EG4 3000 EHV-48 fire. Been running a 12 panel system with 2 EG4-lifepower batteries and the EG4 3000 EHV-48. Added the battery rack and 3rd battery

[FilterGuy]

For a fast fault (1s<t<10s) I have seen one end of a 2' cable be hot and the other just warm. It usually depends on which side of the cable termination was the source of heat.

I agree. I still think the burning wire was a secondary effect due to improper fusing. If a bad connection caused a heat problem that created a short. I would have expected it to take long enough to heat up most of the wire. Also, the reported load on the inverter at the time of the failure was very low....This means that while it was working the inverter was not drawing a lot of current so any heating would have been slow and would have traveled down the wires..

I wish I knew what the circuit in the upper left of the unit was. Was it's failure also a secondary effect or was it the primary failure that cascaded.

My SWAG remains this:

1) Something in the Inverter failed and caused a low resistance short (If the FETs shorted it could cause this).
2) The short created a sudden high surge in current. Even with the 100A Fuses, the current could have been well north of 200A total.
3) A bad connection or a bad crimp on 4AWG battery wires was the hottest part of the circuit and caused the fire.

Question: Can we get a picture of where the cables are connected to the inverter? Is there any indication the heat came from or went into the inverter itself?​

As far as what caused the initial problem that cascaded into the fire.... we have so little information that anything said is speculation.... but what the heck... I'll speculate.

The OP states that the charging seemed to have stopped just prior to the failure
The OP states he had north of 400V on the PV input..... And the 500V design is brand new on the EG4 versions of these inverters. (Could there be a design flaw?)

Could something have fried on the Charge controller? It has a direct connection to the batteries so a short in the charge controller could create a huge current on the battery input. (Could the burnt circuit in the upper left be the solar charge controller?)

 

[Endsofearth]

I agree. I still think the burning wire was a secondary effect due to improper fusing. If a bad connection caused a heat problem that created a short. I would have expected it to take long enough to heat up most of the wire. Also, the reported load on the inverter at the time of the failure was very low....This means that while it was working the inverter was not drawing a lot of current so any heating would have been slow and would have traveled down the wires..

I wish I knew what the circuit in the upper left of the unit was. Was it's failure also a secondary effect or was it the primary failure that cascaded.

My SWAG remains this:

1) Something in the Inverter failed and caused a low resistance short (If the FETs shorted it could cause this).
2) The short created a sudden high surge in current. Even with the 100A Fuses, the current could have been well north of 200A total.
3) A bad connection or a bad crimp on 4AWG battery wires was the hottest part of the circuit and caused the fire.

Question: Can we get a picture of where the cables are connected to the inverter? Is there any indication the heat came from or went into the inverter itself?​

As far as what caused the initial problem that cascaded into the fire.... we have so little information that anything said is speculation.... but what the heck... I'll speculate.

The OP states that the charging seemed to have stopped just prior to the failure
The OP states he had north of 400V on the PV input..... And the 500V design is brand new on the EG4 versions of these inverters. (Could there be a design flaw?)

Could something have fried on the Charge controller? It has a direct connection to the batteries so a short in the charge controller could create a huge current on the battery input. (Could the burnt circuit in the upper left be the solar charge controller?)

I like your swag, bad connection is out, I carefully checked and both wires were tight and no insulation under the set screw. Had to use my impact driver to remove them and the ends are bright, not burned and no sign of arcing. The fire seemed to be outside the box right where the wires went though a bushing (very tight) into the box. I can measure them and see if any burning was inside. I disconnected and moved all the burned parts outside as fast as I could to get it all away from the battery rack. I wish I had an internal circuit design. I will take more photo's tomorrow and post them. I need to fix and get power back as my generator is using a lot of gas.

 

[dianea]

The pictures are what I would expect for terminal connection failures, often caused by low torque, dirty surfaces, improper crimping.

 

[Endsofearth]

Photos of the wires and ends, I think they overheated and the reason the rest of the cable was cool was due to the CO2.

 

[Endsofearth]

Good point. If the inverter failure caused the wire to overheat, the whole wire would heat simultaneously due to the current flowing from end to end. However, if there were a loose connection at the inverter. Say, maybe he torqued down on the insulation and not the bare wire by mistake, then the wire would get very hot but only at the inverter end where the bad connection was at. When the connection finally got so hot it caused a FET to pop. IMO, this is more likely than a random inverter failure for no reason.

