13 panel string putting out just over 200% of what 4 of the same panels put out.

dougl

I have 18 Sharp 165W panels with 13 of them in series going into an EG4 18KPV and I've ever seen this string put out 1,000W. But 4 of the same panels, at the same angle but attached to a micro-inverter connected via AC-coupling to the EG4 will regularly put out just over 400W.
The 13 should be producing 1,600W( 13*165W*0.77eff = 1651W ).
Extrapolating the 4 panels putting out 400W gives a 1,300W number( 400W/4 * 13 ).

Clearly something is off on the string of 13 panels. Anyone have any ideas on what I could do to find the bad panel(s) or what's going on?  They are 19 years old, I do keep them clean and I started seeing a drop in production 5 years ago when they used to be configured as 2 strings of 9 going into a Sunnyboy.  I have a clip-on current meter and could carefully clip a DVM on to read voltage if there was a way to pinpoint a bad panel while operational.

Thanks.

BB.

Some guesses:

Your hardware:
https://eg4electronics.com/categories/inverters/eg4-18kpv-12lv-all-in-one-hybrid-inverter/
https://www.datasheets.com/part-details/ne-165u1-sharp-47163058
Located in San Diego/Southern California?
Array tilt, ideally, around 33 Degrees from Horizontal

A quick look through the data sheets and assuming no horribly cold/sub zero weather, the 13 panel array (single string) seems to be a good match with your EG4 inverter/charger.

Nominally, on a hot day, I would expect the Vmp-hot-array to be around:

13 series panels * 34.6 volts Vmp per panel * 0.80 "hot panel" Vmp derating = 359.84 ~ 360 VDC Vmp-array-minimum in Hot weather and normal operation

I use the Midnite solar array planning tool... The numbers/warnings are for 150 VDC Max input controllers, not the 600 VDC max for your inverter/charger--But it still is quick and easy to use... For example, your max Voc-winter-array voltage (assuming 32F minimum temperature when sun is up), is around 607 VDC... So the "exact" low temperature could run into issues with your controller's Max input voltage (usually you do not want to ever exceed the controller's Max input voltage--Can void warranty, and even run the internal switching devices):
https://www.midnitesolar.com/sizingTool/index.php


Need to look at your array voltage around noon--Is it in the range of 360 VDC minimum and probably around 450 VDC Vmp-array-cool day...

The current from the array, again, clear day, near solar noon with your panels pointing pretty much at the sun within 10 degrees or so--Your array current should range from 4.77 to 1/2 that or 2.385 Vmp (hazy day, etc.).

Typically, if your array voltage is not in the 360-450 VDC range, and the array current is outside (below) ~2.585-4.77 Amp range--Then you probably should look at some more detailed testing (basically, use DC Clamp Meter to confirm your current reading on the EG4 and measure the input voltage--These numbers should be pretty close to EG4 readings (at worst, typically +/- 5% max disagreement with hand held meter (good quality, good batteries, known working condition).

And then go on roof and disconnect each panel and do a Voc/Isc test on each panel (will typically find ~80% of any failed panels).

Also look closely at your panels... Look for cracked cells, browning/blacking of cells/color changing-corrosion of cells and/or interconnects/backing about J-Box, etc...

Your 1,000 Watts is well below what I would expect from your array (typical "worst case" around 1,651 Watts... However, with chargers--obvisouly the state of charge of the battery bank/loads/etc. can affect what the inverter/charger thinks it needs to harvest from the array. If configured correctly, the Grid Tie section should any/all power from the array that the DC charger section does not "need".

-Bill

dougl

Wow Bill, you nailed the system for sure and thanks for pointing out that app, looks amazing. I did the numbers manually and having run the system configured as 13 series panels on the Sunnyboy 2500U for a few years I'd never seen it go over 400V and that inverter has a 500VDC max. We are coastal so 32F is very rare and it warms up quickly as the winds shift from out of the east to out of the west early.

Someone recommended looking at the array with a FLiR camera to find a hot panel or cell and I'm going to try and find a FLiR camera to borrow to try that first.  Removing panels to disconnect them will likely mean snapping clamp bolts and doing that on a ladder has some safety concerns. I might also try putting my phone on a selfie stick and getting pics of the panels to look for defects. Another great idea!

My path looks like this right now:
1) FLiR camera to look for hot zones or panel
2) Selfie stick to get high rez pics of each panel to look for defects like cracks, discoloration, etc
3) try to remove the eastern most 2 panels(shaded in the AM and seriously limit output) and do VoC and IsC tests on them comparing to 1 or 2 of the other panels.

