Below are two images of the first inverter damage .  The major damage seemed to occur on the load output neutral connection to the PCB , also the red capacitor just next to that connection had a good mouthfull. Initially I got surprised because i thought this is the grid input , but i think we can safely say this is the load output .

 

Thanks for the pictures @BritishRacingGreen

The high current through the load N seems like load current to/from the utility network. 

I found a strange thing some months ago. While charging batteries on my Axpert clone from grid I connected a TRFR based power supply on 14V via a LM338 voltage regulator to one of my 12V batteries to give this battery a bit more charge as it was about 0.2V below the other battery. Although the LM338 is only a 5A regulator I measured 10A flowing in the 14V negative and 4A in the positive. It gave me quite a shock and disconnected on the spot. A silly thing not to try again. 

Edited December 29, 20223 yr by Scorp007

I dont see a lot of output surge protection components on the load output , except some capacitors . On face value it appears to me because the damage is heavily on the output terminals , that the inverter got surge from outside into the board . This is as opposed to overload  which would have caused the power chain to have stressed . So a surge came from outside , pcb surge component eg capacitor shorted and allowed  heavy current on the load side  , including the outgoing pcb connection terminal. 

This is the metrics during typical load shedding . Battery was discharging just over 150W  . Which is no surprise , because only DSTV,TV and couple of lights  on. 

The 42.0v metric labeled END is somewhat ominous , if that's a battery metric it is low . But i am sure that's not a contributing factor .

Input grid shows zero , that's important as it voids my initial theory of grid to load short .

 

Thanks for all your input. I will let you know what Mecer has to say. I have measure the E to N on the DB and it measures 0 so no stray current. This replacement inverter has run trouble free for 16 days with 3-4 load sheds per day, like wise the previous inverter ran for a similar period. If something is wire incorrectly it stays that way till corrected and should give an error  on the inverter at some stage during the 16 days. No trips or error codes ever generated and it has always cut over seamlessly which is why the concern. I would have preferred to have errors etc to help identify the problem, it took 2-4 minutes to fry itself at the start of load shedding. Can't figure this out so i don't have a clue what to do differently, but thanks for the support.

5 hours ago, BritishRacingGreen said:

Below are two images of the first inverter damage . 

Interesting. Something seems to have burned near the PCB connections for AC-out L and N. I'd say there was a flame, since it blackened the live and to an extent the neutral wires, but they don't seem to be damaged over their entire length.

Possibly some sort of surge on at the AC-out port caused some capacitors to fail shorted and start the fire. But strange that this didn't cause an overload fault straight away, preventing most of the damage.

1 hour ago, Coulomb said:

Something seems to have burned near the PCB connections for AC-out L and N

Looks like the burgundy cap to me.

13 hours ago, P1000 said:

Looks like the burgundy cap to me.

Could be. It's hard to tell whether it's the cause, or just got blackened by other things burning and emitting soot. Right down on the PCB seems to be very charred to me, so my guess is something down there.

Hi @Coulomb @Scorp007 @P1000 , I got some additional info info from OP , and for the good order I can state :

1. The 5kw MKS4 is mounted in the living room , as it serves the purpose of only powering DSTV, TV and 1 or more non-fixed lights

2. The legacy main db board  feeds a  legacy sub db board where all the house needs are distributed from . This legacy sub db board has been fitted with a single pole 20A breaker , and feeds a 4mm (could be 2.5 not sure) surfix route to a UPS style plug point breakout labelled as UPS SUPPLY . From there it is connected to the AC input of the inverter.

3. The AC output of the inverter is routed to a mini db box in very close proximity to the inverter. It contains an ELD and a double poled 20A load breaker.

4. From the load breaker a surfix cable routes the output to another UPS plugpoint labelled as UPS LOAD . From this point a multiplug is fed via a surge protection module . This surge protection looks not unlike the typical Ellie's type i am used to.

5. From the multiplug the appliance are routed.

The scope of failure is so unusual and destructive , that I have suggested to the OP that there may remain a possibility that the two inverters could be part of a batch problem . This cannot be overlooked , as there is little evidence that any  integration of grid and load could cause such damage.

 

6. It is also important to note that this circuit has actually been serving a 3KW 24V lead acid arrangement (EDIT : 3KW inverter and 100Ah batteries) for what I believe is a period of 2 or 3 years . No Problems. Also note the reason why OP upgraded was solely on the poor performance of lead acid storage .

Kind regards

 

 

Edited December 30, 20223 yr by BritishRacingGreen
Extra info

OP has made me aware of feedback he received from another forum , where the probability of the inverter to inadvertently feed back power onto a dead grid  is taken into consideration.

