 Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me. |
Forum Index : Electronics : Inverter # 4
  Page 4 of 6   |
|||||
| Author | Message | ||||
Tinker Guru  Joined: 07/11/2007 Location: AustraliaPosts: 1904 |
It has been a bit quiet here lately so I thought I'd amuse you with my exploits  .The idea was to find out why my inverter has spurious 'brownouts' when a very heavy load comes on. For this I used my test inverter (inverter #4). So I had it connected up and running with various loads when I noticed I did not have the IFB signal connected to my control board. The load was switched off and the inverter was idling when I attempted to plug in that wire. Bad idea    .This is what happened: Big bang, bright flash and fire. The most spectacular blow up yet . All of the 12 mosfets self destructed, the ones at the low side caught fire by the time I disconnected the power.  This close up shows that the screw terminal idea works well for a blow up but the terminal got too singed by the fire so three of them require replacing. That was the only solder job I had to do, the PCB had no damage and cleaned up nicely. Two of the totem pole transistors (TIP42C) also shortened on the low side but as these are on screw terminals are easy to replace. Here are the casualties:  Lesson learned: Do turn off the inverter before plugging anything in or out. Klaus |
||||
Madness Guru Joined: 08/10/2011 Location: AustraliaPosts: 2498 |
Maybe if you make a rectangular plate to cover the face of the MOSFETs you might be able to hold them together  There are only 10 types of people in the world: those who understand binary, and those who don't. |
||||
| johnmc Senior Member  Joined: 21/01/2011 Location: AustraliaPosts: 282 |
Thanks for the pictures of the latest pyrotechnics, been there many times , I now always shut down and short out the capacitors, before touching any part of the inverter. This is the only way that I have found to be completely fail safe for me.Then bridge the mcb to recharge the main capacitors.( stops welding the mcb contacts) then switch pin 6 of the inverter on to resume inverter operation . Cheers john johnmc |
||||
| Tinker Guru Joined: 07/11/2007 Location: AustraliaPosts: 1904 |
Brilliant idea Gary, it could add extra heat sinking too  .What I found interesting is that despite the mosfet legs were left at their full length they survived, the blow occurs inside the body. I suppose the mess left on the PCB with the short mosfet legs you solder guys have would be much worse .Klaus |
||||
| brucedownunder2 Guru Joined: 14/09/2005 Location: AustraliaPosts: 1548 |
Pardon my ignorance on this idea , would it be possible to insert a fuse of some sort that blows before any of this drama occurs?. Been watching Mark,s build and disappointment several times , scratching my head ,so had to ask. Thank you for the informative pics and description of the disaster.. Bruce Bushboy |
||||
renewableMark Guru Joined: 09/12/2017 Location: AustraliaPosts: 1678 |
That brought back some memories, gladly mine is humming along quietly and I'm going to leave it alone. Unusual how your one restarts from a big load, mine restarts around 5 am, with little or no load, only three times in 20 or so days it's been in service. The only thing that could possibly turn on at that time of day would be the fridge or the heater in my sons room. Sounds like an entirely different problem to mine. Is the wiring big enough to supply the load, maybe check all connections too. Cheers Caveman Mark Off grid eastern Melb |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
[quote]that the screw terminal idea works well for a blow up but the terminal got too singed by the fire so three of them require replacing.[/quote] It seems to be the exploding legs that burn the screw terminal blocks. If the legs are kept original full length, the mosfet body is a fair distance from the terminal block. Anyhow I had this idea of buying some fairly thick wall teflon high pressure tubing and slipping a short length over each mosfet leg. That should contain the fire and smoke and protect the terminal block and heatsink somewhat. The protective metal shield is a good idea too. So I ordered some 2mm bore 4mm od teflon tube from e-bay, but what arrived were five integrated circuits instead that I did not order. When stuff arrives from China its never certain which seller it came from, so I could not be sure which order had been screwed up. Somewhere in the world there is probably someone that ordered some real time clock timer chips, but instead received a small coil of clear tubing..... This is what I ordered: https://www.ebay.com.au/itm/1M-PTFE-Teflon-Tube-2mm-ID-4mm-OD-For-1-75mm-Filament-3D-Printer-RepRap-FT-/142771980470?var =&hash=item213ddf88b6 I have just ordered some more. Cheers, Tony. |
||||
| renewableMark Guru Joined: 09/12/2017 Location: AustraliaPosts: 1678 |
Klaus, just noticed that dc switch with the big red key (page 2), maybe try it without that, they are pretty cheap and suspect quality. Long shot but worth a try. Cheers Caveman Mark Off grid eastern Melb |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Switching the main dc power on and off is definitely a good idea, but it needs to be treated with caution because of the very large energy stored in the electrolytic capacitors. There really needs to be a "slow charge" button to bring up the capacitor voltage before the main switch is closed. And a "discharge" button to drain the capacitors before doing any work on the inverter. Cheers, Tony. |
||||
| Madness Guru Joined: 08/10/2011 Location: AustraliaPosts: 2498 |
If you are patient enough to wait about 30 seconds you don't need to force a discharge. "would it be possible to insert a fuse of some sort that blows before any of this drama occurs?." The power stored in the capacitors is more than sufficient to blow the MOSFETs to pieces. In my experience, a correctly sized breaker prevents any further damage including fires. There are only 10 types of people in the world: those who understand binary, and those who don't. |
