Solaredge ground fault isolation fault

[qrper]

Hello everyone. This is my first post.

My name is mike, and I'm in Ohio. I've been using and designing photovoltaic systems since 1978. This isn't my first rodeo, however that being said, up until this summer all my systems have been off the gird battery based and generally used for communications.

This summer I completely re-did my 48 V dc battery system complete with new batteries and new panels. The 20+ year old Solarex (remember those?) panels were removed and 20 used (Came from a decommissioned solar farm out west) 250 W Trina panels were installed. A new Midnite Solar Classic 150 went in with all new wiring and controls. Life is good! On a bright cool days, I see 80+ A going into my 16 US batteries (L16 clones) A Trace 5548 sine wave inverter handles 80% of the home's loads.

That system works like gang busters.

Until I purchased a electric car.

The old trace can't do 240 V, so I decided to install a grid tie system that would supply the missing 20% the inverter can't do like the 240 V deep well pump and the AC. The gird tie would also allow me to charge the EV when the sun is shinning.

Long story short as they say...

Found out from the local electrical inspector that the panels I had purchased, the used Trinas, won't pass inspection because the "UL" labels had been removed. So I purchased 12 new panels and installed them on the ground mount. Because of several issues with my panel box, and the meter base, I had to have an licensed electrical do the actual hookup to the gird.

Inspector came out, passed the system on the first go.

I was making 3kW back into the grid with 12 Hyundai 250 W panels.

Later, I added another 12 panels. These were the used Trina panels. So I bumped my output from 3kW to 6kW.

Between the new grid tie and the battery based inverter, the following electric bill was $5.01 which was the cost of the meter and billing. I actually had an excess of kWs generated.

Fast forward....

The Solaredge inverter would shut down when it would rain or have a few cloudy/wet days. Error was a ground fault. They also call it a isolation fault.

Disconnected the trina array, and the error went away.

I don't have a Megger, but do have some Fluke 87 meters.

With the + and - leads of a solar panel shorted together, I see 590Ω to the array framework.
Some modules show more leakage, some are perfectly fine. According to the Solaredge trouble shooting guide, I should see 80MΩ on the panel. My Fluke can't read that high of resistance, but 590Ω certainly isn't correct!

If I test from the + lead of the single pv module, to frame (ground) with the neg lead of the PV module just a hanging in the wind, I read from 5 V to as high as 27 V.
Obviously, there is leakage from the + lead to the frame, and the inverter senses this leakage and throws the error.

Has anyone ever seen this happen with the panels? I could see water in a junction box, moisture in the MC4 connectors, a wire break, and so on.

While I haven't checked, I'm thinking maybe the bypass diodes in the Jbox of the panel might be leaky. I popped the cover and the diodes look like they are welded in,

So, I'm up for any suggestions, because right now, It looks like I'm going to have to scrap the Trina panels and install new ones. (BTW the new panels don't give the same results as the used Trinas)

Sorry for the long post, but wanted to cover all the bases

I'm Mike

[bcroe]

Mike, Sounds like you are on the right trail. Maybe those used panels were removed
because of ground leakage issues. I might disconnect the suspect panel string and
with some sun, use an old time mechanical meter to measure voltage from either end
of the string, to ground. The meter must draw small current (the Fluke does not) to
drain off any electro static charge. Or connect a 220K ohm resistor in parallel with
the Fluke.

Leaking bypass diodes will not cause a ground fault.

With a ground fault, I would expect a certain voltage at either end (opposite polarities),
which add up to the Voc of the panels. Regarding the string as a voltage divider, you
may be able to estimate the point of the fault. Multiple faults will be more difficult.
With the meter still reading one end of the string, you could disconnect the most far
away panel. Keep doing the next panel till the meter drops to zero, you have just
unplugged a ground fault panel. Removing it and reconnecting the rest, you may
find more faults. good luck, Bruce Roe

[bcroe]

I would not use an ohmmeter to try and locate a panel ground fault, they
do not handle the voltages and junctions in a panel. On my 400 V strings
I have used this tool instead of a meter, a good part of an amp must flow.
Bruce Roe

PVtestLt.JPG

[qrper]

Mike, Sounds like you are on the right trail. Maybe those used panels were removed
because of ground leakage issues. I might disconnect the suspect panel string and
with some sun, use an old time mechanical meter to measure voltage from either end
of the string, to ground. The meter must draw small current (the Fluke does not) to
drain off any electro static charge. Or connect a 220K ohm resistor in parallel with
the Fluke.

Leaking bypass diodes will not cause a ground fault.

With a ground fault, I would expect a certain voltage at either end (opposite polarities),
which add up to the Voc of the panels. Regarding the string as a voltage divider, you
may be able to estimate the point of the fault. Multiple faults will be more difficult.
With the meter still reading one end of the string, you could disconnect the most far
away panel. Keep doing the next panel till the meter drops to zero, you have just
unplugged a ground fault panel. Removing it and reconnecting the rest, you may
find more faults. good luck, Bruce Roe

Seems great minds think alike. I was wondering the same thing, maybe that's why they were pulled. I tried contacting the seller, (eBay!) and haven't had any reply.