The MKSIV has a so called safety relay which connects or disconnects grid to load . This is of course essential as the inverter uses grid-tie arrangement to entertain  ac blending  in addition to dc sources blending.  The question arises wether a failure mode of this relay could cause inadvertent output on the grid side during power outage. 

My take on this is it is unlikely , or better put , the probability of three conditions to be met to sustain power output on a dead grid , is too low .

The three conditions are 

(a) The safety relay must become stuck in its energized position, or the DSP drive output fails to a state where the relay remains energized.

(b) The inverter load sensor must be faulty in order for no load to be registered to the DSP (ie. not detecting overload / surge) 

(c) There must be no grid power available

Hence my reasoning is that condition (A)  AND (b) AND (c) must be met for a wrong side failure . (B) is taken into consideration as the inverter will detect overload when the safety relay is stuck , aborts and produces an overload error. The  probability of all three conditions to be met , is significantly small enough in my opinion  . I would rate this level of Safety Criticality on level SIL2  , although I have not made calculations.

HOWEVER, as per Safety Integrity assessment ,   this probability only holds water  if the conditions (a) , (b) and (c) are pro-actively DETECTABLE.  And this is where I need some input from @Coulomb and others . For instance let us say the safety relay becomes stuck as being the only fault condition in the inverter. The question is does this fault becomes apparent  to the inverter by means of indirect or other faulty behavior detected by the DSP . We know the DSP cannot directly detect a stuck relay , because a normally closed contact is not fed back to the DSP to evaluate the condition.

Secondly , as per condition (b) , we know that the DSP cannot verify the integrity of the load current sensor , but again , does the failure of  this sensor not become indirectly apparent to the DSP during the evaluation of other functions that's dependent on the load current metric.

I personally don't think that this failure mode is cause to the OP's situation , but I do want to rule out the probability.

 

 

Edited December 30, 20223 yr by BritishRacingGreen

3 hours ago, BritishRacingGreen said:

For instance let us say the safety relay becomes stuck as being the only fault condition in the inverter. The question is does this fault becomes apparent  to the inverter by means of indirect or other faulty behaviour detected by the DSP .

This might not be detected, if the inverter is in SUB output priority most of the time. In this mode, AC-in and AC-out are connected all the time already, with the two grid (safety) relays energised. Grid current is not measured, but load and inverter current is. So if the grid imposed a massive load on the inverter due to stuck-on grid relays, there would be an overload detected in the inverter output. But I assume that's what you mean by (b); the word "load" confused me, as there is a separate detection (via the current transformer) of load (as in AC-out) current.

If the grid relays were stuck on at power-up, this would be obvious, as there would be grid voltage at the output of the inverter when it attempted to soft start the DC-AC converter. There is a fault code reserved for this situation: fault code F53, inverter soft start failed. But inverters typically run for months without being restarted.

3 hours ago, BritishRacingGreen said:

Secondly , as per condition (b) , we know that the DSP cannot verify the integrity of the load current sensor , but again , does the failure of  this sensor not become indirectly apparent to the DSP during the evaluation of other functions that's dependent on the load current metric.

I would think so too, but I can't immediately think of a situation where a faulty inverter current sensor would be detected. I don't immediately see a fault code specific to this situation, either. There is fault code 57, "current sensor failed", but that's only set by a test that happens only at power on, and it appears that it involves the load current sensor (the current transformer), not the inverter current sensor (a Hall Effect device).

Sorry, I'm not giving a definitive answer here; it's difficult to eliminate this possibility.

I know the op and it is a big safety issue to see two units fail like this.

I have the older Axpert 5KVA from around 2015 and at the time managed to track down a service manual for my unit. I realize that this may not be exactly the same as the MK4 version but it does show a block diagram. See below

 

Note both these inverters are not grid tie so they do not synchronize with the grid supply. In non-load shedding the unit will be in bypass mode and the safety relay will be energized (The input is switched to the output) and the inverter relay is open......although it does still need to be able to charge the battery.

What if, when load shedding starts it engages inverter mode too quickly before letting the safety relay de-energize. Could this not put 220V supplied from the inverter back to the input and hence the DB?

I have attached the service manual for the older model Axpert unit as it does have some for the detail circuit diagrams. 

axpert_mks_4-5kva_service_manual_20131104a.pdf

5 hours ago, LeonardF said:

managed to track down a service manual for my unit. I realize that this may not be exactly the same as the MK4 version

Most of the new models have a high voltage Solar Charge Controller, so the block diagram is a little different.

