rate, I should still have at least 15 hours. That still get's me well past sunrise. BUT.... The Enphase will not make any power until I get the frequency back up to 60 Hz.

 

[GXMnow]

The power finally came back on at 4:34 am. The noise from our neighbors A/C coming on woke me up, and I can also see it in the battery summary in the XW. The battery only ran down to 52.2 volts, so it was still above the "Recharge Volts", but it went directly into charging anyways. And because there was no solar happening at 4 am, the PLC was not even looking to make adjustments, it came on at 45 amps of charge current. Just a few minutes later, I manually set it down to just 10 amps to start charging slow until the sun comes up. The battery came up a little more than 1 volt to 53.3 by 8:30 as the solar is now making more than the house is using again.

I made a few notes about thing I need to change in my PLC code, and I now know I need to update my XW. At what version did they fix the odd frequency issue? Getting stuck at 57.6 Hz makes no sense at all. When there is no power at AC1 or AC2 it should run on an internal clock at the correct grid frequency of 60 Hz (or 50 Hz in other regions). It should only vary from that to either track an AC input that is off frequency a little, or when it frequency shifts to cause a grid tie inverter to reduce output. In the 9 hours that the frequency was running low, two of my clocks fell 32 minutes behind. They are both digital displays. My cheapo alarm clock in my bedroom, and the clock on the kitchen range. Our other 2 plug in digital clocks held decent time as always, and so did the microwave oven. While the microwave was on the backup power, I did not try cooking in it with the 57.6 Hz power.

Now that the solar panels have enough light to power them up, I also changed them back to Cal Rule 21 grid profile, but when I went to the pull down list, there were a bunch of new profiles. Check out this screen grab.

The one I have highlighted sounds nice. "AC Coupling FW83pct/Hz PEL300 REC5s"
We all know what AC Coupling means, but the rest sounds interesting.

I may be wrong, but I think I decoded that correctly.

The FW is Frequency Watt. 83pct/Hz, I think means it will reduce it's output by 83 percent for an increase in frequency of 1 Hz. Nice to actually see a number on it.

Then PEL300 I think means "Power Export Limit". For that to work, I bet it requires the consumption side current transformers at the main. Then it will limit to just 300 watts of export from the solar.

And finally "REC5s" I think means "Reconnect after just 5 seconds" when it sees good grid again.

That all sounds great for an off grid setup, but here in So Cal, I might have some problems with the fast reconnect while I am on grid. While the XW is Rule 21 compliant, and it won't export for the full 5 minutes after a good grid, it does connect the loads to the grid much sooner when it qualifies that the power is ok. And the Enphase with this grid profile might then start exporting in just 5 seconds. If the XW battery is low, it should be commanded to use that power to charge the batteries, so it still should not be "exporting", but this still depends on our external controllers to make this happen. My current code has at least a 5 second delay, and could easily be over 7 seconds if it misses the moment the current comes on. So once it sees power export, it will send the command to start charging. Allowing the Enphase to export only up to 300 watts is good, and should work with my setup, but any lower would be trouble. My current code has to see export exceed 150 watts to increase the charge current. Each time that happens, the Enphase production meter would see it is exporting less than 300 and ramp up the microinverter output. Then my PLC will see it over 150 watts again, raise the charge current. Then Enphase will see less than 300 export and ramp up again. It should go to where the inverters are not curtailing any power. My charge current steps are about 80 watts, so I need a window at least that big. I currently have the PLC seek a 100 watt window, from 50 to 150 watts. This looks like it will work nicely, with my current seek window. It might only need a minor tweak to keep it stable. I might move it to seek 0 to 100 watts, to give a little more room for the Enphase to see while I am using the extra power for charging. Then, once the battery is full, it would then export limit at 300 watts beyond what the house is using. Under my current NEM 2.0 agreement, I don't want that 300 watt limit though. If I still have good sun, and my battery is full, I want the credit for the power, but if NEM 3.0 penalizes us for export, this could be a good option.

 

[svetz]

...The output of the XW is still solid at 120.1 volts, but the frequency dropped to just 57.6 Hz,...

Wow! Being in the anti-islanding frequency range from your ac coupling blows. If you couldn't recharge your battery the next day from solar that could have been bad.

Let us know what Schneider says about that "stable" 57.6 Hz. If I'm reading your posts right, you suspect the grid power got wonky enough before failure that it destabilized the Schneider? I think I would have reset it and been worried about inverter component failure.

The one I have highlighted sounds nice. "AC Coupling FW83pct/Hz PEL300 REC5s"

Hmmm... that's pretty interesting. I like your explanations of the meanings!

 

[GXMnow]

Wow! Being in the anti-islanding frequency range from your ac coupling blows. If you couldn't recharge your battery the next day from solar that could have been bad.

Let us know what Schneider says about that "stable" 57.6 Hz. If I'm reading your posts right, you suspect the grid power got wonky enough before failure that it destabilized the Schneider? I think I would have reset it and been worried about inverter component failure.


Hmmm... that's pretty interesting. I like your explanations of the meanings!

I have not contacted Schneider yet.
I will compose an e-mail to them today.

