Zimnismoboy34 Posted October 28, 2019 Share Posted October 28, 2019 Good day all , Seems like i have been lucky and some what unlucky . I got given two axpert 3000va 48v units with 3Kw Solar mppts . One seemed to suffer from a Surge the Movs on the ac input 1 had blown right off. Then affected the mosfets input section . This inverter has 10 of them before the transformer (ir-fb52n15d) and as far as i can tell this is as far as the damage went . I have since replaced these and tried to power the unit off battery , The unit powers on beeps and waits a few seconds till i hear the Ac relay kick in at which point the unit blows the small fuse ive used inline further inspection proves that i have a short on the line of 10 fb52n15d mosfets and further checking finds that one is the cause .... as far as i can tell 5 on each bank of the input of the transformer . I then proceeded to replace the 1 faulty unit and powered the unit on again ... needless to say the same sequence of events took place and the 1 unit on the opposite side of the 5 was blown and in a dead short as well. The other unit presumably wasnt so lucky as this has damaged the mosfets fb52n15d as well as the 3 x FB3307Z Mosfets on the DC input . I proceeded to replace all including the IGBTs on the output incase as they were missing as well from the unit IR-GP4063D . Again the unit has followed the same sequence of events and damaged two of the Mosfets fb52n15d. So now as im uncertain which service manual to use i need to perform testing on the resistors that feed each mosfet incase one has shorted or blown open which ive tried to do with mosfets removed but the readings are not the same as the axpert 5kva 4000w or 5000w unit . If some one can assist me it would be a great help as im partly clueless but also keen to learn and understand how these units function, i have read a substantial amount about repairing them on the australian ev forum and know the coloumb would be the best advice i can get on here . ive tested the transformers that the mosfets feed into and get around 000.3 ohm across the legs on the input on both boards . Not sure if this is shorting the input section and straining the mosfets , I know there are diodes i should check as well on the IGBTS but im taking this one step at a time and hoping that once the input side is resolved we can move on as there is no service manual i know of for the 48v 3kva . Strangely the 3kva board of the 24v and 48v use the same components almost and are identical bar the fact that the 48v board has 3 input capacitors vs the 24v that has 4 . Which leads me to the next question .... could the input capacitors be faulty causing the mosfets to blow ? I thank you in advance and hope my explaination is sound enough to provide feed back on . Quote Link to comment Share on other sites More sharing options...
Coulomb Posted October 29, 2019 Share Posted October 29, 2019 (edited) 8 hours ago, Zimnismoboy34 said: Good day all , Seems like i have been lucky and some what unlucky . I got given two axpert 3000va 48v units with 3Kw Solar mppts . Let me guess, second prize was 6 blown up boards? Quote ... needless to say the same sequence of events took place and the 1 unit on the opposite side of the 5 was blown and in a dead short as well. When the MOSFETs blow, they usually end up a blob of highly hoped semiconductor; in other words, they often fail shorted, both drain to source, but often also drain and source to gate. When this happens, the gate drivers see all sorts of voltages that they are not designed to withstand, so they can fail. If they fail in such a way that the new replacement MOSFET turns on when it should not, then when its opposite number turns on, there is shoot-through, which blows them both up. Quote So now as im uncertain which service manual to use i need to perform testing on the resistors that feed each mosfet incase one has shorted or blown open which ive tried to do with mosfets removed but the readings are not the same as the axpert 5kva 4000w or 5000w unit . I vaguely recall a service manual for a 3 kW model, so scour this forums file section carefully: https://powerforum.co.za/files/category/1-files/ Quote If some one can assist me it would be a great help as im partly clueless but also keen to learn and understand how these units function, i have read a substantial amount about repairing them on the australian ev forum and know the coloumb would be the best advice i can get on here . Use the index to the AEVA PIP-5048 repair topic (now removed from the rest of the main topic, to keep it under 100 pages). There are posts specifically about repairing by yours truly, and some specifically about the gate drivers. As you've noticed, the 3 kVA and 5 kVA models have some similarities (and I have almost no experience with the smaller models). Quote ive tested the transformers that the mosfets feed into and get around 000.3 ohm across the legs on the input on both boards . In other words, it's reading the same as a dead short, which is to be expected. If it really had 0.3 Ω resistance, then at 70 A that would be 21 V drop, nearly the whole battery. Its actual resistance will be much lower than that.The transformer's mutual and leakage inductance will cause the current to be much less than you'd expect from just a resistance measurement. Quote ...as there is no service manual i know of for the 48v 3kva . I've just had a look at my collection, and no, I don't have one of those. Quote Strangely the 3kva board of the 24v and 48v use the same components almost and are identical bar the fact that the 48v board has 3 input capacitors vs the 24v that has 4 . Which leads me to the next question .... could the input capacitors be faulty causing the mosfets to blow ? Certainly, ageing capacitors can cause unavoidable glitches on the DC bus to become large enough to cause the MOSFETs to blow up. The capacitors provided are moderate but not great quality; all the inverters I own or sold on have had a MOSFET and capacitor upgrade, see the main AEVA index for details. I assume that you could use the same recommended long life capacitors. The battery-side capacitors and MOSFETs are the main weaknesses in the Axpert designs. The gate drivers often fail too, but almost always only after a failure of MOSFETs or IGBTs. Edited October 29, 2019 by Coulomb White space forum bug. Quote Link to comment Share on other sites More sharing options...
