New solar and Powerwall install, low production?

[taphil]

Hi, I'd appreciate some analysis on my new installation and if there's a problem with it.

I got a 8.16kW 24 panel system plus 2 Powerwalls installed in March. Located about 20 miles east of Downtown Los Angeles. Telsa Gateway 2, Tesla 7.6kW inverter, and the two Powerwalls are installed in the garage. Layout for 24 panels is shown in picture, they face slightly southeast at about 195 degrees. The system passed city inspection last week, and I paid invoice in full the day after. Now just waiting on PTO.



I flipped the solar on a few times since the install to check the system. It's working in terms of solar charging the Powerwall and supplying the grid after batteries are fully charged, and the Powerwall with solar was able to power the entire house for 100h when I switched off power from the grid, though the battery would not be fully charged every day by sundown, and eventually after a few more days would have hit the 20% reserve in the middle of the night. On a normal night, the battery will discharge 50%, no AC running at all.

However, overall solar production seems low. Tesla's contract estimated 13,800kWh yearly production, and when I plug in my numbers into PVWatts, I get 13,730kWh. For April, it estimates production of 1223kWh, or a daily production of about 40kWh. However, on the sunniest of days in April, my system is only generating about 27-29kWh. The panels have no shading and no obstructions.



I've attached pics of the inside of the boxes. The inverter has 3 wires coming from the solar panel conduit, which I assume are DC positive, negative, and ground. They are connected to the 2+ and 2- labeled inputs, and there's jumper wires above connecting 1+/2+ and 1-/2-. Does this mean all 24 panels are on one string? Or did they forget to run the line for the second string? At peak 12-1pm output, a clamp meter shows +9.5A and -9.5A on the wires coming from the solar conduit, I don't know the voltage. The two black and red output wires from the inverter to the Gateway shows 14.5A on each wire, I don't know the voltage. There is one CT in the gateway clamped onto the red wire from the inverter.



At the time of peak output, the inverter shows strings 1 and 2 active, each at 345V and 5.1A. Strings 3 and 4 are not connected. The AC output shows 14.1A, which corresponds to my clamp meter. The Tesla app shows solar production at 3.8kW, a little higher than what the inverter webpage shows. The most I have ever seen output is 4.1kW briefly on one day, but otherwise peak output hovers at about 3.8-3.9kW on a cloudless sunny day.



Does the installation seem correct? Or is something wrong? I'd appreciate any input so I can get an idea of what's going on before I contact Tesla.

 

[Ampster]

It would be hard to make a diagnosis from just flipping the breaker a few times. I need a few months with my system befire i could reach any conclusion Just changes of weather make big differences in the month to same month comparisons.

 

[nwdiver]

Hi, I'd appreciate some analysis on my new installation and if there's a problem with it.

I got a 8.16kW 24 panel system plus 2 Powerwalls installed in March. Located about 20 miles east of Downtown Los Angeles. Telsa Gateway 2, Tesla 7.6kW inverter, and the two Powerwalls are installed in the garage. Layout for 24 panels is shown in picture, they face slightly southeast at about 195 degrees. The system passed city inspection last week, and I paid invoice in full the day after. Now just waiting on PTO.

View attachment 654892 View attachment 654891 View attachment 654893

I flipped the solar on a few times since the install to check the system. It's working in terms of solar charging the Powerwall and supplying the grid after batteries are fully charged, and the Powerwall with solar was able to power the entire house for 100h when I switched off power from the grid, though the battery would not be fully charged every day by sundown, and eventually after a few more days would have hit the 20% reserve in the middle of the night. On a normal night, the battery will discharge 50%, no AC running at all.

However, overall solar production seems low. Tesla's contract estimated 13,800kWh yearly production, and when I plug in my numbers into PVWatts, I get 13,730kWh. For April, it estimates production of 1223kWh, or a daily production of about 40kWh. However, on the sunniest of days in April, my system is only generating about 27-29kWh. The panels have no shading and no obstructions.

View attachment 654894 View attachment 654898

I've attached pics of the inside of the boxes. The inverter has 3 wires coming from the solar panel conduit, which I assume are DC positive, negative, and ground. They are connected to the 2+ and 2- labeled inputs, and there's jumper wires above connecting 1+/2+ and 1-/2-. Does this mean all 24 panels are on one string? Or did they forget to run the line for the second string? At peak 12-1pm output, a clamp meter shows +9.5A and -9.5A on the wires coming from the solar conduit, I don't know the voltage. The two black and red output wires from the inverter to the Gateway shows 14.5A on each wire, I don't know the voltage. There is one CT in the gateway clamped onto the red wire from the inverter.

View attachment 654900 View attachment 654899 View attachment 654901

At the time of peak output, the inverter shows strings 1 and 2 active, each at 345V and 5.1A. Strings 3 and 4 are not connected. The AC output shows 14.1A, which corresponds to my clamp meter. The Tesla app shows solar production at 3.8kW, a little higher than what the inverter webpage shows. The most I have ever seen output is 4.1kW briefly on one day, but otherwise peak output hovers at about 3.8-3.9kW on a cloudless sunny day.