Photo's of the wire ends, you can see they got hot but were the right length in the connection, no insulation in joint, burning started just below, cables may have been hot full length but the CO2 could have changed that or when the unit failed it heated rapidly. Rebuilding I will be adding more breakers off the battery bank

 

[Hedges]

I think the picture shows tinning of wire has melted, soldering strands together just at the end. So that does appear to be high resistance connection to me.

That could have happened with high current, entire wire heating but resistive connection heated faster. Or could be at relatively low current.

The insulation inside the box seems to be a nicely toasted marshmallow, while outside box it supported flames and became blistered. Did smoke distribute inside the box, possibly causing electrical conduction and damage to components at other location?


I would agree multiple breakers/fuses are needed. If several batteries are individually fused, but combine on a busbar and feed one wire, that wire needs OCP if it's ampacity exceeds sum of fuses feeding it. But as for what caused over-current (if that was the situation), we don't know. If high contact resistance, fuses don't help. Proper torque, temperature measurement, subsequent inspection can help.

I have taken to rotating wires back and forth in the terminal and retightening because I find strands settle. Ferrules could help in some cases. This is fine stranded wire; is the terminal designed for that? Does is specify ferrules? Some terminals clamp a yoke around wire, while others just grind a setscrew into them.

 

[Endsofearth]

I think the picture shows tinning of wire has melted, soldering strands together just at the end. So that does appear to be high resistance connection to me.

That could have happened with high current, entire wire heating but resistive connection heated faster. Or could be at relatively low current.

The insulation inside the box seems to be a nicely toasted marshmallow, while outside box it supported flames and became blistered. Did smoke distribute inside the box, possibly causing electrical conduction and damage to components at other location?


I would agree multiple breakers/fuses are needed. If several batteries are individually fused, but combine on a busbar and feed one wire, that wire needs OCP if it's ampacity exceeds sum of fuses feeding it. But as for what caused over-current (if that was the situation), we don't know. If high contact resistance, fuses don't help. Proper torque, temperature measurement, subsequent inspection can help.

I have taken to rotating wires back and forth in the terminal and retightening because I find strands settle. Ferrules could help in some cases. This is fine stranded wire; is the terminal designed for that? Does is specify ferrules? Some terminals clamp a yoke around wire, while others just grind a setscrew into them.

The pop I heard from 20 plus feet away and smoke was coming out of bottom of unit, I shut off power switch and went for fire extinguisher and wires were burning outside the box when I got back, maybe a minute or less. Shut off solar input and batteries next. So I am thinking failure caused wires to overheat. Charging seemed to have stopped at 95% for an hour or more.

 

[Endsofearth]

The pop I heard from 20 plus feet away and smoke was coming out of bottom of unit, I shut off power switch and went for fire extinguisher and wires were burning outside the box when I got back, maybe a minute or less. Shut off solar input and batteries next. So I am thinking failure caused wires to overheat. Charging seemed to have stopped at 95% for an hour or more.

The terminal is a square box with a flat plate that tightens with the set screw crushing the wire. Waiting to hear back from Supplier.

 

[dianea]

If the termination was soldered, I hope it was silver solder as tin solder often burns in high current applications

 

[Cyanlite]

The OP states that the charging seemed to have stopped just prior to the failure
The OP states he had north of 400V on the PV input..... And the 500V design is brand new on the EG4 versions of these inverters. (Could there be a design flaw?)

I think you're onto something. My SWAG? The inverter is just a rebadged LV6548 that normally has a 250V PV input limited by firmware and hardware. The rebadged version removed the firmware limits thinking the hardware could handle it. But the hardware couldn't handle it, causing a FET to pop, cascading into melting the wires and sparking a fire.

 

[Endsofearth]

I think you're onto something. My SWAG? The inverter is just a rebadged LV6548 that normally has a 250V PV input limited by firmware and hardware. The rebadged version removed the firmware limits thinking the hardware could handle it. But the hardware couldn't handle it, causing a FET to pop, cascading into melting the wires and sparking a fire.

Yes, for it to be over 400 volts that would indicate a no load voltage so the system had failed at least an hour earlier as it had quit charging at 95%.

