Hi all, would anyone be able to tell me what could have caused this failure on an Axpert inverter? Thanks

45 minutes ago, Jack007 said:

This is awfully. Looks more like a bomb

It looks like the solar charge controller had a massive overload, and/or arcing occurred, starting a fire.

Is that a 145 V max PV model?

The damage looks somewhat similar to what happened to a 450 V max PC model. In that case, the fire could have been prevented with a 100 mm length of electrical tape on the inside of the metal case. Two terminals stuck out too far, and made contact with the case. The high PV voltage started an arc that started a fire.

If that is a 145 V max model as I suspect, is it possible that it was wired to much higher voltage panels? Most inverters' solar charge controllers expect high PV voltage these days. Hundreds of PV volts could conceivably cause that sort of damage.

On 2024/05/24 at 5:27 PM, Coulomb said:

It looks like the solar charge controller had a massive overload, and/or arcing occurred, starting a fire.

Is that a 145 V max PV model?

The damage looks somewhat similar to what happened to a 450 V max PC model. In that case, the fire could have been prevented with a 100 mm length of electrical tape on the inside of the metal case. Two terminals stuck out too far, and made contact with the case. The high PV voltage started an arc that started a fire.

If that is a 145 V max model as I suspect, is it possible that it was wired to much higher voltage panels? Most inverters' solar charge controllers expect high PV voltage these days. Hundreds of PV volts could conceivably cause that sort of damage.

Thank you so much @Coulomb

Went to the site again today. Its the 145V 80A model. Jinko 555w panels. 8 groups of 2 in series with 2 inverters (4 groups per inverter). Checked the pv Voltage and each string was below 100V. Setup has been running fine for months. Inverters are second hand and one has been repaired due to generator damage (think the mppt was also replaced). Not sure which one was repaired. Repair was done before installation.

Can i rule arcing or component failure?

On 2024/05/29 at 5:33 AM, Jack007 said:

Went to the site again today. Its the 145V 80A model. Jinko 555w panels. 8 groups of 2 in series with 2 inverters (4 groups per inverter). Checked the pv Voltage and each string was below 100V. Setup has been running fine for months.

So that eliminates the "PV voltage way too high" theory.

On 2024/05/29 at 5:33 AM, Jack007 said:

Inverters are second hand and one has been repaired due to generator damage (think the mppt was also replaced).

The problem seems to originate from the MPPT MOSFEs, and/or the PV connections. I have seen a milder burning/fire here; there the cause was probably a screw making contact with the inverter case, causing arcing and burning of the plastic cover. It's possible that your inverter suffered the same problem; note the two screws attaching the cable from the MPPT inductors (top of inverter) to the MPPT board via thick red and black wires. Though looking at the side of the inverter, it seems like there is about 20 mm from the MPPT board to the top lip of the case. It seems possible that the screws protrude that far above the PCB; the screws anchor to pieces of metal some 8-10 mm high, so there are lugs, washers, and screw heads above that. A straight rule (e.g. level) across the face of the inverter would show roughly how close they would come. Are there clues on the back of the inverter's case near those screws?

Ignore the yellow squares; those are under the relays. The metal that the screws attach to are at the left of the below:

http://forums.aeva.asn.au/uploads/689/SCC_relays_at_output.jpg

The fact that the MPPT may have been replaced suggests that it's possible that the screws were not tight (either these two or the two at the bottom of the inverter,  or both), but I think that there is no way to tell for sure. The fact that the soot seems to  stop on top of the black wire suggests to me that at least part of the problem was higher up. But maybe that's just  due to the soot rising to the top by convection.

So I can't rule in or out arcing or component failure. Assuming that the repair did not involve board level repairs (e.g. replacing the MOSFETs with inferior specification ones), and the fact that these MPPT boards seem to be pretty reliable, I think component failure is somewhat less likely.

Examination of the back of the front cover may make it more or less likely that arcing occurred to those screws.

On further inspection, I believe that the screws were inserted incorrectly [ edit: but see below ]. The screw heads are supposed to be on top, not the nut and the rest of the screw:

 

Until you check the clearance, it's not clear whether this is significant or not. I'd estimate the nut to be 1-2 mm thicker than the screw head, and there appears to be 1-2 mm of thread sticking out. So that makes 2-4 mm less clearance to the metal case.

If only one screw made contact with the case, there should have been no problem, assuming that as is usual the battery is isolated from earth. So the screws would both have had to contact the case. I don't see any evidence of the ends of the screws having arced. If anything, the black lead (not battery negative) does look like it had arcing or at least extra heat.  Maybe that connection was loose? Perhaps give it a tug to see if it feels loose. The red wire looks fine by comparison. Putting the screws in upside down would not have changed how close the lugs would be to the case.