I was working on the array yesterday, it was about 60 here, and thought the very same thing. My Fluke as an input impedance of about 200MΩ which is clearly not enough load. Between my finger and the metal array frame, I can see some voltage displayed on the fluke caused by the dissimilar metals. I have an old Simpson 260 which I'll dig out. If I remember, I think it has an input impedance of about 20K or so.

Okay on the bypass diodes. I was reaching for an answer...

Each panel has one of the Solaredge optimizers hooked to it. Each optimizer makes 1 V when connected to a solar panel. As you may well be aware, when the inverter is happy, it turns on the optimizers and they squirt out 400ish V dc. Are you suggesting I disconnect the panels from the optimizers and then check panels?

One quickie... how do I post a photo? I clicked on the menu and it asks for url.

Mike

[qrper]

This was the test I was conducting using my ohm meter. Reading ohms to ground.

[Mike90250]

.....One quickie... how do I post a photo? I clicked on the menu and it asks for url.

HowToImage_SPT.png

You almost got it, use the IMAGE button, and the dialog box that pops up

UploadImage.png

Chose the UPLOAD tab, then BROWSE
and finally, SEND it to the SERVER, when the upload is complete, you can
then the Blue OK button.

[bcroe]

Actually my test was for bare panels, not sure if it will work with optimizers. Just
leaving the optimizers out and connecting the panels for the ground test might
work better, it will give the high voltages that cause more leakage. Bruce Roe

[qrper]

Actually my test was for bare panels, not sure if it will work with optimizers. Just
leaving the optimizers out and connecting the panels for the ground test might
work better, it will give the high voltages that cause more leakage. Bruce Roe

Okay...

let me get this straight in my head.

I'll do 4 panels in series, that's about to be around 150 V dc.

So... 4 panels in series. the negative lead of the last panel won't be connected to anything, just a dangling in the air.

From the + lead I'll measure with my low impedance analog meter to the frame (ground) or use a light bulb like you did

If I see voltage, then removed one panel at a time from the string until the voltage goes away. That panel is the one that's faulty.
Repeat on the other two strings 'cause I have 12 modules, four modules in a string, and thus three strings.

The new Hyundai panels are on the left, the used Trinas on the right. They are all back including the frames

IMG_1131-a.jpg

Mike

[bcroe]

I would check both ends (both disconnected from other equipment) for voltage to ground, If
no voltage found that stuff is not grounded. With voltage I would connect the meter to the
end that is highest, then start disconnecting panels from the opposite end. You need some
higher voltage for a good test, getting that from the panels.

The voltage should hint as to the ground location. If 150V total gives 94V at one end, that
end is likely 2.5 panels away from the ground. Bruce Roe

[qrper]

I believe I may have been over thinking this. It's not so much a ground fault as it is isolation fault. They're close, but not the same. The weather here today was crap, so I spent most of the day watching youtube videos. Since my problem only happens when the weather is wet, foggy, rainy, or otherwise nasty, I suspect I don't have a 'wire shorting out' namely because the fault goes away when the weather is better. Yeah, it could be water that getting into a wire or MC4 connector or even one of the optimizers.

I'll still do some more tests, but I leaning toward water that somehow got into the panel (s) and causing the issue.

More tests to do!

that will have to wait until the sun comes back out

mike

[Ampster]

I believe I may have been over thinking this. It's not so much a ground fault as it is isolation fault. They're close, but not the same. ......

Could it be on the AC side? I recently had an isolation fault on a SolarEdge inverter and it was the ground wire from the inverter to the main service panel that was the issue.

[qrper]

Could it be on the AC side? I recently had an isolation fault on a SolarEdge inverter and it was the ground wire from the inverter to the main service panel that was the issue.

Nope, I can disconnect one sub array and the fault goes away.

[foo1bar]

Nope, I can disconnect one sub array and the fault goes away.

Since it's Solaredge, you can probably narrow down whether it's the panels or just having the optimizers on that section hooked up.
You can disconnect some (or all) of the panel-optimizer connections and see if you still have the same fault.
You can probably add a single optimizer at a time from the "bad" subarray into the "good" subarray. (either increasing the string length by one, or substituting one optimizer for another.)

[qrper]

Since it's Solaredge, you can probably narrow down whether it's the panels or just having the optimizers on that section hooked up.
You can disconnect some (or all) of the panel-optimizer connections and see if you still have the same fault.
You can probably add a single optimizer at a time from the "bad" subarray into the "good" subarray. (either increasing the string length by one, or substituting one optimizer for another.)

The problem seems to be within those used modules. Even with my Fluke 77 meter, I can read low resistance between the panel's positive and negative leads shorted together and the array frame.

I purchased an Klein megger the other day. It's not a $1k unit, but not a chinese crap shoot either. Should be here friday, and if the weather holds out, will test the panels one at a time.

[Mike90250]

Be VERY careful with a megger. They often use high voltage to test for leakage, and most panels are only rated to 600V. You risk damage to panels if you test above the panel rating. Some +250V panels now come with a 1,000v rating, but know what your test gear is putting onto the panels.