Arcing Occurred During Sunclix Disconnection

sun_day

Hi all! I am totally stunned at this when it happened.

Arcing occurred when we are disconnection the sunclix connectors for our DC measurements. The configuration is...

6 strings of 12 panels total Voc 778V per string. These are connected to a STP20000TLEE-10. Note that the inverter is switched OFF.

The solar panels cables are directly connected to an inline fuse (Product is this: https://www.phoenixcontact.com/online/portal/pi?1dmy&urile=wcm:path:/pien/web/main/products/subcategory_pages/photovoltaic_plug-in_connectors_p-20-06/417d9243-32cf-4277-8af3-bdc0851772d5/417d9243-32cf-4277-8af3-bdc0851772d5). So the DC cables from panels are connected to this inline fuse, which connects to the inverter.

We disconnect the suclix connectors BEFORE the inline fuse.

When we disconnect string 4, 5, 6, no arcing occurred. However, when we disconnect string 2+ and 1+, an arc occurred and the normal flash and smell of burnt material can be seen/smell. The remaining strings were left untouched due to safety reasons.

Anyone have any clue why this is happening? There is no load drawing the current since the inverter is switched OFF. This is the first time we see it happen, at the first time, it is our first time using the inline fuses.

Please enlighten me if you have any idea!

Much appreciation!

BB.

The only way you can get arcs is if there is current flow. The safest way to measure current flow is to get a DC Current Clamp Meter. This one from Sears is only ~$60 and is "good enough" for our needs (debugging):

http://www.sears.com/craftsman-digital-clamp-on-ammeter/p-03482369000P

DC arcs are very impressive when compared to AC arcs of similar voltage and current:

https://www.youtube.com/watch?v=Zez2r1RPpWY

Something sounds very strange--Was the system working correctly before you started working on the system? Something sounds miss-wired or an insulation slice on a sharp corner/conduit end????

-Bill

Estragon

My inverter still draws a small amount of current even when "off" for status lights and remote display. I flip the breakers to kill power completely.

mvas

Was String #1 pushing amps "backwards" through String #2 or
was String #2 pushing amps "backwards" through String #1 ?

Does one string have a lower voltage than the other string?

sun_day

Hi all,

Thanks for replying back!

@mvas - We did string measurements before using Seaward on the strings before the inline fuses are installed. The strings Voc/Isc/Riso are all within expectation. I do not think string 1 amps is backwards through string 2 because if that is really the case, then something is wrong at the inverter side since these are individual connected to the inverter.

@Estragon - I didn't know that. I consulted another friend of mine before and he told me that although inverter is switched off, it is actually still drawing a very very small current due to how the inverter is being designed. Is this true?

@BB. - We did not measure the amps using a clamp meter prior to disconnection since we have never have to worry about an arc occurring. The system was never turn on prior to the disconnection. After this happened, we connected back all the disconnected Sunclix into the inline fuses (no arc occurred during this) and turned ON the inverter. The inverter operate as normal with generation of power to the AC. Inverter did not give any errors during the operation.

I am still at a lost here. Indeed something is strange and when I consulted my engineering team, they said that such arc shouldn't happened unless it is really high voltage. I can't be the one imaginating this since we had 3 people witnessed this.

Any more ideas please?

Photowhit

There's plenty of voltage, but the current found a home. I take it, it was acring across the connections and not to ground.

sun_day

Photowhit said:

There's plenty of voltage, but the current found a home. I take it, it was acring across the connections and not to ground.

Hi Photowhit,

Thanks for the response. Yes the arcing occurred is a series arc not to ground.

Will appreciate if theres more input from everyone! Let me know if you require addition details.

mvas

I still stay, there is something different between String #1 and String #2. With parallel connected strings and the Inverter OFF, if one string has a higher voltage than the other string, then current will flow backwards through the lower voltage string. I am not saying that either string is connected backwards just a difference in voltage. Shading on one or more panels? 

Are those connectors rated to be opened when under load @ 778 Volt & 4 amps?
You had amps flowing and you used the connector to break the circuit.

Here is a picture of a DC arc fire when a so-called Touch-Safe Fuse was opened with the Inverter OFF ...
https://uploads.tapatalk-cdn.com/20170131/f4c44e5318760eca5a3b1b0e28f390c0.jpg
They assumed that since the Inverter was OFF, that there was no load on the PV Panels, but amps were still flowing!

sun_day

Hi mvas,

Thanks!

Just to confirm (apologise since I am not a total expert when it comes to this) what you mean is that there is a voltage potential difference between this two strings, and since they are connected in the same MPPT of the inverter henceforth in parallel, the inverter might also have this potential differencd across its internal circuit which can cause the arc?

I was thinking if it is due to the inline fuse that we used since this is the first time we are using such fuses. I have disconnected so many DC connectors from either DC combiner box or inverter and I am sure that at certain point of time there is a difference in string voltage but no arcing occurred.