View attachment 654902 View attachment 654903

Does the installation seem correct? Or is something wrong? I'd appreciate any input so I can get an idea of what's going on before I contact Tesla.

.... something definitely seems off to me... it looks like they tried to parallel the strings on the roof, which is dumb and lazy IMHO. The whole purpose to 4 MPPT channels is so each string can operate independently. If you parallel the strings on the roof you lose that flexibility. Even if you don't have shade bypass diodes can fail, some panels might be cooler or warmer than others, etc. Regardless it's pretty clear to me that only one string is actually operating. Total current should be ~10A per string for ~20A total unless you have an unusual panel type.

 

[arnolddeleon]

.... something definitely seems off to me... it looks like they tried to parallel the strings on the roof, which is dumb and lazy IMHO. The whole purpose to 4 MPPT channels is so each string can operate independently. If you parallel the strings on the roof you lose that flexibility. Even if you don't have shade bypass diodes can fail, some panels might be cooler or warmer than others, etc. Regardless it's pretty clear to me that only one string is actually operating. Total current should be ~10A per string for ~20A total unless you have an unusual panel type.

or they just flat out forgot to connect the second string. The output is not adding up they are actually in parallel. They clearly can't be in series since 24 of those modules would exceed the voltage input. Power production should be closer to 40 kWh for the day vs 20 kWh.

 

[nwdiver]

or they just flat out forgot to connect the second string. The output is not adding up they are actually in parallel.

Yep. What also annoys me is that they clearly intended to parallel on the roof since the jumper is installed. IMO that's not acceptable. In addition to the reasons I mentioned above you also lose a valuable diagnostic tool. If the strings were separate you could compare the voltage and current values to determine the health of the strings. If a panel failed the voltage of one string would be ~8% lower than the other. If they're paralleled on the roof the voltages will always be identical.... the cost to run a separate line for the other string is <$20..... Even at $2/w I expect more than this.... do better Tesla

 

[taphil]

The panels are just the Q.Peak Duo Blk-g6+ (made in South Korea when I looked at the panels as they were being installed).

So it seems like only 12 panels in series are connected? In the past, the inverter webpage has shown up to 445V at 5A for string 1 and 2, which would make sense if the string were 12 panels in series?

What's the best way to connect the 24 panels? 2 strings of 12 in series? I don't understand the disadvantage of wiring a group of panels in parallel.

 

[nwdiver]

The panels are just the Q.Peak Duo Blk-g6+ (made in South Korea when I looked at the panels as they were being installed).

So it seems like only 12 panels in series are connected? In the past, the inverter webpage has shown up to 445V at 5A for string 1 and 2, which would make sense if the string were 12 panels in series?

What's the best way to connect the 24 panels? 2 strings of 12 in series? I don't understand the disadvantage of wiring a group of panels in parallel.

The data sheet for those panels says their max power current is ~10A. So if that screenshot was near noon you definitely only have one string wired. The current is split at the inverter with the jumper between Channels 1 & 2. I think each channel can only take 10A so if they had wired the array as they likely intended it would have maxed out 1 channel... hence the jumper.

The best way to connect 24 panels to that inverter would be two independent strings of 12. You should have 4 wires not 2. The benefit to this is that each string of 12 can operate independently. If one panel is damaged or shaded the MPP channel can lower the voltage to allow the other 11 to operate normally. You can also compare channels 1 & 2 to verify all panels are healthy.... unless each string has a broken panel....

 

[srbobc]

My identical system is producing about 52kwh per day this month. Northern CA...

 

[nwdiver]

My identical system is producing about 52kwh per day this month. Northern CA...

Do you know if they pulled the same jumper crap or did they actually wire the strings separately?

 

[taphil]

Thanks for all the info. I suspected something was very wrong. I will attempt to enter the Tesla customer service gauntlet tomorrow....

That said, the installation crews on the 2 days of install were nice and professional as far as I had known at the time, and they put all the boxes, Powerwalls, and conduits where I requested.

 

[taphil]

Just for posterity, I thought I'd add the timeline journey to this point. LA County with SCE.

1/15/21 - Signed up for 8kW system and paid $100 deposit
1/16/21 - Uploaded additional pics of my house that they requested
2/19/21 - Home inspection completed (I was lazy and put it off for a few weeks)
2/24/21 - Changed my mind and added 2 Powerwalls
2/25/21 - Tesla started new permit application
3/9/21 - Got new purchase agreement and e-signed
3/9/21 - Scheduled installation for both solar panels and Powerwalls (soonest availability was within 3 days)
3/22/21 - Day 1 installation
3/23/21 - Day 2 installation completed
4/11/21 - Received email in evening stating city inspection would be next day (last minute notice!)
4/12/21 - Passed city inspection
4/12/21 - Received invoice
4/13/21 - Paid via credit card

 

[arnolddeleon]

Yep. What also annoys me is that they clearly intended to parallel on the roof since the jumper is installed. IMO that's not acceptable. In addition to the reasons I mentioned above you also lose a valuable diagnostic tool. If the strings were separate you could compare the voltage and current values to determine the health of the strings. If a panel failed the voltage of one string would be ~8% lower than the other. If they're paralleled on the roof the voltages will always be identical.... the cost to run a separate line for the other string is <$20..... Even at $2/w I expect more than this.... do better Tesla

I would be curious what the single line diagram shows. Given the inverter had the MPPT channel capacity it would seem silly not to use it. Would they have needed to use a larger conduit?