||||
| renewableMark Guru Joined: 09/12/2017 Location: AustraliaPosts: 1678 |
Bruce, yes like Mad said, after changing to an 80A DC breaker no more damage, fets will go pop and you'll normally not even know till you do a resistance check to see why things aren't working. I initially had a bigger one as that's what Oz was using, and it would be fine if nothing goes wrong, but when something does then the only thing that will save the boards is that 80A breaker. Cheers Caveman Mark Off grid eastern Melb |
||||
wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 1156 |
I thought I might write some rambling thoughts I have been having that overlaps with Klaus's pyro display as a topic to discuss & invite comments. I have been considering how to configure my inverter to behave in a more user friendly manner, with regard to switch on and off and protection against catastrophic events. Ideally it should be simple enough for my wife to turn on with a simple on off switch. At first I envisaged 2 switches, 1 low power 1 high power (plus a safety say 100A DC fuse in series). Operation as follows: Connect tested Inverter to batteries & turn on low power switch. All 5V & 12V power supplies are established, but EG8010 power is held off, all FET drivers are inhibited (all FET's held off). Low Power LED is lit green. Main inverter LED is lit red and remains that way as 48V is now trickle charging the main inverter bulk capacitors, via the low power switch circuit. A voltage monitor circuit on the control board (a comparator or uC) is watching the voltage on the main capacitors. When the main storage capacitors have achieved >90+% of input DC voltage, the main inverter LED is now lit green (inviting you to close the high power switch). When the high power switch is closed it also enables 5V to the EG8010 & then the FET drivers are also enabled, starting the inverter. Shutting down is achieved by turning off the high current switch, then the low power switch. Instead of a fuse & high power DC switch, what about incorporating a suitable DC rated contactor that could double as a fuse & on off switch ? Operation now is a little more elegant at turn on. Close low power switch, contactor is initially open but closes when the bulk capacitors are at >90+%. A suitable 200A hall effect current sensor is available for ~ $USD8.50. If a serious overload is sensed the contactor can be commanded to open within 20mSecs. Note if output stage is dead, it is unlikely the main capacitors will charge up so the main contactor will not close. A second half of the comparator can also be testing that the centre of the bridges are at zero volts indicating an upper FET is also not shorted & again not allowing the contactor to turn on. A very suitably rated contactor, rated at 500A carry & 900VDC is available from Digikey for USD$75. It can interrupt 120VDC at 650A over 3,000 cycles. or 1000A for 500 cycles or 2000A for 20cycles. The contactor is a Kilovac LEV200 series P/No LEV200A5NAA. Power for the contactor coil needs ~ 15W to energise, but then only ~ 0.6W for holding current. The above price is for a 24VDC coil, for some reason a 48V coil unit is double the price! Note although a 48V @ 100A forklift fuse could be used, they still take up to 2 seconds to rupture when loaded to 400% of capacity (400A). At 100% capacity 12W is being wasted as heat - this compares to ~ 2W loss for the contactor. Lastly a small (retrofit?) circuit board and 8 pin micro controller circuit could provide the 24V energising & 5V holding currents (using simple PWM) required and read the hall sensor for overload protection. If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| Tinker Guru Joined: 07/11/2007 Location: AustraliaPosts: 1904 |
Thanks for all your suggestions guys but you all seem to miss the point. That inverter worked perfectly fine for my experiments , even if there is no high current DC circuit breaker used at the moment. So I repeat what the operator caused (he's getting old and forgetful ) problem was.No current sensing to the IFB input was connected when the inverter was turned on. Nevertheless it ran fine and powered my 2.1KW fan heater. That load was switched off before attempting to connect the IFB plug. Plug touching socket (its a 3 pole mini phono socket to match the current sensor plug) caused instant bang. No time for any fuse to blow, over current was *not* the problem. What I think is that pin 14 of the EG8010 and both pin 11 (SD) of the IR2110 chips did not like what I did. I do not know why as the SD line is normally kept low via the 1K resistor and is forced high from the CT output in case of an overload. Since there was no load connected the IFB sensing circuitry should have had no output anyway. But I won't try that trick again, have to wait now for a new batch of HY4008's to arrive anyway. That's no hardships I still have three other inverters here to play with. Today I found that 9 out of the 12 gate diodes also blew, I had used IN4148's there and they worked fine. Have not yet tested the gate drivers or the EG8010 outputs to see if they survived. Warp, The HY4008's had the full length 'legs'. Its the source leg that melted off at the mosfet body on some. It would have got hot enough for the plastic housing of the terminal to start smouldering where the 'leg' entered. I'm not sure if a teflon sleeve would help there as I think the 20A terminal rating was exceeded by a fair bit briefly. I will try mad's idea of a cover strip though. Looking at it a 20 x 3mm alu strip right across the 3 (or 6) mosfets would do a good clamping job and just, maybe, conduct a little of that heat away when the two tiny source wire connection inside the chip explodes. It an easy mod with lots of room to do it at my PCB. Klaus |