All very intriguing.

Any other armchair sleuths care to weigh in? Perhaps @BritishRacingGreen ?

Edit: Where the PCB is completely missing, there are TVS devices (Transient Voltage Suppressors). You can see one and a half of them in the top photo. It's possible that these were the parts that started the fire. The TVSs are directly across the PV input. For those to blow, it may have been a component failure, or maybe lightning or some weird problem with the PV wiring could have sent say grid voltage to these parts, but that seems unlikely.

Edit 2: The top photo is from one of my 60 A 145 V max MPPT boards. I'm 99% sure that my screws don't have bolts; the metal thing they attach to is tapped. It seems that the 80 A versions had studs soldered directly to the PCB:

So that means that my inverted screws theory is disproved. In this model, the black wire comes fairly close to a screw holding the heatsink to the board. Maybe if the black wire wasn't crimped properly, or the heat-shrink tubing not applied properly, or the wire was damaged, that could cause the damage to the black wire near where it attaches to the crimp lug.

Edit 3: The black wire's lug seems oddly off-centre somehow. Also, there seems to me more white soot near that black wire that elsewhere. Whatever that tells us.

Edited May 30, 20242 yr by Coulomb

14 hours ago, Coulomb said:

On further inspection, I believe that the screws were inserted incorrectly [ edit: but see below ]. The screw heads are supposed to be on top, not the nut and the rest of the screw:

 

Until you check the clearance, it's not clear whether this is significant or not. I'd estimate the nut to be 1-2 mm thicker than the screw head, and there appears to be 1-2 mm of thread sticking out. So that makes 2-4 mm less clearance to the metal case.

If only one screw made contact with the case, there should have been no problem, assuming that as is usual the battery is isolated from earth. So the screws would both have had to contact the case. I don't see any evidence of the ends of the screws having arced. If anything, the black lead (not battery negative) does look like it had arcing or at least extra heat.  Maybe that connection was loose? Perhaps give it a tug to see if it feels loose. The red wire looks fine by comparison. Putting the screws in upside down would not have changed how close the lugs would be to the case.

All very intriguing.

Any other armchair sleuths care to weigh in? Perhaps @BritishRacingGreen ?

Edit: Where the PCB is completely missing, there are TVS devices (Transient Voltage Suppressors). You can see one and a half of them in the top photo. It's possible that these were the parts that started the fire. The TVSs are directly across the PV input. For those to blow, it may have been a component failure, or maybe lightning or some weird problem with the PV wiring could have sent say grid voltage to these parts, but that seems unlikely.

Edit 2: The top photo is from one of my 60 A 145 V max MPPT boards. I'm 99% sure that my screws don't have bolts; the metal thing they attach to is tapped. It seems that the 80 A versions had studs soldered directly to the PCB:

So that means that my inverted screws theory is disproved. In this model, the black wire comes fairly close to a screw holding the heatsink to the board. Maybe if the black wire wasn't crimped properly, or the heat-shrink tubing not applied properly, or the wire was damaged, that could cause the damage to the black wire near where it attaches to the crimp lug.

Edit 3: The black wire's lug seems oddly off-centre somehow. Also, there seems to me more white soot near that black wire that elsewhere. Whatever that tells us.

@Coulomb I am not as good analyzing forensic evidence as you have displayed  here . But reading your take on possible inferior torqueing and mounting of the nuts and bolts preempts me to to believe that the impedance quality of the joints may be at stake here.  Lets assume the impedance of the resultant   dry joint  was small enough to  not consume enough power to influence operation , and neither to destruct itself. So everything was operating in reasonable harmony until possibly a short circuit was affected at the mppt itself , for whatever reason , maybe battery bus short or component failure (eg electrolytic cap ?) inside the mppt (I dont relate to all the mppt failure modes).  Under normal circumstances this is fine as the mppt and pv panels can handle a short circuit . But unfortunately the dry joint then became the total load burden  , got hotter and its own impedance increased slightly because of the degradation of the joint.  We know from the Thevenin theorem that maximum power transfer occurs when the load impedance matches that of the pv source impedance. So one theory is we reached a condition when that 'sweetspot' impedance got reached , AND that impedance became sustainable enough  over a short period of time. During that time  the pv dumped its max power into that joint , and could explain the vast amount of damage caused by say 2kW  power. Eventually the joint destructed and collapsed.

And under these conditions we don't experience overcurrent or over voltage , so pv fuses everything  will remain intact.

Of course my theory is open for debate.

 

 

Edited May 30, 20242 yr by BritishRacingGreen