installation producing 3x less power and half the voltage

[MichKrom]

Update 2:

I have purchased a "new" SPRx-3300-R inverter on eBay that had the same SunPower's P/N as my original one.

Problem 1: Since my original model did not have "-R" I have contacted SunPower and wrestled confirmation from them that these are equivalent. However, they told me not to call them again as I am just a (unlucky?) owner and I have purchased the replacement on eBay. So far so good.

I have replaced it yesterday.And yes, it now reads correct voltage on the DC side, at 300V unloaded. So yet this DC under-voltage was the inverter's failed DC measurements.
Here are the pictures: https://photos.app.goo.gl/XhUPP62z5YTnsxrK8


Problem2: However I must be super unlucky (cursed? ) as the "new" inverter has problem with AC and reports under-voltage with very inconsistent readings between 160-210V. I have directly measured AC on its leads at 240V so again the new inverter is broken. Additionally I noticed that the claimed "new" inverter logged 8kWh produced so far so I suspect it failed in testing or in first installation. I am returning it (that's a good news - the seller unconditionally accepted the return, yay! so prob just loose $100 to ship it back).

That leaves me with the problem of what to do now:

Does anybody repairs the inverters?

Replacement route: it looks that all inverters on the market now are TL (transformer less) and expect un-grounded PV connection. How can I use one like this with my positive-ground PV system? Here is a possibility: https://www.renvu.com/Learn/Upgradin...verter-systems

[bcroe]

Not familiar with your inverter, but it does sound defective. The advantage here
with 2 inverters, is ability to switch things around to determine faults. Bruce Roe

[MichKrom]

Even a weak section may generate normal voltage with no load current. I find my
faults under best sun, after setting up measurements the night before. Bruce Roe

Bruce, yes, undrestood. If there is a bad connection having significant resistance, it will still measure good voltage under no-load condition, and it will only start dropping the voltage when the load (current) increases, per Ohm law.

My point is that I have measured the voltage at the inverter input (junction box under the inverter) and compared it immediately to what inverter reported: 295V vs 164V. That was in full sunshine, no load (well at least the inverter was off line but powered by the solars's DC....so very small stand-by load, 0.2A) . Why does inverter report only 164V? Looks like it's faulty to me?

[bcroe]

Even a weak section may generate normal voltage with no load current. I find my
faults under best sun, after setting up measurements the night before. Bruce Roe

[MichKrom]

Update: finally got some time and courage and measured the voltage comming out of the array right at the inverter input/connection panel. . I measured 295V while the inverter was reporting only 164V. This was under no load conditions as at the time the line power was off (yea i am in PG&E fire safety disconnect zone, no power for second day). So this points to the inverter being malfunctioning, likely? The other good inverter was reporting 245V.

[peakbagger]

Work at night!. MC connectors used in many solar arrays are not designed for load break and if disconnected during the daytime will usually be damaged by an arc. My friend years ago disconnected his array at the inverter in dark barn during the day. he was quite impressed by the length of the arc and how bright it was. He works on his at night these days.

Just get yourself a good headlamp and you will be all set.

[bcroe]

Bruce,

Yes, I think I understand I need to get me an electrician gloves... Just trying to defer this a bit but I suppose there is no magic, gotta do it.

What do you use to cover up the array of this size? Or is it better to work at night? Michal

Yes, talking about it will never fix it. DO NOT try to make the array harmless to
work on days, covering it will not prevent light on the panel back from generating
the same lethal voltages (at much reduced current, will just kill you, not fry you like
a hot dog). Voltage measurements are possible, disconnecting wires is another
level. I do my wiring at night under lights. Opening a rated switch is possible, I
avoid it much once power is being generated. Bruce Roe

[MichKrom]

Bruce,

Yes, I think I understand I need to get me an electrician gloves... Just trying to defer this a bit but I suppose there is no magic, gotta do it.

What do you use to cover up the array of this size? Or is it better to work at night?

Michal

[bcroe]

I am sorta answering my questions. Here is a great discussion about connecting multiple string arrays in parallel:

So...it's complicated.

I suppose when one or more panels in one string are underperforming (shadow or damage), they will drag the whole array down when the inverter is searching for optimal power generation on I/V curves. And...it may get the wrong spot and get stuck there?

Although it's hard to explain 2x-3x power difference with this...