||||
| Madness Guru Joined: 08/10/2011 Location: AustraliaPosts: 2498 |
That MOSFET clamp was meant to be a joke Klaus. I think you would agree that once you have your Inverter working correctly they run fine unless you play with things. There are only 10 types of people in the world: those who understand binary, and those who don't. |
||||
| wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 1156 |
There was no intended criticism of your set up Klaus & I got your point about what triggered the bang. But I disagree about there being no overcurrent condition - that is surely what caused the FETs to explode. I believe a fast disconnect as I was proposing would not have made things worse, only better. I am sure you would agree that it was not a load over-current situation as there was no load. Yet uncontrolled (excessive) current flowed through the bridge FETs killing them and I can only see one of 3 likely conditions causing that. The FET's turned on for too long during an on period and the transformer/choke saturated probably within milliseconds and it went bang! The other condition (and I think most likely culprit) is cross conduction occurred where an upper FET & Lower FET on one leg turned on at the same time. The third is that during a cycle where the tranformer/choke was at its peak current point and the FETs were controlled to turn off at that instant (prematurely), the stored energy in the inductors/transformer caused some sort of an avalanche or caused FETs to "glitch" on. If cross conduction occurs it then only takes microseconds for the catastrophe to begin. In my humble opinion you plugging in the over-current connecter should not have killed the inverter but it did - the mechanism behind it is still a work in progress to determine why - I wish I still had a CRO with the = to trigger - great for capturing a one off event of an upper & lower FET being driven at the same time. My comments and suggestions are not intended to fix the underlying issue which causes this, but to minimise the mess (and maybe even avoid) when it happens. I read here recently where a breaker even welded shut extending the fireworks. My suggested contactor proposal only needs a small on off switch to turn on the inverter with hopefully a minimum of fuss, it could also trip out and restart automatically - assuming the circuitry is not dead. From what I have read here, turning one of these things on reminds me a bit of the movie scene where Danny DeVito is the poor sucker who has to start the Mobsters car (in case it is wired with a bomb) and turning the key is like playing russian roulette - it doesnt always go bang  I also have 2 questions please, the 12V supply for the upper FET drivers (Tip41/42) can you confirm that you are not using an isolated Dc/Dc converter and that it is only fed from the D1/D2 charge pump circuits on the control board ? The second is why does no-one use the 4 signal transistors on the SPWM outputs of the EG8010 (as per the EGS002) to guard against cross conduction. If the designers of these things don't trust their own code to never cross conduct maybe its a signal that we shouldn't either ? If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| renewableMark Guru Joined: 09/12/2017 Location: AustraliaPosts: 1678 |
Probably overthinking it Wiseguy, I had problems because of stuff ups, if everything was built right, none of that would have happened, if Klaus wasn't fiddling with his machine it wouldn't have happened either. Once they are built and working they seem to be totally fine if you just leave them alone, apart from the slightly annoying self soft start issue. Cheers Caveman Mark Off grid eastern Melb |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Its actually not a bad idea Mad. Anything that reduces the extent of heat and smoke damage has to be a step in the right direction, even if it does look a bit weird. The teflon sleeve idea should contain the heat and cause the legs to blow faster as well as containing the exploding mess. Cheers, Tony. |
||||
| Madness Guru Joined: 08/10/2011 Location: AustraliaPosts: 2498 |
I think what Klaus, myself and perhaps others that I won't name need is an LED sign that turns on when there is power in the caps in bright red letters that says... HANDS OFF There are only 10 types of people in the world: those who understand binary, and those who don't. |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
You could always assemble the whole thing with tamper proof screws. Cheers, Tony. |
||||
| Tinker Guru Joined: 07/11/2007 Location: AustraliaPosts: 1904 |
Warpspeed - the case uses socket head screws , even tamperproof would be no deterrent for me .The whole reason I built this little inverter is for *tinkering* with it, fully accepting there would be the odd bang and me then posting here why, so others can avoid that happening. Fair enough? Madness - Yes, I got the joke but thinking about it a little its actually not a bad idea. I made the top clamping bars today .Wiseguy - We have progressed past the EGS002 solution here. If you like to play with that, have fun. But I will not try to complicate an inverter that runs very well 99% of the time by adding yet more components to go wrong. You can look up the schematics of my inverter control and power board on an earlier page here. Look at the latest posted schematics, some earlier ones have been modified since. I am very pleased about the physical attachment of the mosfets in my inverter. This certainly makes their replacement soooo much easier than you lot of the solder them in brigade. The charring of 3 of the 12 terminals I did not foresee and I will try warp's teflon sleeve idea since I have some here. The PCB has no charring/ burn marks whatsoever. Further testing today showed that the EG8010 and the drivers run fine, no damage there. Klaus |
||||
| Page 4 of 6 |
|||||
 Print this page |