Thanks again for the insights and putting the effort to spec and system so completely.

BB.

A FLIR camera would be wonderful... Always wanted one, but just could not justify the costs.

You may have to look further into the wiring... The panel plug connectors are known to be a weak point in a system... From poor crimps/corrosion/overheating/etc... But there is a good chance that the array would just fail (burn the bad connection out, wire fall out of crimp)--Rather than just a 20% reduction in output (for a single string--Just a guess)--So probably not the first place to look.

Again, checking the current and array voltage under operation may give you some good clues... A "bad panel" may drop a whole Vmp (i.e., around 35 volts per panel), and other guesses from the V and A of the system.

Another 'quick test' you may try... Put a square of cardboard on a long pole and cover one panel at a time... See if you can find a panel that does not affect your array output (a good panel should drop voltage/wattage by ~1 panels worth when covered by the sun... A "bad panel" with shading would have much less of an output drop). Just some SWAGs here that may save you some ladder time.

Take care,

-Bill

dougl

BB. said:

Some guesses:

Your hardware:
https://eg4electronics.com/categories/inverters/eg4-18kpv-12lv-all-in-one-hybrid-inverter/
https://www.datasheets.com/part-details/ne-165u1-sharp-47163058
Located in San Diego/Southern California?
Array tilt, ideally, around 33 Degrees from Horizontal

A quick look through the data sheets and assuming no horribly cold/sub zero weather, the 13 panel array (single string) seems to be a good match with your EG4 inverter/charger.

Nominally, on a hot day, I would expect the Vmp-hot-array to be around:

13 series panels * 34.6 volts Vmp per panel * 0.80 "hot panel" Vmp derating = 359.84 ~ 360 VDC Vmp-array-minimum in Hot weather and normal operation

At noon today with the sun azimuth direct to the panels(~164), in 95 degF temp and hiproof pitch of 18.4deg I read 223.8 VDC at 3.82 A

The EG4 18KPV showed 223.7V at 866W.

Need to look at your array voltage around noon--Is it in the range of 360 VDC minimum and probably around 450 VDC Vmp-array-cool day...

The current from the array, again, clear day, near solar noon with your panels pointing pretty much at the sun within 10 degrees or so--Your array current should range from 4.77 to 1/2 that or 2.385 Vmp (hazy day, etc.).

BB.

As a first guess, it sounds like you have a few panels that have failed. My guess is that multiple failed panels are still passing current via the "bypass diodes" accounting for the loss of Vmp-array overall string voltage.

Probably time to think about what happens next... Such as an entirely new array.

-Bill

dougl

BB. said:

As a first guess, it sounds like you have a few panels that have failed. My guess is that multiple failed panels are still passing current via the "bypass diodes" accounting for the loss of Vmp-array overall string voltage.
-Bill

I did the test covering the panels but could not get a helper so I had to rely on the mobile display of the EG4 and I found what looked like possibly 3 questionable panels. My test method was to get a nominal measure (1 minute ) from the system/app, cover the panel and wait for the EG4 monitor app to update and record the new voltage and power output.  I saw 80-100W changes on most panels but 3 panels I saw wattage changes of only 10W-30W.

I replaced one panel today having saw > 4.0A when doing the Isc test. I did not do a Voc test on the panel but maybe I should have.  After replacing the one panel which showed only a 10W difference in overall power output when covered vs uncovered I do not see any difference in overall power output of the system compared to the same time yesterday. Sunlight seems consistent. I would have expected to see >100W boost compared to yesterday.

I have 4 more of the 165W panels which are known good and will swap 2 of those in over the next few days when I get a window to do it. The swap should go quicker since this swap required cutting a connector, opening a junction box and hard wiring since it was at the end of the string.  When I did an Isc test on the removed panel I got 0A so it was a bad panel.

BB.

Three bad panels sound about right... You could lose upwards of 30-40 volts per "dead" panel... Your present Vmp-array is something like 100-120 VDC "less" than I would have expected from a 13 panel array...

Glad moving the shade from panel to panel allowed you to (it seems) pretty quickly locate the "failing" panels. You are the first "guinea pig" I have suggested this this test method.

