Enphase Failures

[solarix]

Hey, this is Solarix with another episode of the "Go Solar Show".... We just got down off the roof of a six year old Enphase/REC system. Three of the microinverters (out of 28) had failed and one panel has very low output. Enphase had sent replacements to the homeowner and we finally had the time this morning to get up on his two story roof to swap them out. The system was originally installed by an out of town outfit that is long gone and so they called us. Well... Enphase has redesigned their stuff since then and the trunk cabling is now different and you have to pull off surrounding panels to replace that. The mounting slots are smaller on the new brackets so the original mounting hardware doesn't fit. And boy do we like the new integrated grounding panel clamps that eliminate the old solid copper ground wire (leash) that used to be used (like on this old system). The best part is that after re-energizing the system and looking at the monitoring online, we find that the original installers messed up the array mapping and the many of the microinverters are not actually located where they were said to be! Two of the ones we replaced are still showing bad. The Enphase s/n label is just a little paper thing that can come off and is only on the top surface (where you can't see it unless removing the PV panel its under) so who knows where the real failed panels are? We're gonna have to charge this homeowner a ton to get under each panel and figure out what is where so that if (when) the next microinverter failures occur, he'll be ready. (The internet monitoring takes quite a while to update its data so its not very practical to deduce where the microinverters are by covering the panels one at a time to see which ones drop out)

Shall I "dis" on Emphase some more?

[sdold]

That sounds like a bad day. Were the failed inverters M190s?

[AzRoute66]

Perhaps explain to the homeowner how he might reduce the service call charge if he goes out there with some large throw pillows and finds those panels on his own time... I would do it.

[bcroe]

Perhaps explain to the homeowner how he might reduce the service call charge if he goes out there with some large throw pillows and finds those panels on his own time... I would do it.

A better approach might be a DC clamp on ammeter. Bruce Roe

[AzRoute66]

Everybody owns something they can throw on a panel. Most people don't own a meter, much less a DC current clamp on. Those that do own such a meter are generally the type that clamps it on before calling the cavalry.

[bcroe]

Everybody owns something they can throw on a panel. Most people don't own a meter, much less a DC current clamp on. Those that do own such a meter are generally the type that clamps it on before calling the cavalry.

In my opinion, if they plan to do any work on an array, or even keep an eye on performance, they
ought to get this meter. Honestly, I don't see how partially blocking a panel even works, never did
it here (have fixed 5 faults), maybe you could spell out the procedure you have in mind? Bruce Roe

[sensij]

In my opinion, if they plan to do any work on an array, or even keep an eye on performance, they
ought to get this meter. Honestly, I don't see how partially blocking a panel even works, never did
it here (have fixed 5 faults), maybe you could spell out the procedure you have in mind? Bruce Roe

When each panel has individual monitoring, as they do in Enphase systems, blocking a panel even partially will reduce its output relative to its neighbors, and allow that panel to be identified on the monitoring system. Knowing which panel is which is necessary when the monitoring indicates a problem with a panel, so that the right one can be serviced or replaced.

As Solarix is saying, if the update rate for the monitoring system is slow, it can take several hours to work through even a small array and identify each panel. Good installers record the position of each serial number so that this slow functional identification method isn't necessary.

[J.P.M.]

Everybody owns something they can throw on a panel. Most people don't own a meter, much less a DC current clamp on. Those that do own such a meter are generally the type that clamps it on before calling the cavalry.

I've got a piece of plywood cut to size. If used, I just lay it over a panel ad attach w/ bungie cords. I don't use it much anymore, but it works great.

[tyab]

"The internet monitoring takes quite a while to update its data so its not very practical to deduce where the microinverters are by covering the panels one at a time to see which ones drop out"

Next time try using the Enphase Installers Toolkit app (Android and IOS) It provides fast feedback since it is communicating directly with the local Envoy and not the servers.

"The Enphase s/n label is just a little paper thing that can come off and is only on the top surface "

I saw a hint on some other forum to write the Enphase SS number on the case front, back, and side with one of those thick sharpies prior to installing them. When we did my installation, wife was marking the inverters (front/back/side), filling in a location chart and kids were handing them in order to me for install. My M250's were all silver so this is highly visible but not sure what one would do if you had the black painted (or the newer plastic) ones.

[AzRoute66]

In my opinion, if they plan to do any work on an array, or even keep an eye on performance, they ought to get this meter. Honestly, I don't see how partially blocking a panel even works, never did it here (have fixed 5 faults), maybe you could spell out the procedure you have in mind? Bruce Roe

I agree if they plan to work on the array/electrical, they should have a meter and clamp on is nice. But most don't plan to work on it any more than they plan to buy an auto analyzer and work on their own Buick.

The technique I would use to find the bad ones differs from the technique I would use to re-map the entire array but only by a little bit.

So this guy has 28 panels so it will be gruesome even if you had a meter. Maybe harder with a meter depending on how 'tight' the array is. If I were looking for three bad panels in a field of 28, I think I would start with five pillows (or jackets or whatever he uses to block the panel).

1. I would place 4 pillows then check output. Either they will be on four good panels or there will be a bad one in there. If all four were good ones, pick up all four pillows and place them on the next four panels. Continue.

2. At some point, you will cover one or more bad panels (Group 3 for instance). Leave one of the pillows (Pillow A) where it is, and bring in the other pillow to move four onto the next group. At this point, Pillow A can now be used to map the exact number of the panel that it is on, and it can be observed to be on a good or bad panel at the same time you check Group 4. If Pillow A is on a good panel, move it onto another in Group 3 while moving the four pillows to Group 5. If Pillow A is still on a good panel, toss it onto either of the remaining two in Group 3 and you have found a culprit.

And so it goes. You should be able to ID all offending panels in about 10 moves max, and if you have been taking good notes of which inverters were which in the groups, you are more than halfway (about 3 more moves) to recreating a full and accurate array map.

[AzRoute66]

Since I'm here. To remap the entire array with the same methodology take 8 pillows.

1. Place the four pillows on group A.
2. Leave A1 and A2 in place and place four pillows on Group B.
3. Move A1 to A3. Leave B1 and B2 in place and put four pillows on Group C.
You now know Group A: A1 was covered then uncovered, A2 was covered both times, A3 was uncovered then covered, A4 was never covered.
4. Move B1 to B3. Leave C1 and C2 in place and put four pillows on Group D.
You now know Group B: same reasoning as Group A.
5. Move C1 to C3 . Leave D1 and D2 in place and put four pillows on Group E. And so it goes.

Another obvious solution only requires 7 pillows, but more moves - just in case pillows are in short supply.

[bcroe]

OK, pillows or plywood locate the panel on the individual panel monitoring system, then
the monitoring system measures panel performance. That can work.

When I bought my 62, an old time mechanic told me to get some appropriate tools to work
on it. He was right, too bad they never taught that in school. Here is my 79 resting on my
most expensive "car" tool. Bruce Roe


79onLift.jpg

[sensij]

Since I'm here. To remap the entire array with the same methodology take 8 pillows.

nice... but why limit yourself to binary? With the bypass diodes in most panels, you have quaternary states to work with.

[AzRoute66]

Bruce, I wish you were my neighbor. I could borrow your clamp on ammeter, and 'stuff'.

[AzRoute66]

nice... but why limit yourself to binary? With the bypass diodes in most panels, you have quaternary states to work with.

I am not that smart. And it would require potholders instead of pillows. I am going to need the pillows when I am done.

I would do that though, if the array was in proper working order and I was sure I could tell the difference between 3, 2, 1, or 0 panel columns covered. It would save me one whole move.