The connectors are definitely rated for such thing as these are specialise PV connectors. Unless the connectors are faulty for that particular string?

mcgivor

Been thinking about your dilemma after reading posts, there has to be a load in order to get an arc, if you were to connect one leg, say negative and using a volt meter between the 2 disconnected positive leads, do you read a voltage, potential difference? If so, there would be something downstream creating a load. When connected, using a clamp on ammeter you would read a current, if disconnecting in thr reverse order perhaps the last ones disconnected would exhibit the same arc, probing in the dark here.

sun_day

@mcgivor

Thanks for your inputs! Appreciate it! As agreed with you, arcing will be caused when there is a load. The DC cables are directly from the PV array to the inline fuses. Nothing is between them except for MC4 connectors for the home run cables.

To make it more clear to everyone who has been helping me so far, below is the disconnection sequence we took,

String 6+/- > No arcing during disconnection
String 5+/- > No arcing during disconnection
String 4+/- > No arcing during disconnection
String 3+ > Arcing occurred during disconnection
String 3- > Not disconnected
String 1+ > Arcing occurred during disconnection
String 1- > Not disconnected
String 2+/- - Not disconnected

Again note that STP20000TLEE-10 only has one MPPT. Therefore all 6 strings are connected to the same MPPT.

BB.

Unless there is a short circuit because of miss-wiring (i.e., a solar panel string wired backwards--Panels are "giant diodes", wired backwards they become a dead short and can ruin the string from over current or pop series breaker/fuses that should be wired in each string)--Then either you have sliced insulation, floating connection hitting metal, or an unknown load (such as gt inverter taking current).

A DC Current Clamp Meter is going to be the easiest/fastest/safest way to figure out where the current is going.

-Bill

sun_day

@BB.

Thanks! I will definitely use the DC current clamp meter next week when I am back at the installation site.

Something to note is that I mentioned the DC side are all in good condition because we measured the 6 strings using Seaward (which checks Voc, Isc, Riso and polarity). The measurements from Seaward reported fine. This was before the inline fuses were installed and connected.

After the arcing occurred, we connected back everything and actually turned on the inverter. The inverter work normally and generate power. If there is insulation fault somewhere along the DC side, shouldn't the inverter pick up the error?

mcgivor

Think I understand what is happening, if you were to disconnect  only one polarity at a time, negative or positive, but only one polarity at a time, I'll bet there would be no arcing, reason being there would be no reverse current feeding back towards a diode, the bypass  diodes in this case. 

Estragon

Another theory FWIW.

The inverter is drawing some small amount of current, the supply of which is shared between strings. As strings are disconnected, the amount of current increases as the remaining strings supply that current until there is enough current to sustain a small arc. The arc becomes a load (heating air) which increases as the connectors are pulled apart until the load is more than the full pv current available.

With no breakers to flip to make sure current is zero, I would throw a tarp or something over the panels to disconnect (or do it at night).

sun_day

@mcgivor

The strings were disconnected one at a time. By the time we are at string 3, string 4 to 6 (positive and negative) are already disconnected. The remaining strings that are still connected to the inverter were string 1 to 3 (positive and negative). This is when we disconnect string 3 positive, the arc occurred.

@Estragon

I re-read the posts above and realised that your inverter is switched off but the LED light and display are still on as it draw currents. For my case, when the STP20000TLEE-10 is switched off, the LED light and display are essentially switched off too (these are feed when DC is connected to the inverter).

Many thanks for all the inputs guys! Lets keep this going!

Estragon

@sun_day
Just because the lights are off, I wouldn't say for sure there is no load. As BB suggested, you would need to measure to be sure.

Anyway, at nearly 800v, I would be covering panels in any case. I'm paranoid around even 48v though so...

mike95490

Easy check when you power it back up -
 Monitor the inverter output wattage, each string should contribute x amount of power to the output, and since it's 6 parallel strings, each should contribute equally.  Engage one string, after the inverter boots up and is producing power, add another string, and in 10 sec or so, that power should present itself on the output too.  Easy to find the oddball string(s) that way,

And you may have to replace some connectors, the ones that arced, likely fried the low resistance gold plating, and will heat and burn up in a couple days.

mvas

mcgivor said:

Think I understand what is happening, if you were to disconnect  only one polarity at a time, negative or positive, but only one polarity at a time, I'll bet there would be no arcing, reason being there would be no reverse current feeding back towards a diode, the bypass diodes in this case. 

  I disagree with your theory that the Higher Voltage String #1 pushes reverse amps through the by-pass diodes of the the Lower Voltage String #2.  String #1 cannot force any significant level of amps through the bypass diodes of String #2, because all of the by-pass diodes in String #2 are reverse biased from String #1's point-of-view.  Reverse Bias Leakage is very small.  But with no-load, ie Inverter OFF, the Higher Voltage String #1 will push reverse amps through the individual solar cells of the Lower Voltage String #2.

  Forcing reverse amps through the individual solar cells is how "Snow Melt" mode works.

