[RE-wrenches] Megger for array testing

[Matt Lafferty]

Hi Drake,
 
Disclaimer #1: i do not recommend megger-testing an array (or module or
string) unless you have an approved procedure from the module manufacturer
for the installed configuration. 
Disclaimer #2: i am not the foremost authority on the list for this topic...
Mr. Brooks? (Note: Bill's post just came in, but i want to share this
anyway)
 
i have, however, conducted these tests, as part of the commissioning
requirements, on about a dozen systems which were installed under R&D
grants. Back in the day, pretty much all systems bigger than 4 or 5
kilowatts were required to be megger tested, if they were "on the radar" in
any way. This was a function of the fact that public dollars were being
spent in the HOPE that PV could some day become a mainstream technology. And
the fact that PV modules were so expensive. The combination of these factors
meant that, in any meaningful sense, only projects that received subsidies
were built, and grant requirements were in play. One of the conditions was
that these tests be performed. This is a logical requirement in many ways. 
 
Each system, that i've done this on, was different from the others. In each
case, the procedure and analysis methodology was different. Sufficiently
different, that i learned early on to NOT MAKE UP MY OWN PROCEDURE.
Manufacturer procedures varied in lead placement, array sections to be
tested, surface wetting techniques, test voltages, temperature compensation,
and Pass/Fail values. i've made up lots of procedures over the years, but
i'm not touching this one. I don't understand it well enough.
 
The basic theme and theory was the same, pretty much across the board. But
the actual methods... The steps... They were different. The very limited
information i was able to extract from manufacturers' engineers varied. Some
warned about reverse-biasing diodes and others didn't think it was a
concern. Some insisted that test-lead arrangement was super-critical and
others didn't. Some wanted to test at voltages lower than the system voltage
and others said 1kV was fine on a 500V circuit. Some wanted to connect one
pole to ground. Most didn't.
 
Applying test voltages to, and measuring the resistance of, an installed,
illuminated source circuit or array, is very different than applying test
voltages to, and measuring resistance of, a single, unilluminated and
shorted module. Myriad additonal factors and considerations must be
accounted for. 
 
My observation is that at least half of the manufacturers' engineers were
only guessing, and crossing their fingers that everything would be fine
after performing megger testing on a source circuit or array. Another
observation is that at least half of these engineers had never contemplated
megger-testing an array as a completed system... Only as a single module
within the hi-pot test regime for manufacturing and listing purposes. As a
group, they all were resistant to provide a documented procedure and
acceptable test values. None, as in Z-E-R-O, were able to predict actual
test results with any reasonable degree of accuracy. As in, off by megohms
in many cases. Even though the subjects under test were designed, and
largely manufactured, by their companies. Even though these guys were
responsible for the actual system design, in most cases. Even though my
experience leads me to believe these guys are at least half full of crap, i
do believe this test is valid and has merit. i believe that all arrays
should be tested for their dielectric resistance during the commissioning
process, in fact.
 
If we are going to megger our field-installed conductors... And we
absolutely should be... It's a simple step. In order to do this, as an
industry, we just need a better understanding of the characteristics and
test procedures. My requests that manufacturers publish a procedure for each
module they make, with acceptable test results in common circuit
configurations, have resulted in squat. Going forward, i hope somebody
forces them to do it, 'cause they ain't likely to bother otherwise.
 
Under controlled conditions, such as those found during module
manufacturing, predicting a test result within a reasonable degree of
accuracy has a much higher chance for success than a prediction under
uncontrolled conditions in a multi-subject combination. I get that.
Nevertheless, we are deploying these products into an installed
configuration. There is shipping to the distributor. Shipping to the
installer. Shipping to the jobsite. Unpacking and handling. Getting it to
the roof. Bolting it down. Wiring it up. All before it actually gets to
start its productive life.
 
It is reasonable to be able to expect and require that these tests be
performed during the commissioning process. After all, when would you, as an
installer carrying the warranty, rather find out if something has a higher
chance for failure during its lifetime? At the time of installation when you
can still get your hands around the neck of somebody or five years later
when the sales-slug you know works somewhere else?
 
The fundamental purpose of megger testing an array is to measure the
dielectric resistance between conductive circuit elements (positive and
negative) and conductive non-circuit elements (module frames & racking). The
higher the resistance (megohm) value, the greater the dielectric resistance.
Busbars and cables and switches and receptacles... They all need dielectric
resistance in order to work. The entire theory of useable electricity
depends on dielectric resistance. Without dielectric resistance, you have a
short circuit.
 
Each model of module is a little different in its construction. As a
consequence, the absolute values for each are different. This may or may not
mean there is a different minimum acceptable value for a given class of
module. Resistance is resistance, after all.
 