5 hours ago, LeonardF said:

Note both these inverters are not grid tie so they do not synchronize with the grid supply.

Actually, they do, so that they can reverse the inverter (DC -> AC) into a unity power factor synchronous rectifier (AC -> DC). This is also what causes them to occasionally burp puffs of power into the AC "input".

6 hours ago, LeonardF said:

In non-load shedding the unit will be in bypass mode and the safety relay will be energized

It could well be, but it depends on settings and conditions.

6 hours ago, LeonardF said:

Could this not put 220V supplied from the inverter back to the input and hence the DB?

Yes, it can and it does. Hopefully most of the time it gets absorbed by the non-essential loads before it gets back to the actual grid. Though this is not such a shocking thing that Eskom seems to regard it as. Having all relays closed is a common occurrence. One of the firmware's jobs is to carefully balance the amplitude and phase of the inverter against the grid, to push power in the right directions and with the right magnitude for all conditions, while load and solar power can be fluctuating. It's quite tricky. 

9 hours ago, LeonardF said:

not synchronize with the grid supply. In non-load shedding the unit will be in bypass mode and the safety relay will be energized (The input is switched to the output) and the inverter relay is open......although it does still need to be able to charge the battery.

The MKS4 as well MKS3  is grid-tied, but the machine does not export power on its ac input. 

Theres two reasons for this. It allows:

1 the power chain to be simpler in that it is fully bidirectional, and hence illiminates the need for an additional battery charging subsystem which  requires, amongst other a bulky transformer. 

2.it allows for the powerfull functionality of  ac blending. So if the combined power of pv and battery is not enough to supply the load, the grid will augment the tequired balance. 

3. It introduces batteryless mode. This means if no battery is connected, the load will be supplied by a 'mixture' of pv and grid. However this is not feasible when the grid is not available, because pv on its own will struggle to meet the continuous load match if left to its own devices. 

The notion of bypass is therefore an old legacy term,  there is no more hard bypass (line) or hard battery mode. 

Edited January 10, 20233 yr by BritishRacingGreen

On 2022/12/29 at 8:40 AM, BritishRacingGreen said:

Below are two images of the first inverter damage .  The major damage seemed to occur on the load output neutral connection to the PCB , also the red capacitor just next to that connection had a good mouthfull. Initially I got surprised because i thought this is the grid input , but i think we can safely say this is the load output .

 

Hi @Coulomb i am revisiting this thread within context of the MKS4 weak pcb bonding from neatral to earth. Look exactly where the burn took place, exactly above the thin track.  I maybe not be on something here, as surely the track is to thin to cause harm like this. Or can it?

14 minutes ago, BritishRacingGreen said:

Hi @Coulomb i am revisiting this thread within context of the MKS4 weak pcb bonding from neatral to earth. Look exactly where the burn took place, exactly above the thin track.  I maybe not be on something here, as surely the track is to thin to cause harm like this. Or can it?

Or, could it be that the blue capacitor between output neatral and earth blew up and caught fire, maybe when the thin bond broke. 

The location of this cap is shown below :

If you look at the burn image, this cap looks destroyed. 

Edited March 8, 20233 yr by BritishRacingGreen

On 2022/12/28 at 1:40 PM, garyrutt said:

I bought a Mecer Axpert 5.6K MKS IV 5600VA / 5600W 48V MPPT Inverter (SOL-I-AX-5M4) in November. Within 2 weeks after installation it caught alight a few minutes into load shedding. Mercer identified the fault as a "Blown AC Output" from a power surge??? This 5kva unit supplied power to a TV, DSTV box and 3 lamps so i asked how this was possible but the evidence showed the blown AC Output. Mercer gave a replacement unit which was installed the same way and 16 days later on Boxing day, 3 minutes into load shedding it went up in smoke again. Now i need to get it back to Mecer for evaluation. I have since brought in an independent Electrician who confirmed the installation was correct. The question i have, the inverter has an AC supply in, an AC Supply out and a 48v DC connection the the Li battery. What can go wrong? Where is the inverter protection to prevent this..... so many unanswered questions so hoping someone on this forum can suggest a failure mechanism. Thanks

@Coulombhere is the interesting part. If you read the OP opening description, the failures occured exactly a few minutes into loadshedding , meaning the grid was disconnected, so the relay bond was the only bond active at the time!  whats more it happened twice, with his original MKS4 and with replacement MKS4.! 

The owner informed me he installed a Sunsynk today, so he is sorted, but these failures still haunts me.  Stepping down for now. 

