Positive side of string reading half the negative side

[SouthFaceRenewables]

System specs: 27 Hanwa Q cel 400W modules, 4 total strings on 3 different roof faces, Tesla 1st generation MCI, (2) Powerwall + batteries with backup gateway.

Scenario: Modules were removed for a roof replacement. Use clamp on amp meter around noon on a sunny Colorado day, temperature outside was 60F. Got amp readings of ~10amps (MPPT V posted on powerwall + is 10.77A). Told client everything was up and running. While we are cleaning up client shows me their tesla app and states that production is only about 60% when compared to a sunny day in February 1 year ago (10kwh last year, 6kwh today). As I'm in the junction box on the roof checking connections to the DIN rail blocks, I decide to measure voltage from + to ground, and - to ground. 2 of the strings have + voltage readings of only ~ half of their - counterpart. I have taken a few courses through SEI, and never had this topic come up. I'm at a loss with regards to how to troubleshoot further. Would greatly appreciate someone sharing their knowledge with me.

 

[tortuga]

System specs: 27 Hanwa Q cel 400W modules, 4 total strings .

I am not a PV expert but you can use some basic ohms law to troubleshoot any DC series circuit.

Those panels appear to be 400Wmpp and if I am not mistaken 37.13Vmpp
therefore
using ohms law P/E = I
we get
400W / 37.13V = 10.77A

Since they are in strings the voltage and wattage will increase proportionally with the number in the string
but the amperage will stay the same with any number of panels (10.77A)

Scenario: Modules were removed for a roof replacement. Use clamp on amp meter around noon on a sunny Colorado day, temperature outside was 60F. Got amp readings of ~10amps (MPPT V posted on powerwall + is 10.77A).

10.77A means at least one panel is working in the string.

client shows me their tesla app

I hate apps

As I'm in the junction box on the roof checking connections to the DIN rail blocks, I decide to measure voltage from + to ground, and - to ground.

Whats do you mean by ground ?

2 of the strings have + voltage readings of only ~ half of their - counterpart.

How many panels are in each sting?
I would expect DC string voltage to be the voltage of each DC source X N in the string.
So if I have ten 37.13V battery's all in a string I would expect to get 37.13 X 10 = 371.3V

Note you wont see those numbers in the real world they are just from the data sheet.

 

[jaggedben]

I wanna say there's a ground fault but the details are lacking. What were the actual voltages? Where the measurements taken with the inverter operating? Did the inverter continue to operate?

Was the lower kWh production not because you just turned the system back on, so the client was comparing a partial day's production to a full day?

With two paralleled strings a ground fault will make both strings have the same voltages to ground, but with three or more it gets funky.

 

[SouthFaceRenewables]

I wanna say there's a ground fault but the details are lacking. What were the actual voltages? Where the measurements taken with the inverter operating? Did the inverter continue to operate?

Was the lower kWh production not because you just turned the system back on, so the client was comparing a partial day's production to a full day?

With two paralleled strings a ground fault will make both strings have the same voltages to ground, but with three or more it gets funky.

Thank you for taking time to assist a fellow installer. Answers to your questions are:
1) Voltage readings for string A which were taken at din rail terminal blocks on the roof (5 panel series string, 45watt panels, expected voltage 225): Positive to ground-210v, Negative to ground-100v (This is what is stumping me. I would expect negative to ground to read the same 210v as the positive to ground reading)
2) Measurements were taking with the inverter operating, and it continued to operate.
3) the production being compared was a snapshot of real time production. Picture a bar graph where the Y axis is KWH produced and the x axis is time in 30 minute or 1 hour intervals.
4) No strings on this project are paralleled.

I’m headed to the site tomorrow to have Tesla remote into the system. I will let you know what they find.

 

[ggunn]

Thank you for taking time to assist a fellow installer. Answers to your questions are:
1) Voltage readings for string A which were taken at din rail terminal blocks on the roof (5 panel series string, 45watt panels, expected voltage 225): Positive to ground-210v, Negative to ground-100v (This is what is stumping me. I would expect negative to ground to read the same 210v as the positive to ground reading)
2) Measurements were taking with the inverter operating, and it continued to operate.
3) the production being compared was a snapshot of real time production. Picture a bar graph where the Y axis is KWH produced and the x axis is time in 30 minute or 1 hour intervals.
4) No strings on this project are paralleled.

I’m headed to the site tomorrow to have Tesla remote into the system. I will let you know what they find.

Turn off the inverter and disconnect all the strings from the inverter; measure + to ground and - to ground on all the string home runs. The voltages on intact strings should all appear to be undefined because there should be no path to ground on any of them. If there is a ground fault on a string the + to ground and - to ground voltages will sum to the string Voc and the ratio of the voltages will show you where the fault is.

 

[jaggedben]

Okay, I'm going to talk about a ground fault because I think you need to understand that better but now I think you probably just have a string or two unplugged.

Thank you for taking time to assist a fellow installer. Answers to your questions are:
1) Voltage readings for string A which were taken at din rail terminal blocks on the roof (5 panel series string, 45watt volt panels, expected voltage 225): Positive to ground-210v, Negative to ground-100v (This is what is stumping me. I would expect negative to ground to read the same 210v as the positive to ground reading)

In a true ground fault, if the string is isolated from other strings, the voltage to ground will be proportional to the number of panels between the conductor your measuring and the ground fault location. So for example, if you have a 5 panel string and the fault is three panels from the positive home run, then positive to ground will be three panels of voltage and negative to ground will be negative 2 panels.
Also I don't know how your Tesla MCIs work so I can't say they won't necessarily mess with your voltage readings when you turn the system off.

That's probably all irrelevant though, because you have a transformerless inverter and you were measuring the DC voltage to ground while operating. Also your strings weren't isolated (see below). That means you're referencing the DC to the grounded AC neutral, through the inverter. It's kinda anyone's guess what you'll get on an inverter with multiple strings of different sizes connected through multiple MPPTs.

2) Measurements were taking with the inverter operating, and it continued to operate.

That suggests the problem is not a ground fault. Hopefully the inverter would stop operating on a ground fault. Well, unless your EGC isn't properly installed between your array j-box and inverter.

...
4) No strings on this project are paralleled.

...

When the inverter is operating, the switching electronics are connecting all string positives and negatives to the same busses inside the inverter. So they're going to be effectively paralleled. (In fact in my experience the negatives are usually directly connected to each other inside the inverter, so only the positives are isolated by the MPPT electronics when the inverter is off. But this may depend on inverter design.) So you can't assume that one string isn't affecting others in this scenario. That's why ggunn's troubleshooting suggestion in the last post included disconnecting all strings from the inverter.

This also means that if you had a string that wasn't actually connected, you could still possibly get voltage on its terminals in the j-box, from the other strings.

If you turn off the inverter, disconnect the wires from the inverter in your j-box so that all the strings are isolated from each other, and measure positive to negative voltage on each string, you can confirm if any are not connected all the way through. And you can also measure to EGC from the array to see if you have any ground faults, and which string.

 

[ggunn]

Turn off the inverter and disconnect all the strings from the inverter...

This should be obvious, but disconnect all the strings from each other as well.

 

[SouthFaceRenewables]

Thank you both! I believe you have given me a strong direction to take my troubleshooting when I return to the site tomorrow. I will let you both know what the resolution is once I identify it.