 A Blocking Diode in each string would prevent reverse amps from flowing into any of the lower voltage strings in a multi-parallel PV Panel configuration. What are the odds that every string has the exact same Voc when there is no load?  The larger the difference the larger the amps.  I am surprised that NEC does not require a DC Rated Disconnect on each and every string.  Opening a PV connector with one hand on each connector while under a load at 778 Volts is extremely dangerous.

 Analogy -> Connect a full charged battery in parallel with a fully discharged battery. Huge amps will flow, even with the Inverter disconnected.

Simply measure the amps is ALL wires to discover where the amps are still flowing even with the Inverter OFF.

mcgivor

mvas said:

mcgivor said:

Think I understand what is happening, if you were to disconnect  only one polarity at a time, negative or positive, but only one polarity at a time, I'll bet there would be no arcing, reason being there would be no reverse current feeding back towards a diode, the bypass diodes in this case. 

  I disagree with your theory that the Higher Voltage String #1 pushes reverse amps through the by-pass diodes of the the Lower Voltage String #2.  String #1 cannot force any significant level of amps through the bypass diodes of String #2, because all of the by-pass diodes in String #2 are reverse biased from String #1's point-of-view.  Reverse Bias Leakage is very small.  But with no-load, ie Inverter OFF, the Higher Voltage String #1 will push reverse amps through the individual solar cells of the Lower Voltage String #2.

  Forcing reverse amps through the individual solar cells is how "Snow Melt" mode works.

 A Blocking Diode in each string would prevent reverse amps from flowing into any of the lower voltage strings in a multi-parallel PV Panel configuration. What are the odds that every string has the exact same Voc when there is no load?  The larger the difference the larger the amps.  I am surprised that NEC does not require a DC Rated Disconnect on each and every string.  Opening a PV connector with one hand on each connector while under a load at 778 Volts is extremely dangerous.

 Analogy -> Connect a full charged battery in parallel with a fully discharged battery. Huge amps will flow, even with the Inverter disconnected.

Simply measure the amps is ALL wires to discover where the amps are still flowing even with the Inverter OFF.

@mvas I agree with your disagreement of my theroy, it's a mind bender and not being there makes it all the more difficult to diagnose. If there is a disconnect before the inverter, it should be turned off to isolate the array, thereby removing all potential load, or as you stated a disconnect for each string. Working on potentially live and loaded circuits is potentially hazardous, electricity cannot be seen heard or smelt, until something goes wrong, which is why in the industry there are lock out procedures followed before work commences. In the case of PV, which is a different beast, covering the panels as a form of isolation is the best was to avoid a potentially dangerous situation. 

sun_day

Hi gentlemen,

In this case, there is no DC isolator unit to disengage. Panels DC cables are connected straight to the inline fuses to the inverter.

Right now the mystery of why this is happening to string 1 and 3 and not to 4 to 6 is not concluded yet. Any mis-wiring, insulation fault such as wire being cut somewhere along the cabling routing shouldn't be an option here due to the initial test measurement being successful.

Of course for my part I will be going down next week to clamp the clamp meter to check on the amps of all the strings. But will like to have a solid theory understanding to assist with my onsite diagnosis next week.

BB.

Normally, with 3 or more parallel strings, you need one breaker or fuse per string.

This is to prevent one shorted string from getting over current from the other good strings and possibly catching fire.

-Bill

mike95490

Are all strings equal - same # of panels on each ? 
Do you use a combiner box ?  All those parallel strings ?

Photowhit

He's said he's using "in line fuses" a few times.

BB.

Sorry--I skipped over the disconnected before in-line fuses as the fuses were removed--My fault.

The in-line fuses are something like ~15 amps?

-Bill

mike95490

Photowhit said:

He's said he's using "in line fuses" a few times.

Sadly, inline fuses are not usable as disconnects like breakers are.   Fuses are a couple bucks cheaper, but when there is diagnosis to do, they are useless.

Photowhit

mike95490 said:

Photowhit said:

He's said he's using "in line fuses" a few times.

Sadly, inline fuses are not usable as disconnects like breakers are.   Fuses are a couple bucks cheaper, but when there is diagnosis to do, they are useless.

I was just addressing the combiner box discussion, no reason to have both, so the fuses I figured were it.


I actually think Sun_Day won't be doing this again,...lol.  He said it's "...our first time using fuses."

sun_day

@BB.

Each string has an inline fuse for positive and negative.

@mike59490

Strings are all equal. No combiner box is used. The strings are connected in parallel internally in STP20000TLEE-10 (unless I am wrong).

@Photowhit

Unfortunately I have no say on whether we will continue to use this. However once this problem can be identify, I will definitely feedback to request them to install a combiner box.

mcgivor

Suggestions for circuit breaker, something like this, many different ones available, different A ratings etc.

http://www.schneider-electric.com/en/product/A9N61660/c60pv-dc---2p---25a---800-v---b-curve/?range=61095-c60-for-dc-circuits&node=684682433-miniature-circuit-breaker

Estragon

@sun_day

I assume this is a negative ground system with panels, framing, and equipment grounded, and with DC negative tied to ground in (only) one place.

I don't understand why there are fuses on the DC negative wires from the strings though?