Tester for "Insulation resistance low"

[Zee]

It is the rainy season.
I am getting service calls for resi. systems with "low insulation resistance" error codes / ground faults in the PV array.

Does anyone here have a favorite or recommended tester for such jobs?
I mean the resistance tester.

I have run the DC V tests to no avail.
(measured + to gr, and - to gr in order to locate the faulty wiring/panel)

TL;DR:
Systems installed 2006 and 2013.
-One has bubbling of the back sheets on some, but not all, problem panels. Sharp brand.
No other damage or wiring problems.
Intermittent faults..... they clear within days of rain.

-On the other, I have picked up panels and cannot find any visible damage. (no crushed, nicked, chewed wires.... no gaps in MC conn.s.....no damage to panels).
Measured no voltage from panel + or - leads to panel frame....so it seems no internal panel shorts or faults.
These faults persist for weeks.

 

[Zee]

Additional Q's:

Why don't the partial DC voltages add up to total string V?
E.g. in a string of 8 (no MLPE's)
+ to - 331 VDC
- to ground 158 VDC
+ to ground 140 VDC
(....adds up to 298 VDC)

also:
+ to - 332 VDC
-to g 203 VDC
+ to g 122 VDC
(....adds up to 325 VDC)


Also, what exactly goes wrong in these "insulation resistance low" cases? In particular when no visible damage can be found.
I read the the insulation may deteriorate over time?
Does it break up/crack/microcrack and allow moisture to penetrate?
Does this mean it actually absorbs water?

Does it change chemically and become less resistive/more conductive?

Or is it faults with connectors and nor conductor insulation?

What is the fix? Replace all panels?

 

[petersonra]

This has nothing to do with it being a pv system. It does have to do with intermittent insulation faults. They are a real pain to find some times. The ir tester can only test it right this second and if right this second there is no fault it will not find anything.

Sometimes you have to resort to changing out the most obvious wires one by one until the problem goes away. This is maddening to the end user but if you can't find the fault, it is the only way.

The other issue is that sometimes the apparatus that is reporting the fault is itself faulty and there is nothing wrong with the conductor insulation at all. For instance, since this is happening when it is wet out, it might be you are getting condensation on a board where the testing originates.

 

[solarken]

This is happening on two different residential systems, correct? Not more than that? What inverter model? What are the string arrangement/sizes of the two systems, i.e. two strings of 10 modules? And how often does it happen? Every time is rains? You might want to try disconnecting all but one string and letting it run for a few days thru a rainy period that you are confident would cause the issue to show, then connect a different single string, and repeat, etc, until you can isolate it to one string. From there you may be able to find it, or you may need to jumper out half the string (making sure to stay above min startup voltage and divide and conquer until you identify the module or wiring that is causing it. Significant work, but maybe not as much as removing, inspecting, and replacing every module and every wire in the array. I don't have experience with an Insulation Tester but I too am interested in the best way to tshoot this problem because it will likely happen for some of the systems I have installed as they age.

As far as the voltages to ground not adding to the voltage across the + and - leads, I don't think that suggests anything is wrong for ungrounded or functionally grounded arrays.

 

[wwhitney]

Why don't the partial DC voltages add up to total string V?
E.g. in a string of 8 (no MLPE's)
+ to - 331 VDC
- to ground 158 VDC
+ to ground 140 VDC
(....adds up to 298 VDC)

Did you take those measurements one at a time, or did you hook up 3 meters to get simultaneous measurements?

If one at a time, the simplest explanation would be that the presence of the meter taking measurements to ground slightly shifted the voltages to ground. Which I think is plausible if + and - are basically floating to ground and only coupled capacitively.

Cheers, Wayne

 

[Zee]

This is happening on two different residential systems, correct? Not more than that? What inverter model? What are the string arrangement/sizes of the two systems, i.e. two strings of 10 modules? And how often does it happen? Every time is rains? You might want to try disconnecting all but one string and letting it run for a few days thru a rainy period that you are confident would cause the issue to show, then connect a different single string, and repeat, etc, until you can isolate it to one string. From there you may be able to find it, or you may need to jumper out half the string (making sure to stay above min startup voltage and divide and conquer until you identify the module or wiring that is causing it. Significant work, but maybe not as much as removing, inspecting, and replacing every module and every wire in the array. I don't have experience with an Insulation Tester but I too am interested in the best way to tshoot this problem because it will likely happen for some of the systems I have installed as they age.

As far as the voltages to ground not adding to the voltage across the + and - leads, I don't think that suggests anything is wrong for ungrounded or functionally grounded arrays.

Yes 2 systems.
#1
installed 2006
FRONIUS 3800w PRIMO (transformerless, less than 5 years old)
3 strings of 8
24 @ SHARP NT-185U1 panels
4.44 kW


#2
installed 2013
SMA 4000 TL US
2 strings of 9
18 @ Canadian Solar CS6P-250M , 250 Watt modules
4.5 kW

FWIW, "Ground-fault monitoring for SMA SB 4000TL-21: Insulation monitoring: Riso > 500 kΩ"
--------------------
On both systems, faults shuts down INV later in rainy season, every time it rains heavy.
Then it may last for weeks or month.
Major loss.
----------------------
On system #1, I voltage tested string ends to ground......
and then panel counted per panel V to fault location....
and then shuffled no less than 7 panels around on roof physically.
I tried to get all bad panels into one string and leave 2 good strings.
But to no avail. INV triggered each time.
....And...... now getting new error codes. ("low DC V")
===> Will replace all panels. I am fine with that, as is customer. They had 2 decades. And I spent many hours.
So my questions on #1 are out of curiosity only!

