Burnt disconnect board Xantrex GT

Here is a picture of what failed on on a Xantrex inverter after we replaced the top end. The original was a 2500 watt, the replacement a 2800. For some reason my guys did not replace the whole unit with the new style disconnect. They just used the new top end on the old disconnect. I’m dealing with that ill advised decision separately.

But that’s now what caused this failure. It wasn’t a wiring issue as it worked fine for 4 days. As this is on the DC side of the disconnect board, it would point to a DC sugre, but there is no ecidence of such and the dc arrestor mounted in the outdoor discinnect is undamaged.

There is a small amount of moisture at the bottom of the disconnect. The basement was humid due to drying out from flooding that was only 6” deep and never came anywhere close to the inverter. I suspect that little bit of moisture may have been due to internal condensation leaking from the top section, although everything was dry when we looked at it.

What looks like a water line of the breadboard isn’t, it is heat damage. It looks like all the traces became overheated. There is monitoring on the systems and it failed about 9:30 am on Saturday. The 15 min kWh was 0.08 just before the failure. In the 4 previous days of perfect operation that rate had been well exceeded. Unfortunately that monitoring does not show kW.

To me the only theories that make sense are either surge, or a short circuit that occurred and allowed too many amps to flow on the traces. No tripped breakers or blown fuses. The DC Surge Arrestor is fine, and the AC surge arrestor is still lit, so if surges occured, they were below the level required to scarifice those devices.

What am I missing?

Thanks!

Comments

  • BB.
    BB. Super Moderators, Administrators Posts: 33,631 admin
    Re: Burnt disconnect board Xantrex GT

    I would call your installer or Xantrex/Schneider directly... So far, the manufacturer seems to have been OK on fixing units that have failed in service (although, it takes a month or two to get the replacement inverter).

    The current generation of inverters have a huge set of knife switches to replace that small rotary AC/DC switch that is in your unit.

    -Bill
    Near San Francisco California: 3.5kWatt Grid Tied Solar power system+small backup genset
  • System2
    System2 Posts: 6,290 admin
    Re: Burnt disconnect board Xantrex GT

    Hi Bill,

    We are the installer. We put about 125 GT's in the field over about 16 months before switching back to SMA once we stared seeing problems. And I've replaced the bulk of them, some units more then once for all the normal failures common to the GT. This isn't a failure mode we've ever seen before.

    As you said Xantrex is good on replacing them, but it can take months to get the replacement since Schnieder bought them.

    Like everyone else, Xantrex GT's will haunt me for ever.
  • Solar Guppy
    Solar Guppy Solar Expert Posts: 1,989 ✭✭✭
    Re: Burnt disconnect board Xantrex GT

    Looks like the FR4 got cooked big time ... I'd guess something went south on the AC as that path looks like the output of the disconnect to the input to the inverter. How was the neutral and ground connected ( its cropped from the photo )

    I had a unit cook the PCB, lose DC wires overheated the connector on a GT5.0
  • BB.
    BB. Super Moderators, Administrators Posts: 33,631 admin
    Re: Burnt disconnect board Xantrex GT

    I only have the GT 3.x on my home--and it is the third one in six years. Not looking real good right now.

    I assume that your array is not "oversized" by a whole bunch with respect to the inverter (Isc-array too high?).

    I do not have enough knowledge of the product to guess why that happened. There are just too many possibilities. (we even had copper dendrite growth between traces take out circuit boards in our computers years ago).

    -Bill
    Near San Francisco California: 3.5kWatt Grid Tied Solar power system+small backup genset
  • Joe_B
    Joe_B Solar Expert Posts: 318 ✭✭✭✭✭✭
    Re: Burnt disconnect board Xantrex GT
    Looks like the FR4 got cooked big time ... I'd guess something went south on the AC as that path looks like the output of the disconnect to the input to the inverter. How was the neutral and ground connected ( its cropped from the photo )

    I had a unit cook the PCB, lose DC wires overheated the connector on a GT5.0

    That looks to me like a serious over current condition, To flash a trace that size off of a PCB would require serious amps. You are lucky it did not sustain an arc and burn a bunch of stuff up.

    EDIT: Now that I look at it more closely, it looks like an over voltage issue, looks to me like an arc went across that burnt area.
  • System2
    System2 Posts: 6,290 admin
    Re: Burnt disconnect board Xantrex GT

    We only have nine 200 watt modules connected to that inverter, and the previous one ran 4 years (with only one replacement for the capacitor disconnecting issue).

    I just got this information which sounds like it could be on the right track:

    "I just noticed something in the picture, look at the damaged location, it is the space between the two traces, positive and negative. It looks to me like an arc started between the two traces. The traces burnt until the gap was too large and it went out. I was under the impression that the green color you see on the trace would insulate the bars but may not from moisture. I don't think a surge did this, the more I look at it, it appears to be an arcing event between positive and negative. The unknown is what started the arc, my first guess would be the moisture condensing on the board finding a tiny open section of buss bar coating and starting the arc, then burnt off additional coating allowing it to keep burning."
  • BB.
    BB. Super Moderators, Administrators Posts: 33,631 admin
    Re: Burnt disconnect board Xantrex GT

    The green is typically solder mask and only there to keep solder on the pads/around the holes when making the product. It does not add (substantially) to the electrical properties of the PCB.

    Normally, we used a minimum of 1/8" for AC line separation to ground/isolated circuits. Normally, that would be for a 600 VAC rated/1,800 VAC high pot.

    Normally, the insulative properties through the PCB (the solid, not the curable layers) is pretty much not a problem with arcing.

    And horizontal (between traces/copper), you would assume that the PCB material would be a pretty good insulator--However, it is not guaranteed that there are no voids (air pockets) between traces/copper--so you should assume air as the dielectric material (hence, back to the 1/8" gap for 600 VAC rated circuits.

    It has been too many decades since I was in board design, but a clean circuit board will have, very roughly, 50 volts per 1/1,000" arc over, and >500 volts per 1/1,000" vertically through the PCB material (if I recall correctly--it has been too long).

    So--My assumption is that either a copper whisker, some debris (even a bug) may have (with moisture) created a path for current flow... And with DC, once there is any sort of current flow, an arc can be started and sustained very nicely until somebody notices (or the sun goes down).

    And with solar panels being (more or less) constant current sources, any short circuit will not clear a breaker/fuse (no extra current available to pop the protective device). And why the NEC and others are looking at Arc Fault detection for solar PV power systems.

    -Bill
    Near San Francisco California: 3.5kWatt Grid Tied Solar power system+small backup genset
  • halfcrazy
    halfcrazy Solar Expert Posts: 720 ✭✭✭
    Re: Burnt disconnect board Xantrex GT

    That is a new failure for me to I have about 6 of these installed and have changed 5 of them at least once and some 2-3 times. This has to be one of the biggest disappointments in product I have ever had the pleasure of messing with.
  • RCinFLA
    RCinFLA Solar Expert Posts: 1,485 ✭✭✭✭
    Re: Burnt disconnect board Xantrex GT

    Looks like electro-migration between the PV + and - lines.

    There is a moisture line still visable across the board that goes right through where the burnt PCB was. A thicker PCB resist coating may have saved it but typical PCB coating is not a totally hermetic seal.

    Any contaminate salts (like from a sweaty hand) on the PCB, plus moisture, plus a large DC bias will start a metal migration.