SMA Ground Fault troubleshooting advice. [RE-wrenches]

Wed Nov 21 12:18:23 PST 2007

Dave/Wrenches:

Locating ground faults in the array is fairly straight forward albeit a
little dangerous.
A DVM and some spare fuses (just in case) is all you really need.
Be very careful when you are doing all of this because you will be
working with live wires hanging in space and PV negative is no longer
necessarily at ground potential.  If the equipment grounding on the
array is not up to snuff it is also possible to see hazardous voltages
on the module frames or structure so measure everything and be really
careful.  Treat everything as hot and potentially lethal until your
meter says otherwise. Remember each time you disconnect something the
whole situation can change so measure again to be sure.  Also, depending
where the ground fault is and the order in which you pull the wires out
of the inverter (or disconnect) it is possible to see double the array
Voc.  This can be a very bad thing since the spacing on all components
was designed for 600 Vdc max and you could be at double that.  A 960 Vdc
arc at  5 to 8 Amps is pretty ugly.

Start by opening the DC disconnect and removing the GFDI fuse in the
inverter.
Measure the array open circuit voltage form the PV plus in the dc
disconnect to the PV negative in the inverter.
Lets say it is 480 Vdc.
Now measure PV positive to ground and then the PV negative to ground.
An array without a ground fault should give you roughly equal values of
around 240 Vdc but the measurement should decay down towards zero.  What
you are seeing is a discharge of the capacitive coupling between the
array and ground.

In an array with a ground fault you will measure a fixed voltage to
ground on both the PV positive and PV negative.
The sum of the PV positive to ground and PV negative to ground
measurements should equal the open circuit voltage.
The difference in this case is that the voltage will be fixed and will
not decay towards zero.
The voltages you measure actually tell you where to start looking for
the ground fault.

For example: with 12 modules in series and an open circuit voltage of
480, each module is equal to 40 Vdc.
If you measure PV positive to ground and get 0 volts then you should
see a PV negative to ground voltage of - 480 Vdc.  This tell you the
ground fault is in the PV positive output conductor.
Alternatively, if you were to see PV positive to ground at 400 Vdc and
PV negative to ground at -80 Vdc this tells you the ground fault is
between the 2nd and 3rd modules "up" from the PV negative.

To figure out which string the ground fault is in, isolate each string
(remove the wires from the disconnect/inverter) and repeat the PV
positive to ground and PV negative to ground measurements.  On one
string the measurement will float around and the other will give you a
fixed measurement to ground.  The fault is in the string that has the
fixed measurement to ground.  My suggestion is to pull both positive and
negative wires for string 1 and measure.  If you are lucky you picked
the right string the first time out and you don't need to pull the
second string at all. 
It is unlikely but also possible to see a ground fault in the "middle"
of a module.  In this case you might see 380 Vdc from PV positive to
ground and -100 Vdc from negative to ground.  This would indicate a
fault in the middle of the third module "up" from the PV negative. 
Normally this means the module itself has a ground fault due to a nick
in the backskin, fault in the edge seal, etc. You can verify this by
isolating (unplugging) the suspect module then measuring PV positive to
ground and PV negative to ground for that isolated module. 

Lastly, In some very rare case the ground fault will actually be in the
string combiner, dc disconnect, or (heaven forbid) in the inverter
itself. You can verify this if the array voltage measurements show a
ground fault when connected to the combiner/disconnect/inverter and then
both strings float when measured in hanging in free space.  While
unlikely, if you find this situation reconnect the PV back up then
disconnect one thing at a time and measuring relative to PV positive and
PV negative to ground until you find where the fault is.

Best Regards,

John Berdner

>>> ipl at sover.net 11/21/2007 07:40:51 AM >>>

Wrenches,

I have a grid tied system with a SB3800U inverter being fed by 2
strings of
(12) Evergreen 180's (4.32kW). UniRack rail mounted on garage roof.
Good
installation all wiring neat and well supported. Only DC bond to ground
is
in the inverter etc. And no external monitoring equipment.

The system has been operating very well (over 5,000 kWh over 2 months
in
Northern Vermont), this is the first problem.

We are in the middle of a snowy spell and the array is covered in snow,
but
I should be able to check out the array within a few days. According to
the
error message on the Sunny Boy the GFDI 1 amp fuse is open. The array
has
made it through lots of rain, ice and snow over the past year.

My suspicion is in the MC connector(s) themselves. We had a bad batch
of MC
cables purchased from a large distributor 12 to 18 months ago. The MC
connector would pull off the wire when tested. Of course not all of
them
where tested before they got into the field, so one could be on this
roof.

Obviously I will check connections in the combiner, the dc disconnect
etc,
could be a moisture issue. Of course it could be a squirel I guess.

Any advice on troubleshooting this ground fault would be appreciated.
Safety
wise do we need to cover the array with a dark tarp?

Dave

--
[Non-text portions of this message have been removed]

- - - -
Hosted by Home Power magazine

To send a message: RE-wrenches at topica.com

Archive of previous messages: http://lists.topica.com/lists/RE-wrenches/read

List rules & how to change your email address: www.mrsharkey.com/wrenches/etiquette.php

Check out participant bios: www.mrsharkey.com/wrenches/

Moderator: michael.welch at homepower.com
--^----------------------------------------------------------------
This email was sent to: michael.welch at re-wrenches.org

EASY UNSUBSCRIBE click here: http://topica.com/u/?bz8Qcs.bz9JC9.bWljaGFl
Or send an email to: RE-wrenches-unsubscribe at topica.com

For Topica's complete suite of email marketing solutions visit:
http://www.topica.com/?p=TEXFOOTER
--^----------------------------------------------------------------