Here is the funny thing, when the grid failed, the error was grid frequency high, and it was at 60 Hz average, but it was bobbing between 59.9 and 60.1 Hz. It is a small enough range, it should not be a big deal, but it was causing a noticeable flickering in some of my LED lights, and I could hear a slight wobble in the hum from the fan in my bathroom. While I still had some sun, and the Enphase inverters kept going on and off because of the wobble, it never wen more than +/- 0.2 Hz from 60.0 on my Fluke, and the display in the InsightLocal kept showing 60.0 or 60.1 with a slower update rate from my Fluke.

When I was down to where the Solar should be making about 500 watts, I was only seeing 200, I figured I could play around and not lose too much power/energy. When I told the Enphase Envoy Local Installer to update the Grid Profile, I was watching it propagate, updating 1 to 4 inverters about every 2 minutes. I was not watching the XW at this point. I expected to see them start producing, but as each one woke up with the new grid profile, it just sat at zero watts production, with no log entries coming in at all. After over 5 minutes, and still no production, I could still see sun hitting my panels, that's when I looked at the XW and saw the 57.6 Hz showing. As far as I know, the grid power was down the whole time. There is no new log entry of it trying to connect. It never even logs when it does connect. Enphase logs EVERYTHING! It would be nice to have the XW log when a bulk charge starts, when it hits absorb, when it qualifies the grid. It logs none of those.

So my guess is that when the Enphase updated the grid profile, it somehow caused the XW to see a weird fluctuation that caused it to change it's running frequency. But once things were stable, why was it still locked at the wrong frequency? IT should know when there is no grid input, and everything is in range, it should ramp to the correct frequency. Once I contact Schneider and get a reply, I will let you know what they say.

That being said, what versions are people running in their XW-Pro and what is working well without odd bugs? I know I am running an old version.

 

[pvdude]

I'm running 1.11.01bn49.
However, I have read somewhere here that "49" breaks sell back!
As I don't sell back, no issue, every feature I use works fine, but my system is very simple, compared to yours.
(DC coupled, no sell back, FLA batteries, grid support)

 

[svetz]

... it was bobbing between 59.9 and 60.1 Hz. It is a small enough range, ... +/- 0.2 Hz from 60.0 on my Fluke

Digital meters try to "average" over a sample period, so it could have been bouncing far beyond that (unless your fluke is one of those slick ones that shows you upper/lower bounds). if it was beyond the bounds for the ride-through time I can see the microinverters going into anti-islanding.

...So my guess is that when the Enphase updated the grid profile, it somehow caused the XW to see a weird fluctuation that caused it to change it's running frequency. But once things were stable, why was it still locked at the wrong frequency?

As the microinverters were in anti-islanding before and after the change, don't see them causing the fluctuation. The grid might have been nutso though.

 

[400bird]

I'm running 1.11.01bn49.

That version broke EPC, but GXMnow isn't using EPC, so that should be fine. I had no other issues with it.

I'm running 1.11.00 bn28
(I'll need to double check that specific number, but it was the release directly prior to 1.11.01)

 

[GXMnow]

My Fluke is a 76 True RMS meter. Only 4 digit, but the first digit is a full digit, not the normal 3.5 digit. The digits can change pretty fast, so it was giving a new reading about 2 or 3 times per second. It's possible the frequency moved a little further than 0.1 Hz, but most of the time it was at 60 Hz +/- 0.04, the highest and lowest it ever showed was just the 0.10 Hz deviation. A few times I saw 59.92 and 60.08 so less than a 0.1 Hz difference. I would venture to guess the biggest swing was less than 0.5 Hz during that. I just can't imagine what made it drop to 57.6 Hz.

I got a reply from Schneider, basically telling me to update to 1.11 Build 49. I looked through the release notes and found "Fixed bug where XW Pro frequency shifts when changing the battery’s max charge rate" Hmmmm

 

[Shimmy]

Here is the funny thing, when the grid failed, the error was grid frequency high, and it was at 60 Hz average, but it was bobbing between 59.9 and 60.1 Hz. It is a small enough range, it should not be a big deal, but it was causing a noticeable flickering in some of my LED lights, and I could hear a slight wobble in the hum from the fan in my bathroom. While I still had some sun, and the Enphase inverters kept going on and off because of the wobble, it never wen more than +/- 0.2 Hz from 60.0 on my Fluke, and the display in the InsightLocal kept showing 60.0 or 60.1 with a slower update rate from my Fluke.

That sounds like high THD to me.

 

[GXMnow]

There is no doubt that the power coming from the iQ7 inverters is not a steady pure sine current. When I am on battery only from the XW-Pro, the power is very clean. None of my LED lights show any flicker at all. But when the iQ7's are pushing power into the system, a few of the LED lights show a very distinct flicker/pulsing. From what I have been able to determine, it has to do with the MPPT search algorithm in the iQ7's.

Each AC cycle (might be a couple cycles), it varies the amount of current a little up or down to test which way makes more power from the solar panel. So each cycle is a clean low distortion sine wave, but one cycle might be 0.75 amps rms, but then the next cycle could be 0.77 amps, and the next is 0.73 amps. I don't have a full block diagram or code on how the Enphase inverters work, but I do have a decent idea of what it is doing.