Zimnismoboy34 Posted October 30, 2019 Author Share Posted October 30, 2019 Hello and many thanks for your wise input . Ive had a look at the driver side of the mosfets and found that there seems to be some resistors that arent quite what they once used to be . The problem im facing is that the 3kva 48v doesnt use the same resistor divider on the gate pin between signal and gnd . Effectively i have a 330ohm resistor marked 33R0 which i know and can measure but the other resistor which has the marking 30c doesnt match up to both the 24v repair manual and the 48v repair manual . Assuming that my more consistent readings of these resistors which indicates a value of 19.86k ohms ... should be the correct reading on this resistor ? unless im mistaken ? Secondly the 25c2655l npn amplifier and 25a102ol arent available any more and there are two pairs of each on the 2 x 5 mosfet circuits . the closest i can find is this https://www.mouser.co.za/Search/Refine?FS=TRUE&N=18338129+4294689400+4294690470+4294690434+4294604607+4293990080+4294605144+4292906361 an im uncertain if the turn on low voltage is adequate as originally they used 0.5v .... Will await your advise and i do thank you in advance . Quote Link to comment Share on other sites More sharing options...
Coulomb Posted October 31, 2019 Share Posted October 31, 2019 (edited) 8 hours ago, Zimnismoboy34 said: Effectively i have a 330ohm resistor marked 33R0 Actually, the position of the R is like a decimal point. So that's a 33.0 Ω resistor, which makes sense as a series resistor (larger MOSFETs use 22 Ω, some use 10 Ω). If it measures 330 Ω, then it's faulty. Quote the other resistor which has the marking 30c doesnt match up to both the 24v repair manual and the 48v repair manual . Assuming that my more consistent readings of these resistors which indicates a value of 19.86k ohms ... should be the correct reading on this resistor ? unless im mistaken ? That's a different resistor code. From http://kiloohm.info/eia96-resistor/30C, 30C represents 20.0 kΩ. It's a way of getting precise resistor values (like those in the E96 series) to have a small enough code (three alphanumeric characters) that can be printed on most SMD resistors. This is a common value for a gate pull-down resistor. 19.86 is 0.7% less than 20.00. Quote Secondly the 25c2655l npn amplifier and 25a102ol arent available any more Ah. I suspect it's a 2SC2655; "2SC" is a common prefix for transistor part numbers. Here is a datasheet: https://www.mouser.com/datasheet/2/408/Toshiba_2sc2655-488363.pdf . It looks to be a lot stronger than the ones you found, and if they are in a "tall TO-92" package, they would be pretty strong (else they would have used something in a smaller and cheaper package, probably surface mount). My guess is that the other one is a 2SA1020 (note the letter-Oh rather than the zero at the end). It is presumably its PNP complement; indeed it says so on the datasheet for the 2SC2665. It seems to be obsolete, but you should be able to find a site that will suggest a replacement. Nearly any PNP transistor with similar specifications, designed for switching, and in a similarly sized package (mainly for heat dissipation) would probably work, but one that is designated as a complement to the 2SC2655 would be best (so there is more of a chance that the capacitance, speed, gain bands etc are similar). Edited October 31, 2019 by Coulomb Quote Link to comment Share on other sites More sharing options...
Zimnismoboy34 Posted October 31, 2019 Author Share Posted October 31, 2019 Well we making progress i know what to order ... thank you for that information . I wanted to confirm on the BJT transistors as they are pairs and seem to be scarce i am looking at alternatives but the gain bandwidth seems to be vastly different the original part is 100mhz and any newer components are either 50mhz or 200mhz . Yes original part was Gain Bandwidth Product fT: 100 MHz how would affect the switching ? Quote Link to comment Share on other sites More sharing options...
Coulomb Posted October 31, 2019 Share Posted October 31, 2019 3 hours ago, Zimnismoboy34 said: Gain Bandwidth Product fT: 100 MHz how would affect the switching ? Considering that the switching happens at some thirty-something kiloHertz... not that much. Seriously, the fT figure is important for radio frequencies. Switching involves square waves, and ideal square waves have infinite harmonics, but any modern transistor would I think be fast enough. The consequences of a "too slow" transistor in a gate driver application would be that the MOSFET turns on a little too late, or off a little too late. The former could lead to a slight reduction of efficiency; the latter could in theory lead to shoot through and disaster. But there would be dead time designed into the system, and individual transistors vary quite a lot in their actual characteristics, so there has to be enough safety in the design to accommodate that variation already. Short answer: 50 MHz or 200 MHz should be fine. Even 20 MHz would be fine. I would not give up too quickly on trying to find the exact correct devices (or alternatives that claim to be suitable replacements). ___ 1 Quote Link to comment Share on other sites More sharing options...
