 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 5 of 6   |
|||||
| Author | Message | ||||
| Warpspeed Guru  Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
[quote]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.[/quote] Klaus, it seems we are constantly thinking along similar lines, so I will lay another strange idea on you. I really admire your floor standing "big box" approach to housing your inverter and have been thinking about how to make something like that myself. I don't have access to any large sheet metal equipment such as a decent guillotine or large pan brake, so decided to cheat... What I am putting together at the moment consists of a 6mm flat steel plate bolted on top of four industrial strength castors, with all the heavy magnetics simply bolted on top of that flat steel plate. Its simple and very rigid ! My 5Kw inverter requires four output transformers of different sizes so its a very simple way to support the 90 odd Kg of all that combined steel and copper, and get it mobile. On top of that sits just the outer shell of a two drawer filing cabinet, with the top hinged as a lid. What was the original open front now becomes the open rear. So the outer metalwork is already fully made up and complete, and even painted. All I had to do to it was fit a piano hinge across the back for the lid. Now the sneaky part is that the inverter heatsink and all the electronic guts of the inverter are attached only to the hinged lid. With the lid down the heatsink is vertical for best convective airflow (without the fan running). But with the lid hinged in the up position, everything rises up right out clear of the surrounding metal enclosure, to work bench height, and is all fully exposed for very easy access to all the parts. A pair of gas struts take all the 10Kg weight of the heatsink and electronic guts, and holds the lid in the open position. In the open position the heatsink and circuit boards are horizontal and very easy to see and access. It has been dead simple to make. The only tricky part is getting the mounting positions for the gas struts right so the struts work with the right amount of force as the lid angle changes. That requires some trial and error but it has not been as difficult as I thought it might be. Cheers, Tony. |
||||
Tinker Guru Joined: 07/11/2007 Location: AustraliaPosts: 1904 |
Tony, I'm looking forward to see pictures of that creation. 6mm steel plate is a bit on the heavy side but if its what you had at hand... A tip. I just opened up inverter #3 (3KW version) to trial some brownout cures. Anyway, while that wheeled inverter cabinet is fine for use of the inverter, its a bit too close to the floor for my old bones to work on it. So I wheel it under the chain hoist and hoist it up onto a small wheeled bench so its at a comfy chair height to inspect and work the innards. I use slings to lift it as I already had some. If your cabinet is strong enough you could fit strong handles on top for lifting but with 90kg slings might be a better idea. BTW, I do have a small (600mm) manual guillotine/ V groove folder/ roll bender combo but it maxes out at 1mm thick material. The inverter case uses 1.6mm alu sheet for stiffness. I had that made for beer money by a local alu fabricator. Klaus |
||||
wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 1156 |
Maybe I am not good at communicating my thoughts so I will clarify a couple of points. I have no desire to build a unit which will go bang one day for no apparent reason. Recently at least Tinker, Poida, Mad & John have all had failures which so far the root causes are unexplained - satisfactorily to my technical mind at least. I'm sorry but I consider there are bugs lurking or and the design is marginal somewhere - I still suspect the EG8010 itself or something to do with its implementation. My final question in my last post was not regarding or promoting the use of the EGS002 it was regarding the fact that the designers of the EG8010 IC apparently do not trust their own software code to not present cross conduction, as they saw fit to connect 4 transistors to guard against just that event on their EGS002 implementation. You may have moved on from the EGS002 but you left behind some safeguards in doing so perhaps (the 4 transistors)? Sometimes extra bits improve things. Despite any criticisms and an apparent lack of interest in my thoughts of how I might improve the design I will continue in my quest unperturbed to develop a reliable inverter with a friendly user interface. The first stage is completing a thorough test of the EG8010 under all stimuli and various conditions. I will eventually report back with my findings which may or not support its current implementation. There is no question that there is a lot of talent here, the 4+kW inverters when working are efficient run cool the basics are fine. The information that will help me wind my first toroid etc are all priceless - thanks to all for the help & information provided to date. If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| Tinker Guru Joined: 07/11/2007 Location: AustraliaPosts: 1904 |