The solar panel input is charging a capacitor bank. If there is no load on the output, the capacitor would charge up to the open circuit voltage of the solar panel. The inverter pulls current from the capacitor bank to put out the AC waveform. Each cycle it measures the voltage and current from the solar panel. It then calculates the current for the next cycle of AC and determines if it got more energy from the solar panel or not. If it got more energy, it will step further in the same direction, if it got less, it will step in the opposite direction. When it finds the true maximum power point, it still needs to try a little more current, it will see the voltage drop too much, making the power drop. Then it will step back to MPP (Maximum Power Point) by reducing the current a step, it will then see the power come back up. So then it drops the current another step on the next cycle. The panel voltage comes up a little more, but not enough, so the power dips again. So next cycle is pulls a little more current again, and hits the MPP. If the amount of sunlight hitting the panel does not change, it will keep oscillating one step above and one step below the MPP.

With 16 panels, I would expect the oscillations and variations in MPP across all the panels to sort of average out, but it does not. In a perfect world, when some of the panels increase current, you would expect a few to decrease current. While on grid, this is never a problem, and I never see any light flicker. But when running on the XW-Pro as the grid forming device, the pulsing is obvious. The actual voltage variation is virtually nothing. I can't see it with my scope, let alone a meter. The power looks very clean, but there is a tiny bit of jitter. This last time, I did not put the scope on it. So I was not able to verify what was going on. By the time I thought about it, I had already caused all the Enphase inverters to go offline with the 57.6 Hz from the XW-Pro.

I hope to do the firmware update this weekend. I have to shut down a bunch of things before I cut power to the backup loads panel. And I will setup my laptop to talk to the XW.

 

[svetz]

There is no doubt that the power coming from the iQ7 inverters is not a steady pure sine current.

The IQ7's don't generate their own waveforms, only the IQ8s can do that. Whatever you're seeing is from something else. If out of spec, they should go into anti-islanding.

.... From what I have been able to determine, it has to do with the MPPT search algorithm in the iQ7's....

The MPPT is working on the solar DC side. It can cause the amps out of the microinverter to vary, but shouldn't affect the phase or LEDs. I don't doubt your LEDs flickering, but don't see why.

While on grid, this is never a problem,

The IQ7s are inverting while on-grid, if it was solely an IQ7 issue it should happen here too. So, an interaction between the XW and the IQ7s?

...when running on the XW-Pro as the grid forming device, the pulsing is obvious. The actual voltage variation is virtually nothing. I can't see it with my scope, let alone a meter.

For flickering, my guess would be the XW didn't have a frequency lock and the IQ7s were having difficulty phase-matching (e.g., every wave is just a little different). But if it was bad enough to make LEDs flicker they should go into anti-islanding. Well, possibly not if it's in the throttling range 60 to 63 Hz.

Might be worth calling Enphase to see what they say.

 

[GXMnow]

The IQ7's don't generate their own waveforms, only the IQ8s can do that. Whatever you're seeing is from something else. If out of spec, they should go into anti-islanding.

At the start of the power outage, the XW-Pro disconnected from the grid and it was producing a solid in spec local grid.

The iQ7's don't produce their own grid, but they do have to internally generate a waveform to match the grid they are on. The current they produce has to also be in a sine wave, in sync with the "grid". Zero current at the start of the waveform, the ramp up to 50% current at 30 degrees in, then 70% current at 45 degrees, maximum current at 90 degrees... You get the idea. This is part of the 5 minute qualification time. It needs to sync up with the grid so the current being added is at the right phase.

And this is where the jitter becomes a problem. If the voltage waveform is moving around in time, it is no longer perfectly in sync. How much jitter is too much? Was the XW-Pro actually jittering, or was it the current from the iQ7s causing a small voltage change which was causing them to think the frequency was shifting? It's hard to tell where the errors are coming from. But it does look like Schneider had to update their frequency control. I will update after I get the new firmware installed.

 

[Shimmy]

The IQ7's don't generate their own waveforms, only the IQ8s can do that. Whatever you're seeing is from something else. If out of spec, they should go into anti-islanding.

You still end up with voltage distortion from a parallel current source just due to system impedance. Since utility impedance is so low, the effect is less of an issue. The XW-Pro is unlikely to adjust its voltage waveform to improve power quality.

My guess is that at low power output the effect from the microinverters makes things worse. You have a shorter window where you produce current (just at the peaks); MPPT might play a role in it. I also doubt that the actual frequency is changing, but more that the waveform makes it difficult to count cycles.

 

[svetz]

...How much jitter is too much? Was the XW-Pro actually jittering, or was it the current from the iQ7s causing a small voltage change which was causing them to think the frequency was shifting?...

I hear what you're saying, that there's an imbalance, it only happens when the microinverters are on, and it takes two to tango.

What I'm saying is the microinverters have no problems syncing while on grid, it's only when the XW is the source of the waveform. Even worse, the XW is designed for AC Coupling, so fluctuations and voltage changes from the microinverters have to be a part of it's engineering.

Ergo, ??
; -)