Zimnismoboy34 Posted November 7, 2019 Author Share Posted November 7, 2019 thank you for the assistance . Ive decided to try replace the drive circuit which after testing was totally blown . Explains why some mosfets were turning on and not others . I will replace all resistors and have tested the two TO transistors and they seem to be fine . Will test and revert hopefully no more bang ... Quote Link to comment Share on other sites More sharing options...
Coulomb Posted November 7, 2019 Share Posted November 7, 2019 Check PIP inverter repairs and hardware modifications to see if you can get more clues. The schematic traces might be helpful, but of course they are for a different model so use as a guide only. Quote Link to comment Share on other sites More sharing options...
Zimnismoboy34 Posted November 8, 2019 Author Share Posted November 8, 2019 right we are 1 up and working thankfully it wasnt as bad as the other and only the series of 10 mosfets on the input had issues with the drivers, after replacing all the faulty driver resistors and 1 x 25v 470uf capacitor near the power supply for the microcontroller i assume, ive run some tests and the unit is working and functioning as should be . charge from solar, and from ac with inverting from battery although have not load tested just yet this is where the real test will come in ... now on the other unit it was not so lucky probably suffered a reverse surge of power into the output . The IGBTS were all blown, and a quick check of the circuitry around the igbts reveals some of the driver resistors might not be so happy . The photo im attaching is from a 24V model i have just recieved that was dealt the same fate sadly, and guess what the driver circuit is IDENTICAL which i assume is all down to the boards both being 3000va but one is 24v and the other 48v . So at present i have 1 working unit 3000va 48v 1 igbt and input mosfets blown 48v 3000va unit including the IRFB3077 are also faulty and shorted across pins . and 1 x 24v 3000va unit again blown igbts and mosfets on input . My question is what are the diodes and what model nos should i replace them with as im certain they are faulty . and how would i test the 8 pin Through hole optocoupler which i assume it is ?driving each IGBT. Once again thank you for the patience on the forms helping members like myself and giving them the opportunity to get out of mischief and do some good ... lol Quote Link to comment Share on other sites More sharing options...
Coulomb Posted November 9, 2019 Share Posted November 9, 2019 12 hours ago, Zimnismoboy34 said: 1 up and working Well done. 12 hours ago, Zimnismoboy34 said: My question is what are the diodes and what model nos should i replace them with as im certain they are faulty . There are up to 5 diodes associated with each IGBT gate, as per my partial schematic trace (for a 48 V model, but it seems things re similar, except with a 100Ω in place of a 47Ω gate resistor). I note that perhaps the resistor R145 in the schematic trace should be 4R70 (i.e. 4.70Ω) instead of 47R (47.0Ω). I can see some diodes with the A2 marking code; those are the 1N4148W as per the schematic trace. The MELF (glass with copper inside) diode is probably the 18 V zener, MMSZ5248 if the marking code is J3). I say this because one looks like it has its anode connected to the emitter of the IGBT. 12 hours ago, Zimnismoboy34 said: and how would i test the 8 pin Through hole optocoupler which i assume it is ?driving each IGBT. I think you can only test the photodiode; it should test as a moderately low voltage LED (so a diode with perhaps 1.5-2.0 V forward bias). You could try putting a small power supply across the opto (pins 8 and 5, 8 positive), and see if applying ~5 mA to the opto LED causes the output to pull up. Yes, one opto for each IGBT gate. Quote Link to comment Share on other sites More sharing options...
Zimnismoboy34 Posted December 8, 2019 Author Share Posted December 8, 2019 Good day many many thanks for all the tips , using the sample 24v board i had i managed to get a reference and replaced a diode on the igbt driver side as well as checked all other components including the optos and managed to confirm all was well before firing it up ... much to my own amazement i have both units up and running .. i cannot thank you enough Quote Link to comment Share on other sites More sharing options...
Zimnismoboy34 Posted December 8, 2019 Author Share Posted December 8, 2019 ive since had a 5kva 4000w 48v come into my hands and it would appear that the board doesnt tie in with the schematic somewhat . This has the Q32 but no Q31 ? Based on this none of the igbts have any of there name plates on this side of the board, every single one is blown off . I want to give it a shot and repair it but cannot for the life of me work out what each should be ? QB2 QD2 QA1 QC1 Q32 D13 Q30 Q29 Q27 Q28 Q30 THROUGH TO Q28 ARE GP4063D D13 IS 1F19AP (RHRP1560) QB2 THROUGH TO Q32 NO MARKINGS from what i gather QB2 QD2 QA1 QC1 SHOULD ALL BE THE SAME WITH Q32 BEING DIFFERENT ? Please kindly assist if possible if any one can reference these parts of a known board it would really help . Quote Link to comment Share on other sites More sharing options...
Coulomb Posted December 9, 2019 Share Posted December 9, 2019 7 hours ago, Zimnismoboy34 said: This has the Q32 but no Q31 ? Yes, they dropped Q31 at some point. 7 hours ago, Zimnismoboy34 said: from what i gather QB2 QD2 QA1 QC1 SHOULD ALL BE THE SAME Yes 7 hours ago, Zimnismoboy34 said: WITH Q32 BEING DIFFERENT ? Possibly. There is a service manual for these in the file section of this forum. It has part numbers. Well done with the other repair. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.