However, system #2 bothers me for 2 reasons:
- It is just past 10 yrs old. Not much.
- - I installed it.
I will return and remove what seems to be 1 faulty panel in each string.... and see if INV starts up.
I have my doubts.

 

[Zee]

As far as the voltages to ground not adding to the voltage across the + and - leads, I don't think that suggests anything is wrong for ungrounded or functionally grounded arrays.

I am no engineer.
However, when there is a fault on a floating string, the string IS grounded at that fault location.
All testing protocols for these systems call for positive to ground and negative to ground VDC testing and then dividing by panel Voc and panel counting.
Thanks for thoughts.

 

[BillK-AZ]

I use a 1000VDC insulation tester (Ideal 61-795 purchased 2008). I use this on systems without MLPE, works well to locate modules or pinched wires on systems where SMA inverters report insulation test failure. Test by string, then by half a string, etc. to locate module. Some failed modules have no visible faults.

 

[Zee]

Did you take those measurements one at a time, or did you hook up 3 meters to get simultaneous measurements?

If one at a time, the simplest explanation would be that the presence of the meter taking measurements to ground slightly shifted the voltages to ground. Which I think is plausible if + and - are basically floating to ground and only coupled capacitively.

Cheers, Wayne

One VDC measurement at a time.

Since PV is a power source, shouldn't the Dc string voltage be constantly "kept up"?
How would the voltmeter deplete the VDC?
Thanks for reply.

 

[Zee]

I use a 1000VDC insulation tester (Ideal 61-795 purchased 2008). I use this on systems without MLPE, works well to locate modules or pinched wires on systems where SMA inverters report insulation test failure. Test by string, then by half a string, etc. to locate module. Some failed modules have no visible faults.

Nice. Cost? Do you then just junk the panel even if no visible fault?
Have you ever used the voltmeter method I mentioned? How do you like vs your insulation tester?

 

[wwhitney]

One VDC measurement at a time.

Since PV is a power source, shouldn't the Dc string voltage be constantly "kept up"?
How would the voltmeter deplete the VDC?

I'm not suggesting that the voltmeter would affect the voltage between DC- and DC+. I'm suggesting that if those voltages are floating relative to ground, then the voltage relationship to ground is just due to incidental capacitive coupling or whatever, and may be easily changed. In particular by connecting the voltmeter itself, without changing the voltage between DC- and DC+.

Cheers, Wayne

 

[ggunn]

I am no engineer.
However, when there is a fault on a floating string, the string IS grounded at that fault location.
All testing protocols for these systems call for positive to ground and negative to ground VDC testing and then dividing by panel Voc and panel counting.
Thanks for thoughts.

Yes, the fault to ground is the center of a voltage divider network, and comparing V+ to ground with V- to ground or either of them to Voc of an unfaulted string (assuming the strings are all of the same length) will tell you between which two modules in the string the fault has occurred.

 

[jaggedben]

Additional Q's:

Why don't the partial DC voltages add up to total string V?
E.g. in a string of 8 (no MLPE's)
+ to - 331 VDC
- to ground 158 VDC
+ to ground 140 VDC
(....adds up to 298 VDC)

also:
+ to - 332 VDC
-to g 203 VDC
+ to g 122 VDC
(....adds up to 325 VDC)


Also, what exactly goes wrong in these "insulation resistance low" cases? In particular when no visible damage can be found.
I read the the insulation may deteriorate over time?
Does it break up/crack/microcrack and allow moisture to penetrate?
Does this mean it actually absorbs water?

Does it change chemically and become less resistive/more conductive?

Or is it faults with connectors and nor conductor insulation?

What is the fix? Replace all panels?

It sounds to me like module glass or backsheet failure that is allowing in moisture that is compromising isolation from ground. I've seen this, and while I forget the details it definitely confused the heck out of me, too. (I had two failing modules next to each other, which made it even harder to narrow down.) If there is water getting into the module creating a conductive enough path to the frame and racking then essentially a module could be grounded at/near both its positive and negative ends. So when you measure the string positive and negative to ground you don't measure the affected panel either time. Or some portion of it.

Sorry man, I know it sucks.

 

[BillK-AZ]

Nice. Cost? Do you then just junk the panel even if no visible fault?
Have you ever used the voltmeter method I mentioned? How do you like vs your insulation tester?

These testers are still sold, $460-525. Good instrument, bought when the norm was 600vdc systems. Only recently have I found failed modules, well out of warranty. If a unique failure I give the modules to the ASU research facility. Need to replace with best available module.

The DVM method works OK for lower resistance faults to ground, The insulation tester finds faults under 4000 megaohms, well beyond a DVM. The insulation tester does not work very well for locating the fault by ratios. You get different values based on polarity to ground, and for repeated tests due to charge buildup. On two jobs we found the fault was pinched PV wire, DVM showed no faults. On one occasion we completed a job started by another contractor (never actually looked at the roof, high parapets) and a pallet of modules was stored for months outside, backskin up, and found several with standing water. The wet modules got mixed in with the others, initial start-up of the inverter faulted on resistance failure and two modules were replaced. After two years of operation another module caused an insulation failure even though it passed earlier testing.

 

[Zee]

Interesting. Thank you gentlemen.