 

 

Edited March 8, 20233 yr by BritishRacingGreen

14 hours ago, BritishRacingGreen said:

If you read the OP opening description, the failures occurred exactly a few minutes into loadshedding , meaning the grid was disconnected, so the relay bond was the only bond active at the time!  What's more it happened twice, with his original MKS4 and with replacement MKS4.! 

I like your theory. Perhaps the OP had some moderately heavy leakage from active to earth in some faulty appliance. Or perhaps it was wired wrongly, with the actual load between active and earth. With a proper, strong bond from neutral to earth, it would work normally, though the earth leakage breaker should have been screaming blue murder. With the MKS IV's way-too-thin connection to earth fusing, if the faulty appliance had load from active to earth, it would have stopped working, yet put the full mains across the blue capacitor. It should be rated for continuous operation at mains voltage though.

So we need a lot of faults all at once to make this work, though you've identified two of them: the too-thin track on the MKS IVs. We need the faulty appliance, a faulty or non-existent earth leakage breaker on the inverter output, the blue capacitor burning up, and possibly the OP not noticing that one of their appliances stopped working.

Possible, and intriguing. I get the feeling that I'm missing something that makes this theory, or a minor variant of it, explain the behaviour with fewer faults.

On 2022/12/28 at 4:22 PM, garyrutt said:

Hi I'm also using a Vestwoods 48v 100Ah Li battery.VT48100E-A2 51.2V battery with a Mecer Axpert MKS IV 5.6kw inverter. How did you get communication between the two and what protocol did you select on Inverter?

 

On 2023/03/08 at 7:46 PM, BritishRacingGreen said:

Or, could it be that the blue capacitor between output neatral and earth blew up and caught fire, maybe when the thin bond broke. 

The location of this cap is shown below :

If you look at the burn image, this cap looks destroyed. 

I encountered this particular defect on 3 inverters (all EASUN SV IV) - The PCB is exactly as in your pictures, with that thin trace connecting Neutral to Earth in battery mode. Sometime burns like that.... sometimes is more messy (burned hole in the PCB)

possible rootcause for 1st inverter: User had some faulty wires on PV panels. Panels were mounted on metal support, when there was a rain outside there was current flowing from panels to earth. I think that reverse current somehow destroyed the thin wire.

Possible rootcause for 2nd inverter: User was trying to feed power to another Inverter (OUTBACK type). I measured 30V AC, 50Hz on Generator Input of the OUTBACK inverter (I dont know this inverter so I did not investigate more). There was also a problem with an AC cable he was using - there was short circuit in the cable and the cable heated up. (probably L was shorted to Earth in the cable) - he had no Differential protection mounted and no fuse on AC output.

Later edit: Now I wonder if that thin wire is a mistake or is intentionally done like this. On EASUN SV2 I have not seen it.

@Coulomb you were mentioning Fault 53 in a previous post. Could you please explain what conditions are checked for this fault? (Long story short: I started repairing an Easun SV IV inverter - display board was defective, changed uC, reflashed, OK, replaced burned mosfets, replaced IGBTs, replaced burn resistors, one optocupler+diodes. Now I powered ON and is stuck at Fault 53. I measured 360V on the BUS, so I suppose the DC-DC converter is working OK. No AC out but I've seen 50Hz displayed for a second on the main display. Any ideas what to check next?

Edited July 27, 20242 yr by sethmad

2 hours ago, sethmad said:

@Coulomb you were mentioning Fault 53 in a previous post. Could you please explain what conditions are checked for this fault?

Fault code 53 is inverter (not bus) soft start fail. So the bus voltage is OK, and conditions are right to turn on the main inverter (DC->AC converter). Early on, this is done with the output relay off, I believe. The target voltage is set to 5 VAC, and the inverter output voltage is measured. If 10 volts or more lower than the nominal voltage (usually 230 V, but it could be set to 220 V o 240 V), the target voltage is set to 5 VAC higher. This continues until the output voltage is less than 10 volts short of the nominal voltage. Normally, this should cause the DC->AC converter output to ramp up to nominal voltage in just under one second. If it takes 5 seconds, then the inverter soft start fails. If it happens in less than 5 seconds, the inverter soft start passes, the target voltage is set to the nominal value, and the output relay turns on, powering the load.

I'm reading his from old firmware, because this is the only place I've commented it reasonably well, but I'd expect the same basic test to happen in all models.

So I'd say you still have something wrong with a DC->AC IGBT gate driver, the IGBTs themselves, or maybe the buck transistor is open circuit.

Consider using my "functional test" of the gate drivers here. You need two power supplies, and it's a bit tedious, but you can see the actual gate voltages for both IGBT on and IGBT off conditions, on the bench.

Edited July 27, 20242 yr by Coulomb