Ah Wiseguy, I see you need to play with the EGS002 boards for a while longer yet before you might come to the conclusion that, yes, they do in fact blow up just the same, 4 extra transistors or not. I will encourage you to build your own version of control board, perhaps in time we can combine the best features of all versions posted here and come up with a truly failsafe arrangement. You might agree that we have got the power side pretty right by now, it certainly handles what it is asked to do. I'm on the trail of the odd auto soft restart problem, as soon as I can nail down a definite reason for it you can read about it here. Its not a destructive event, just annoying not knowing why it happens. My methods differ from yours, I experiment with the complete inverter not just its brain - that's too deep for my limited knowledge. Having several home brew inverters here does help to compare, and improve as the next one is built. If you concentrate all your efforts just on one version, you might end up on a long way to a dead end. Just getting my power board working reliably took many trials and re designs. And my control board is now at version #4.... As you can see from poida's post, the complete bought Chinese inverter with the EGS002 drive board also has problems.... Klaus |
||||
| wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 1156 |
I agree that they blow up but I cant concur yet whether its for the same reason. I don't "need" to play with the EGS002 it just happened to arrive before the adaptor boards for mounting an eg8010 onto. I am using it as a convenient holder for the 8010 while I torture it for further information with my CRO. As an aside, the EGS002 uses 10uF capacitors for the bootstrap supply driving the upper gates of the bridge FETs to on. When multiple FETs are connected in parallel each with 5K resistors to ground and we are also fighting the "miller effect" of the FETs input energy requirements, that seems a bit light on. The 100uF capacitors in your (Mads?) implementation of the buffers driven by the IR2110's feels a lot better - this is subjective though as I havent got to modelling the output stages as yet. PS I eluded to the fact the Power part of the inverter appears to work a treat, using Warps choke design and tuning information and saturation tester is all excellent. But the inverter definitely needs a bulletproof good brain !! If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Mike, I don't think anyone here has ever had one of these inverters work at the very first attempt. It takes heroic courage and perseverance to work through the many possible difficulties and failures that can occur. There are all kinds of potential problems to overcome, such as sub standard counterfeit Chinese parts, blind faith in the functioning of completely unknown software of possibly highly questionable origin. Problems with the physical layout that can introduce noise and destructive transients. And lastly the pitfalls of home brewing transformers and chokes from secondhand parts, as well as workmanship defects introduced by faulty soldering and crimping. These inverters are definitely not projects for beginners. Its only really the vast help and support from other forum members that makes such a complex project even possible. Klaus, The inverter metalwork is all in many pieces at the moment, waiting for the fourth smallest transformer to be wound, so I can drill the mounting holes in the base plate before painting and final assembly. The 450mm x 600mm steel plate adds only 15Kg to the 90Kg mass of the four bare transformers. Its simple to make, just requires some tapped holes, and very strong. |
||||
Madness Guru Joined: 08/10/2011 Location: AustraliaPosts: 2498 |
Neither Tinker or myself rely on the IR2110's to drive the MOSFET gates when using larger numbers of MOSFETs. Before using Totem pole drivers to do the heavy work of driving the gates we both had loads of problems. Since I started and subsequently Tinker started using the TIP41c & TIP42c transistors there have been no unexplained blowups. (Tinkers recent failure was due to connecting the IFB while it was running which I consider to be an explained failure) Since I have started using the Totem poles I have not had any failures of any component on the control board. Any problems I have had since then have been caused by things like a GTI failing due to an earth leakage in my regulator. That caused the GTI to fail and feed 450V DC back into the Inverter Toroid from 5 KW of PV panels. A lot of time, blood, sweat and tears have gone into getting these Inverters working reliably by Oztules, Tinker and myself. If I was starting out fresh I would be using that as a starting point as we have proven that it is possible to build reliable Inverters that will just keep working. I can assemble my PCBs and they just work, no smoke or shell shock, just put it together with due care and attention to soldering and turn it on. I have had one running nonstop for 8 months power my house before I had to shut it down to do some changes to my switchboard. If improvements can be made from there then great but personally I think trying to start out from scratch is pure masochism. As for adding a bullet proof brain I fail to see the point of why it is needed. We already have overload shut down, this is far faster than any relay type shut down would be on the DC input, the only way that could possibly work is between the capacitors and the MOSFETs and then it would need to handle the very large instantaneous currents. Also I have added brain to mine which will shut down if there something like the voltage is too low (which will kill the Inverter) plus I have added thermal and other triggers to turn off the Inverter safely. 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 |
