Mike, I had two of the 2000w G1 with limiter's do the same after approx a year of use. I had used them in the same manner as you for a year in 2015, dying in early summer 2016. Did exactly what yours did - blew a main internal fuse. Testing all the mosfets turned out most had blown, sourced some new ones and replaced them. Was told by the manufacturer that this had been addressed in the new model ( G2 - the one you have ) but obviously this is incorrect, seems the only thing they have done is moved the Mosfets to the middle to attach to the heatsink internally and also the one on the base. where as in the G1 they were on the outer edge and attached to the cover which was a finned aluminium design. After loosing the 1st GTIL, I did run external cooling over the 2nd to try and keep the temperature down incase this was the cause of the 1st failing, but this 2nd one still eventually failed anyway. As already mentioned by others I think the main cause is poor design and the components utilized within. I did at the time fault the inverter and flag the Mosfet issue back to the manufacturer - who offered to send me some new Fets.Instead I sourced these from a local electronics company to save the shipping costs etc from China. I have since moved to a more comercial inverter ( Victron Quattro ) which is made like a tank ( and also weigh's like one) but thats due to the transformers and other HD components inside. This also performs far better than the GTIL and can run as a Grid tie or standalone all day + has a built in charger for any battery types - only downside is obviously price, but for safety and peice of mind I decided to go this way in the end. I did have mine fused individually on the DC with circuit breakers which did trip at the time but only after the GTIL had blown its Fet's etc. Your experience of the GTIL unfortunately also confirms for me that the issue I had with the G1 hasn't been cured with the G2, as promised by the manufacturer who offered me a trade up to this G2 model as a solution + additional costs on my behalf. Decided at that stage ( summer 16) to buy a more dependable unit, and after much research bought my Victrons. These have been rock solid, like all my other Victron products - which does prove you get what you pay for in the end. Check all your Mosfets, and I'm sure you'll find a few faulty ones - Good Luck on the repair!! & happy tinkering....
i'm not real impressed with the design of the inverter, the MOSFETs being positioned the way they are makes them virtually impossible to get to. That being said, I noticed that on the inside of the top cover there appears to be a spot of smoke to indicate where the problem is at.
That smoke/black mark in the lid (3:59 in vid) looks like it might line up withe the component that you spotted with the hot mark round the solder joint (11:20 in vid).
check the gate pulldown resister and diode if they have blown as well just replacing the mosfet wont fix it and the new mosfet will blow again also check the transformer on that line for a dead short
Mikes DIY Tesla Powerwall it's on the under side of the board on the gate leg it is the surface mount with an R on it and the diode is just above the resistor
I love those eight Transformers on the side? Little brown components on the bottom side or capacitors... And a huge amount of solder underneath, probably to take up all the current that's throwing instead of having wide paths...
Mike, that's a bummer but you should be able to replace the mosfets as I think they are the most likely cause of the short. They sit directly across the incoming 48VDC and then chop that at a high frequency to generate a high voltage DC rail, probably around the 400V mark, that 400V DC is then chopped to form a 230V AC line that gets filtered and sent out to the grid. That odd component you mentioned in the end of the video is just a simple jumper, they needed to get the gates of the mosfets back to the controller so they used a big jumper to do that I think. As fleaspotter already suggested I think you should be able to access the screws for the clamps that holds the mosfets to the heatsink by reaching between the transformers with a long screwdriver, I would start by removing the two mosfets you already identified as possible faulty and test those. They should be standard off-the-shelf components and I think you should be able to simply replace them as I don't think anything else would have died. Mosfets tends to fail closed and then blow a fuse or track, all your tracks looks fine so I think you should be fine just replacing the faulty mosfets. If you're really unlucky the mosfets would have taken the driver with them but I don't see why that should happen? I would replace both mosfets in that pair and take it from there.
Circuit breakers (and HRC fuses) are great as they break the connection so fast the failed components don't explode. Standard fuses are so slow to blow making a mess of failed components, that's why the green fuses were not blown. Sad to see him die :( It will most probably be the MOSFETs as you thought but as they are in parallel they will all be dead. That looks like a pain to work on as you will have to de-solder every component on the heat sink and remove it as a whole as the screws are hidden behind the transformers. Just hope the mosfet drivers are still good.