-Bill

dougl

It got more interesting today Bill.  Summarizing, 13 165W panels in series configured in 3 rows( 1=7, 2=4, 3=2 ). Originally 3 or 4 panels which showed low wattage drops when covered and I had replaced 1 previously with a known good panel and validated the removed panel put out no current doing the Isc test so was truly bad but was passing current(bypass diodes).  Surprisingly I saw no change in output power after replacing that panel.  Today I was ready to remove the 2 row 3 panels which tested good(Voc and Isc tests easy to do) and use them replace what was seen as bad panels in row 1 and row 2.  I started disconnecting panels on the left/east side doing Voc/Isc tests. The first one measured 25V Voc while a good panel tested 38V Voc so it was moved out and replaced. Another panel showed a similar low Voc so it too was moved out of the row and replaced.  I then tested all 4 in row 2 by doing Voc/Isc tests and they measured good.
I tested the left/east 3 in row 1 doing Voc/Isc tests and they were consistent/good.  I did a combined Voc test of the remaining 4 in row 1 which were still in series and their voltage equalled 4x what one of the good ones tested so it seemed I now had 11 working panels in the string.  11 * 165W = 1,815W  * .77(eff factor) nets 1,397W

But I connected the PV string at the EG4 and after 15 minutes at 12:30p(solar noon) it was only putting out 935W at ~4.0A.

Either the EG4's MPPT is not setup to work well with these old panels or I missed something. 

What's also odd is that I had placed the 2 "bad" panels on the 3rd row rack to get them out of the way and had shorted their outputs together. When I walked by them I put my current clamp on their wires and read 4.77A on both of them. Their output voltages were lower but since it was easy to connect them back into the string because I had only jumpered that series connector when removing the 2 working panels from that row. So I put them back into the series and saw the output jump a few watts( maybe 3W ). I guess that's not odd, their low voltage didn't add much power and they didn't drop any power so bipass diodes are working for the failed cells in the panels.

I don't get why I again only saw 935W for the 11 'good' panel string when it should have hit 1300W-1400W. 
At 12:30p I was seeing 424W off an array of 4 of these 165W panels feeding a 700W micro-inverter. 424/4=106W and 106W * 11 good panels = 1,166W
I guess that's not too far off from 935W( 231W ) so maybe the losses are in the long #10 wire, solar disconnect and the EG4 PV disconnect.  Two hours of effort, no additional power gain and not really anything learned.

I'm purchasing an MPPT meter for ~$90 which can handle upto 1600W and up to 60V so will connect to each panel and test their operation MPPT data.

UPDATE:  using the MPPT meter the 2 panels removed on Saturday were clearly bad with Vmp of 16V and 20V and Voc of 26V and 25V. REALLY interesting was that 2 of the panels I have in another small array showed MPPT values of 28V/4.0A and 28V/3.7A  so now I know what's holding the big string back. One or more panels with low Imp output which is pulling the whole array down. Now I have to go back up and isolate every panel, test them and swap out any panels which are low provided I only have 3 or fewer of them since 3 of the 4 on the smaller array look like they have Imp over 4.0A.

dougl

I'm leaning towards Bill's recommendation and 86'ing the current 165W panel array and replacing with new panels.
I measured 4 165W panels and 3 of the 4 look ok but add that with the fact that 1 panel is 100% dead(0V/0A) and 2 are half dead( ~24V instead of 34V ) added with the fact that there's at least one more weak one in the 11 panel array which is keeping it producing under 1kW it's time to recycle these panels. But, it's not so easy since new panels are much larger in physical size. 

The old Sharp NE165W panels are: 1575 x 826 x 46mm
I could fit 7 across the first row/rails and 4 on the 2nd row/rails.

New panels like the Aptos DNA-120-MF10-440W
1903 x 1134 x 35mm  so I would be lucky to be able to fit 6 of these on the existing roof and rails. It's not a big house and it's got a hip roof. I also wonder how the EG4 MPPT will handle such a low voltage string since Vmppt is 34V so that's only 204V.  Granted, the spec says 100-600VDC but will it be efficient at the low end of ~200VDC?

There are some good prices now that I've started looking.

BB.

The specs

MPPT OPERATING VOLTAGE RANGE 140-500 VDC
NOMINAL MPPT VOLTAGE 360 VDC

Startup is 100 vdc minimum... No problem.

Typical worst case mppt voltage

6 x 34 vdc Vmp x 0.81 hot panel derate = 165 VDC minimum array Vmp

Looks ok... But you might look around for good price ok higher Vmp panels (are around, but may be difficult to find) or smaller panels (more in series) but may cost more per watt...