 

[Endsofearth]

If the termination was soldered, I hope it was silver solder as tin solder often burns in high current applications

Tinned, not silver soldered, they came pre-cut and tinned. I have rolls of silver solder, I will cut and silver solder in the future unless silver solder would be too rigid and keep the wire from flattening in the connectors.

 

[FilterGuy]

The terminal is a square box with a flat plate that tightens with the set screw crushing the wire.

This type of connection is probably adequate for the max 80A the inverter will normally draw, but it is not the best type of connection. When the internal short happened, there was almost certainly a hot spot in the connection.

Also, in my earlier comments, I was thinking there were 100A fuses on each battery. If it was just the breakers then once the short occurred, the current could be *way* north of 200A. Breakers tend to be very slow to break unless the current is a multiple of the trip value.

 

[Kornbread]

Does anyone else find it odd that the eg4 LiFePower4 batteries seem to have issues with starting some of the larger inverters, but in this scenario they dumped enough current to turn 4awg wire into crispy french fries?

 

[dianea]

Tinned, not silver soldered, they came pre-cut and tinned. I have rolls of silver solder, I will cut and silver solder in the future unless silver solder would be too rigid and keep the wire from flattening in the connectors.

Always scrape or cut away the tin completely off to bare copper. Unless it's the real elemental silver, which has even better conductivity than copper. Silver terminals are the gold standard. Gold plate is ok, because it doesn't have an oxidation layer. Clean copper to copper crimping is good and is possible to cold weld under enough pressure. If the terminal is properly torqued directly onto a quality bus bar, it's a lifetime guarantee...

 

[dianea]

Does anyone else find it odd that the eg4 LiFePower4 batteries seem to have issues with starting some of the larger inverters, but in this scenario they dumped enough current to turn 4awg wire into crispy french fries?

The problem may be resistance of poor terminations at the end, which become hot spots and "fuse" out the circuit. My EG4-LL does pretty good and can even warm up 2awg cables at 100amps. I checked everything at maximum load with a thermal camera to be sure no hot spots or bad connections could create fires later. This is usually done every year at large facilities, so we don't have property damage or downtime due to fires.

 

[Cyanlite]

The problem may be resistance of poor terminations at the end, which become hot spots and "fuse" out the circuit. My EG4-LL does pretty good and can even warm up 2awg cables at 100amps. I checked everything at maximum load with a thermal camera to be sure no hot spots or bad connections could create fires later. This is usually done every year at large facilities, so we don't have property damage or downtime due to fires.

Anything to protect against that? 100amps continuous probably isn't safe for #2 cables. But then again, I'm not a licensed electrician, I'm just a random poster on the Internet.

 

[dianea]

Anything to protect against that? 100amps continuous probably isn't safe for #2 cables. But then again, I'm not a licensed electrician, I'm just a random poster on the Internet.

I would do a thermal scan for hot spots, if it's not hotter than body temperature, it should be good. The insulation should have a temperature rating, like 80C, and the flame resistant leaded insulations are much higher (and expensive.) Look for the "UL" label printed on the cable, which requires specific insulation testing, including high voltage tests (usually 7.5-30KV) on every inch of cable made. I used to check and document the calibration of the testers that made cable with a NIST traceable calibrated meter.

 

[shopman]

Does anyone else find it odd that the eg4 LiFePower4 batteries seem to have issues with starting some of the larger inverters, but in this scenario they dumped enough current to turn 4awg wire into crispy french fries?

This was an issue of resistance. A short in the inverter caused current to flow at a controlled rate, a rate that was under the limit of the BMS and circuit breaker so the battery did its thing and provided power. This is very different from inrush current where the instantaneous demand for current can far exceed what the battery can provide, thus the BMS shutdown.

 

[Shimmy]

Does anyone else find it odd that the eg4 LiFePower4 batteries seem to have issues with starting some of the larger inverters, but in this scenario they dumped enough current to turn 4awg wire into crispy french fries?

Not at all. #4AWG will melt at a sustained current over ~200A. The system could provide a sustained current of 300A. The #4 was not adequately protected, especially when the third battery was added. The system seems to have behaved exactly how one would expect after a DC short.

Root cause is presumably the high PV voltage relative to the system rating, and the inverter should be replaced under warranty. (And EG4 should stop including cables from the battery to the inverter unless they can guarantee that they are adequately protected.)