It seems that the screws are accessible between the transformers for each pair of MOSFETs. Check the video at 8:19, I think it's the best recorded position to see this.
You're right, you might be able to access the clamps for the mosfets through the gap between the transformers, that would make it a lot easier to replace the mosfets.
I noticed the brown skid mark on the inside of the case. What does this line up with? Electrolytic capacitors drying out are worth checking, granted your kit is not that old, but if they were getting heated that would age them prematurely. I think I would put the ohmmeter on the output, measuring in and work back down the circuit, perhaps desoldering legs of components until something changes.
You are going to have to get that heatsink off to get access to the MOSFETs. Once you have the heatsink out if the problem is not immediately obvious, then test the MOSFETs for a shorted condition.
Mike look at 3:52 you can clearly see a burn mark on the underside of the cover, line that up with the components below and you will narrow your choices... good luck!
If you look in the lid, I saw a dark mark which maybe dust but it also could be a plasma mark where a component has sputtered out of its case. This unit appears to have 8 identical inverters in parallel. The resistor R80 did look dark and maybe the mosfet it is connected to is short and it's pair on the same transformer. Because of the parallel nature of this unit I would guess that one of the 8 has gone short circuit and this would stop all the others from failing. Because you have 8 identical invberters this will enable you to check component values in the blown circuit with the still good ones. The brown caps on the rear are 470n 250V film caps. www.es.co.th/Schemetic/PDF/MET-683-400JP10.PDF
The brown components on the bottom of board are 47nF 250V capacitors. It appears that design uses two 48V to 400V DC to DC converters. One to give you +400V and the other to give you -400V. The +/-400V is then chopped to give you 240VAC. The each DC-DC converter has 4 identical channels in parallel. Each channel has a transformer and a pair of MOSFETs to switch the 48V. The eight 2 leaded packages on the back are diodes to rectify the DC-DC converter to provide the +/-400V rail. As the fault looks like a dead short on the input side so problem is likely to be one of the MOSFETS screwed to the big heatsink. This inverter has been designed to a price so is very hard to service. I have repaired a 300W grid tie solar inverter but would not recommend repairing this 2kW unit. Fault finding will be an absolute beast as you have to remove and test every one of the 16 MOSFETS. Best alternative is to scrap it and buy a properly designed inverter
If you look closely to the underside of the board you can see that those 400VDC rails are connected together so there is only one 400VDC rail and not two. You don't need two anyway to generate a 230VAC as that have a peak of 325V so can easily be chopped from a single rail . You don't need two to generate 230VAC rail. :-)
Thanks Peter for the details on how it works :) Have a watch of part 3, i just videoed tonight. Slowly getting there but the real question is will it work again.
I have a GTIL2. It does not start up. There is a blue LED (LED1) on the main board near U11 which blinks a code as an aid to troubleshooting. Does anybody have a copy of the table of "blink codes"? I have no idea how extensive it is. Maybe it would help here.
replace the fets like in the powerjack inverters. Use overvoltage protectors on AC and DC. 1000W inverter mean 500W real power on cheap china inverters. Remove all fluorescent tube from the house, they produce high voltage spikes!!!
We have all led lighting , but unlike a offgird inverter there is a small delay between when it senses the load and when it outputs more power to cover it. I guess that is a bonus of grid-tie.
Looking at the screen module from here at 3:56 it looks like the component at U3 has melted. Either that it's covered in solder mask. This screams of a blown FET though.
This is a small PCB and U3 is most likely a local voltage regulator for the display board, The size of the cables to that board would not carry 60A to cause his trips to fail. The reason it looks like it does is because of the conformal coating of the whole PCB after manufacture to stop the ingress of moisture and little bugs making it home.
No idea what to do, I doubt it will be this easy but maybe just try running it with a new fuse? I was just about to buy one of these, would you still reccomend it? Was also thinking of using the same charge controller setup as you are, with 2 panels for each charge controller & a 72v battery bank, would you suggest anything different? There are some alternative 72v boost controllers on ebay, search "72v mppt", one is black the other is blue.