-Bill

dougl

Ya, looks like the Vmp will be too low. I have to find out what that "Nominal MPPT Voltage 360 VDC" is all about.
Having been on the roof too much recently I noticed it's also on its last leg and I'm replacing panels and need to move racks I might as well put a new roof on and if I'm moving racks I have more panel choices. I found some Aptos 400W for a great price of $120ea but with a Vmp of 31.28V even with a 7 panel string the derated voltage is 177V and nom-voltage is only 219V.  There is a raised rack on the back slope of the roof where I was going to put another 7 panels and modify the vertical posts for 2 or 3 seasonal adjustments to keep the future battery topped up through the late fall, winter and early spring.

But if the array and EG4 are not going to do well at ~200V/string and needs 360V then I have to start looking for ways to get more panels or find some 72 cell panels instead of the 60 cell panels I'm finding locally.   

I never knew MPPT was this picky and thought the electronics just worked it out optimally. And I was able to turn 4 of the Aptos panels sideways along the longest bottom row, put 4 vertically on the 2nd row and 1 sideways on a 3rd row for a total of 9 panels. I can now probably setup the rack on the back side of the roof for a 50 deg tilt for fall/winter/Spring and do ok. With these on the inverter MPPT inputs the inverter can throttle back export once the battery is full.  Thanks for keeping my mind fed Bill.  

BB.

According to specifications (assuming mfg's specs are correct, and unit is functioning correctly), the MPPT range is 140-500 volts... If you are below 140 or above 500 VDC, the MPPT conversion will be "less efficient".

If the array voltage is below 100 VDC, the EG4 will  not start. And above 600 VDC, the EG4 can be damaged and fail.

The "optimum" MPPT voltage of 360 VDC is a quick array design point that will work in "any" weather conditions (from very cold to very hot). And will allow the highest Wattage/nominal voltage array with the lowest current (Pmp=Vmp*Imp).

I.e., a 5,000 Watt array running Vmp~140 volts will have (purely made up numbers to show the "math"):

5,000 Watts / 140 volts = 35.7 Amps Imp-array-std
140 VDC * 0.03 typical voltage drop (3% typical drop) = 4.2 "max" desirable voltage drop

Whereas a 360 volt array will have:

5,000 Watts / 360 VDC Vmp-array = 13.9 Amps Imp-array-std
360 Volt array * 0.03 typical max drop = 10.8 volt wiring drop (support longer cable runs and/or smaller diameter copper cable).

The "high voltage" array has less than 1/2 the current and cause use much smaller diameter (AWG) copper wire (smaller wire for smaller current, and higher supported voltage drop at higher array voltage).

Does this make sense?

-Bill

dougl

Ah, so the 360V is just a 'sweet spot' since it's high enough for lots of power at lower current but also not going to see voltages spike over the 500V limit on cold morning nor drop below the 150V low threshold.

a 150V-500V will be tested soon. I connected 4 340W panels having Vmp 34V so ttl Vmp=136V and I saw the EG4 get next to nothing from them at a time of day when micro-inverters were getting well over 500W from them. I recently found some 400W panels with Vmp=31V( I know, kinda low ) but I will test 6 of them(Vmp=186V) on the EG4 soon.

Thanks again for explaining and finding the spec stating the EG4 MPPT range is 150V-500V when I thought it was 100V-500V.

dougl

What I found out was that 7 of the original 19 Sharp 165W panels are bad with most of those showing only 1/2 of their nom SC voltage.

As for the lower PV voltage into the EG4 goes, I ended up moving my rack down the roof enough to get a 7th panel in the string so a Vmp of 217V(31V*7) and they've been working fine and getting to 2550W peak at a 30 deg tilt. They are bi-facial and seem to hold up generation about an hour longer then the old Sharps flat on the roof.

BB.

Glad you where able to sort that out and get a good array up and running.

Just for completeness, "SC Voltage"??? Do you mean Short Circuit Current or Open Circuit Voltage (or something else)?

-Bill

dougl

Right, my head was elsewhere when I wrote SC Voltage. I was trying to picture the data on the MPPT meter I'd purchased and made the latest tests with and wasn't thinking for sure. Previously I had done tests covering panels and watching output power while connected to the inverter and it did show a number of bad panels( 3 of them ).
Was driving me nuts why the 11 remaining were not putting out well over 900W so I bought the MPPT meter and individually tested again and it showed the 4 panels putting out half the voltage(Open Circuit Voltage) and I didn't bother looking at nor recording the MPPT wattage.  Re-racking it all and replacing with new panels.

Will see what Sharp US says about 7 out 18 panels going bad in 20 years. Actually sooner because I had seen a 25% drop in output in 2017 just didn't act on it other than trying to figure out a way to evaluate without disassembly.