@taphil , ask for a copy of the plans for your project. It's good to have it saved away. I find it annoying that is not part of the document set available to download from the portal.

 

[CrazyRabbit]

9 to 10 amps is correct for a single panel, voltage will be around 34 volts when at optimal power point. so if they had all 24 panels in ONE string the voltage would be around 34*24=816 volts, open circuit voltage much much higher. that would fry the inverter. also OP measured about 9.5 amps, so strings are not in parallel. conclusion, they only wired up half the panels!

 

[Rapidrezults]

After reading this post, I was curious and went and took a picture of my recently installed inverter. I got a 8.16Kw system with 24 panels and 2 Powerwalls. I made sure they ran 4 separate strings and 4 seperate MPPT inputs for reasons mentioned above but never actually double checked the work (didn't think I would need to). Here is a pic for reference. I am still waiting for inspection so my system has been turned off.

 

[nwdiver]

After reading this post, I was curious and went and took a picture of my recently installed inverter. I got a 8.16Kw system with 24 panels and 2 Powerwalls. I made sure they ran 4 separate strings and 4 seperate MPPT inputs for reasons mentioned above but never actually double checked the work (didn't think I would need to). Here is a pic for reference. I am still waiting for inspection so my system has been turned off.

That's weird.... do you have a complicated roof arrangement? Typically longer strings are better... it's just the parallel strings that can cause problems. It IS good that they didn't do any parallels...

 

[Rapidrezults]

That's weird.... do you have a complicated roof arrangement? Typically longer strings are better... it's just the parallel strings that can cause problems. It IS good that they didn't do any parallels...

I guess you could call it that? I'm really new to the solar scene so any input would be appreciated. Here is my layout. The front house cluster is 63 Azimuth the back two clusters are 243. They split the front cluster into two seperate strings I guess. Pitch is 22. I have my pool thermal solar where the big box is drawn, and I wasn't willing to part with that even though it was prime real estate for PV.

 

[nwdiver]

I guess you could call it that? I'm really new to the solar scene so any input would be appreciated. Here is my layout. The front house cluster is 63 Azimuth the back two clusters are 243. They split the front cluster into two seperate strings I guess. Pitch is 22. I have my pool thermal solar where the big box is drawn, and I wasn't willing to part with that even though it was prime real estate for PV.

View attachment 654975

Ah.... that makes sense. I would have done 4 strings there too. SW String, SE String and the 14 panels facing NE is too many for 1 string... so 2 strings of 7.

 

[dareed1]

I ran the numbers for the OP's location. The peak insolation on your roof, assuming a roof pitch of 21 degrees, is 1030W/m2, so if your system ran at 100% efficiency, the 24 panels should yield 8.4kW at peak. It appears thus that you have one string connected and the system is yielding 92% efficiency. Roof pitch is not very important for your house because the direction the panels face is not straight south.

For comparison of symptoms and some idea of how this can be done incorrectly, we have a 24 panel system installed in December with a Delta inverter. They wired the two strings in parallel on the roof and that gave a flat topped power curve, unlike your smoothly peaked one. The limitation was the 12Amps/MPPT channel in the Delta inverter. Subsequently they ran two sets of wires down from the roof and connected them to two MPPT channels, but put a jumper across the positive terminals of the two channels. The inverter now showed nearly identical voltages and currents for the two MPPT channels. This arrangement provided a peaked curve with the expected power output at peak, but the two strings are south and west facing so the MPPT logic would get confused in morning or evening with just slight amounts of shading. On April 12, a tech came and discovered the jumper, removed it but did jumper between one of the channels and the unused channel. This gives one channel with 9.5A at peak, but two other channels with 4.8A at peak. The string connected to the two jumpered MPPT channels still has some tracking difficulty. Instead of involving Tesla, at this point I plan to remove the jumper myself so that the setup will be 2 strings on 2 MPPT channels.

Anyway, if the Tesla inverter acts roughly like the Delta, it appears that only one string is connected, as others have said.

 

[CrazyRabbit]

the should use all the mppt channels, as long as the voltage requirements can be met.

 

[wwu123]

The pros use thermal imaging to spot bad cells or panels, I guess the idea is that non-producing cells run hotter than working ones. Those cameras are expensive, but a cheap $10-20 IR gun might show a clear difference between 12 entire panels are out vs 12 that are working. The consumer guns typically have like a 12:1 spot ratio though, meaning if you're 12 ft away, it's measuring temp of a 1 ft circle where it's pointed, so you might have to be up on the roof to isolate each panel temp.

Not gonna solve the problem though, just suggesting another way one might be able to get some confirmation cheaply. I used one recently to verify for my (baking) wife that our aging oven was 25 deg lower than what the display was set to.