Yes it is complicated, predicting how the MPPT circuit will respond to multiple bad
strings is not something worth messing with. That is why at the beginning I told
you to try each string by itself. And I will add, after you find one or two bad
strings, do the troubleshooting on one at a time operating. Bruce Roe

[MichKrom]

I am sorta answering my questions. Here is a great discussion about connecting multiple string arrays in parallel:
https://forum.solar-electric.com/dis...tring-voltages

So...it's complicated.

I suppose when one or more panels in one string are underperforming (shadow or damage), they will drag the whole array down when the inverter is searching for optimal power generation on I/V curves. And...it may get the wrong spot and get stuck there?

Although it's hard to explain 2x-3x power difference with this...

[bcroe]

Low string voltage most likely means some sections in bypass. Might be a whole panel, or
might be a section. If you throw a complete cover over a panel and an operating string
voltage does not change, that panel is bypassing. You might be able to look over panel
wiring paths and make a neat cover for just section of a panel. Then check covering each
section to see if it affects output voltage or shut down, if not X it as bad. There must be a
few bad ones from your description.

Inserting 3 way taps and measuring voltage around the string is my usual method. Bruce Roe

[Mike90250]

erase the phrase "Blocking Diodes" from your head. Most gear built since the 80's do not need blocking diodes. Some troll is passing that around the web to cause trouble.

[MichKrom]

(posting again, sorry if duplicate)

Yes, I am not the original owner. House came with it. The solar installation was put in operation in Jan'08.

The pictures were taken at 7PM so shading definitely occurs then, from the West. In the early morning (say 8am) the trees may give some shade (they are due E). Otherwise, full Monty

>> I doubt that will produce much more than a hazardous condition and additional confusion.

What "hazardous condition" can occur when covering panels? I am hoping to pinpoint which panels are limping and if any? And I am trying a non-electrocution-prone approach first

Questions:
1) Since there are two strings in each array in parallel, should these be connected via blocking diodes? (There is no mention of blocking diodes in design docs. I do not know if there are blocking diodes in combiner box.)
2) If there is no blocking diodes and one string is weaker than the other (shaded or failing), would this cause problems (power form stronger string gets consumed by weaker side and potentially damages them)?

Thanks for all the discussion here, I am learning "the cables" !

[J.P.M.]

Nothing that strong but there are some that have different tint.

Here are the pictures:

https://photos.app.goo.gl/BgcbwVtjwuQmejRf6

Plot thickens:
- at ~2PM 190V/7A and 250/9A - note that panels are rated at 5A so with two strings in parallel it should be possible to get to 9A
- at ~4PM both arrays started to produce the same amount of power (according to inverters) still with different voltages but similar amps (190V/250V and both at ~4A and reported power about 1kW)

Something is not adding up.

I will be doing some more experiments:
- covering pannel by pannel and observing the changes
- turning the house ooff and using the SmartMeter to measure power sent to the grid

The photos are interesting for what they show besides what looks like some defective panels. I'd call Sunpower or the vendor (if they are still around) on that issue.

1.) You do know that one bad panel in a string will affect output more than just losing the output of that one panel, right ?

2.) Same applies to shading. A string or and array that's partially shaded will have output reduced by more than the % of the string or array that's shaded. You will notice the same effect if you partially cover the array, panel by panel. Without some additional measurements which might be quite dangerous, I doubt that will produce much more than a hazardous condition and additional confusion.

I noticed some trees and shades in the photos.

You say the array is old but you are new to solar. Am I to take that to mean you are not the original owner of the array ?

[MichKrom]

I think I have misread the inverter readings.

The only time a no-load readings are available is after rebooting the inverters and before they try to go on-line (they have 305 second wait period). I read 224V and 294V respectedly per array and zero Amps at this time. Otherwise, when inverters went through initial "wait" they will try to put a load on the arrays all the time (and hence report non-zero amps). Now, at noon I read 187V/0.3A (off line) and 245V/9.4A (~2kW output). So there is definitely sth wrong with battery 1 but I now cannot rule out connections or panels.

It is weird that when the bad one eventually goes on-line it can produce some output (usually around 4A but I have observed it up to 7A; just always <180V). Also, since the arrays face due-south but have 30% angle, I suppose best output is in the afternoon (instead of noon)? This would be consistent with the bad array turning on later during the day as it needs more sun to overcome its deficiency?

Another question: should not the two strings in an array be connected via blocking diodes? My desin docs do not mention anything like that (they did state 2 string with 7 panels each - so my guess was right). If they are not via blocking diodes what happens when a panel fails in one string? Would not power from one string then go to the other?