One key function of any test is to compare the results of the test to
expected results... More accurately, to compare the test results to an
acceptable threshold value. In our case, absent some other authority, that
value has to come from the manufacturer. Furthermore, this threshold value,
which i'll refer to as the minimum acceptable resistance, changes downward
as the surface area of the array being tested increases. This phonomena is
consistent with wires, too.... The larger and longer the conductor, the
lower the theoretical minimum acceptable insulation resistance is.
 
Test lead placement is a fundamental piece of the puzzle. But it's only part
of the puzzle. You need to know several things that are specific to your
application... For example, the procedure for testing a single module may
differ greatly from testing a string of 10 in series. Testing two paralell
strings of 10 may have a different procedure... In any case, the expected
values for each setup will be different. Based on my experience, the key
items you need to know, from the manufacturer and for your specific
application, are:

1.	

	Any pre-megger tests to be performed
2.	

	Test lead placement (May be different for pos to ground than neg to
ground)
3.	

	Circuit conductor configuration
4.	

	Test voltage
5.	

	Time on test
6.	

	Acceptable irradiance during test
7.	

	Number of times to repeat the test
8.	Wet or Dry 

9.	

	If Wet, what solution to use
10.	

	Temperature compensation
11.	

	Irradiance compensation
12.	

	Minimum acceptable resistance for each test (May be different for
pos to ground than neg to ground)

If you are testing at the inverter, you will need to account for the
resistance of the insulation of your wiring. You will also need to be sure
that each module frame has a good connection to the grounding conductor you
are using as a reference.
 
In my experience, i have seen differences in which test lead connects to
which circuit conductor. Some had different Pass/Fail values for positive
and negative and some did not. Some included temperature compensation and
others didn't. Some were done wet and others done dry. Test times varied.
Some said a single test was fine and others said three passing tests were
required before the system passed. And they all had different minimum
acceptable resistance values. Most systems passed the tests. Some didn't on
the first try and required troubleshooting and module replacements. One
never did pass and had to be removed. One thing i have NEVER SEEN is two
identical test results on the same circuit.
 
As you push for a procedure, expect them to request that you sign an NDA
before they give it to you. You can decide how you choose to respond to
that. For what it's worth, i'm somewhat confident that they will work with
you on this. In large part, this is because it's Sanyo. They are a grown-up
company who really believes in quality products and customer relationships.
Can't say that about many others. If you run into resistance, get them on
the phone, cough, then say "Solar Ark".
 
i certainly wish you the best of experiences with this and hope you find the
problem(s) one way or the other.
 
Pray for Public Array Megger Procedures!
 
Solar Janitor
 
  _____  

From: re-wrenches-bounces at lists.re-wrenches.org
[mailto:re-wrenches-bounces at lists.re-wrenches.org] On Behalf Of Drake
Sent: Wednesday, November 17, 2010 9:24 AM
To: RE-wrenches
Subject: Re: [RE-wrenches] Megger for array testing


Thinking more about the Megger test of an array, is seems impossible,
despite that fact I've been told it could be done.  If I put the tester
positive to the 500 V DC negative, that would put two 500 V sources in
series for 1000 VDC.  If I put the positive to positive, the resulting
voltage to ground would be zero.  What is the trick?



At 10:36 AM 11/17/2010, you wrote:


Hello Wrenches,

Looking through the archives on using a Megger  for testing modules, it
appears that it is best to keep voltages to 500VDC.  I read all that I could
find. 

Talking to a Sanyo representative, I was told that keeping test voltage
under 600 would not void the warranty or be an issue with UL.  He said they
have tested Sanyo modules up to 1000 VDC with no problems.

As a final test on a system, I want to Megger test the array from the
terminations at the inverter.  The approach I'd plan to use is: 

*	Clamp the Fluke 1587 negative to the enclosuer ground. 

*	Put the positive test lead to the negative conductor from the array
and test 

*	Put the positive test lead to the positive conductor from the array
and test 

All tests would be done at 500 V.  The modules are illuminated, but there is
no significant reading from either positive or negative terminal to ground. 

*	The conductors would be removed from their terminals at the inverter
and be in free air. 

*	The conductors would be continuous through the string of 10 Sanyo
210 modules. 

My questions are: 

*	Does anyone see a problem in this approach? 

*	Since the array often has 500 VDC open circuit readings and shows no
indication of a fault whatever, would this test accomplish anything more
than the standard voltage to ground tests I've already done? 

*	Is there any danger to diodes or other PV components? 

Thanks in advance.

Drake  


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.re-wrenches.org/pipermail/re-wrenches-re-wrenches.org/attachments/20101117/17ae3fcf/attachment-0003.html>