I'm not sure I do - I am seriously interested to hear the explanation (or theory) of the chain of events from that which actually caused the demise of the output FETs Ok I dont need further explanation for that..... That does help with my confidence level - a lot & that there is much more right stuff here than wrong - I'm not hanging around this forum for no reason. (and the reason also isn't to annoy people) Your overload shut down doesn't stop the battery current into dead FETs - I do not intend to dispense with fast normal overload shutdown. Why cant the relay be before the capacitors ? - the total damage would be limited by the capacitors stored energy which will dissipate very quickly into dead/dying FETs. I proposed to remove the fault current from the battery ASAP to limit the pyro display & flames. The contactor I suggested a couple of posts back easily breaks 2000A @ 900V DC within 20mSecs for around $100 (ridiculously cheap for what it is). Nothing wrong with belts and braces - "more stuff to go wrong" can also be more stuff to make things better If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
The IR2110 chips were only ever designed to dive a single pair of fairly small mosfets (one upper, one lower mosfet) located right next to the IR2100 on the same circuit board. These I.R. isolated gate driver chips are just not capable of driving multiple very large geometry mosfets directly, and expecting that to also work through a long gate drive cable extension is just wishful thinking. Just as Mad says, the correct solution is to fit high power gate drivers as close as possible to the power mosfets, and preferably one high power gate driver per mosfet. The TIP41/TIP42 solution is excellent, but I personally prefer TC4452 gate driver chips (or similar) mounted into good quality machined contact IC sockets. Either solution will work fine, its just personal preference. There is also much to be gained by proper design of the output transformer and choke. As a retired power electronics design engineer, I have never been a great fan of using software to drive very high powered inverters directly, and many other old and grey battle scarred hardware engineers I know, hold similar fears. Microcontrollers are great as an add on for supervisory, measurement, and logging features no argument about that. All the features, bells and whistles, great stuff. There are too many problems at initial startup and final shutdown where various power supply rails can rise and fall independently, and trusting fast enough software protection is tempting fate a bit. Using software to do the whole thing is just not safe. I have worked with some very clever software people over the years, but they tend to live in a utopian world where there is never any noise, the +5v rail never varies, and everything is totally predictable. When the software very occasionally and unpredictably crashes, they blame the hardware, and they are totally correct in that assumption. The problem with the very occasional software crash is that in an inverter it can cause catastrophic damage, even start a fire. It can be done, but it needs a lot of thought and cooperation between the software guys and the hardware designers. Cheers, Tony. |
||||
| wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 1156 |
Tony, I concur completely with everything you wrote. I don't rely on rails etc in important stuff - I put monitors for under & overvoltage and staged startups etc. My only extra I am considering is to put opto-drivers before the TC4452's all close to the output stage to completely isolate sneak current paths back to the control board. I also intend to have isolated DC supplies for the upper mosfet drives - never been a fan of bootstraps with large gate loads. What is your theory for the occasional dropout restart of the inverters in the early morning hours for unexplained as yet reasons ? If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| Madness Guru Joined: 08/10/2011 Location: AustraliaPosts: 2498 |
I have an 80A DC breaker that does stop any further damage very effectively, so if something goes wrong the FETs short and don't explode, there is no fire, no burnt tracks just replace the FETs and it is good to go. I see no advantage to adding some complicated and much more expensive relay. Nothing that is before the capacitors will prevent damage as the energy stored in the capacitors is more than enough to kill the MOSFETs. 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 |
The advantage was two fold, it is much more robust than the 80A breaker, allows for variable trip level and would allow for auto restart if the output stages are still ok. We intend to be away for perhaps a week or two at a time - if I have confidence in the unit I would like to leave it turned on. If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| gaspo Regular Member  Joined: 25/06/2018 Location: AustraliaPosts: 65 |
I have a question regarding the use of TIP41/42. In Tinker's schematic he is using 22R resistor connected to the bases of the transistors. At 12V that's about 0.5A. Wouldn't be lower output current more beneficial for the IR2110? The TIPs have hfe of at least 15 (I measured mine at > 100), so lowering the base current to 100mA they could still push > 1.5A to the mosfet gates. Could Darlington transistors (ie. BD681/682) be used instead to lower the IR2110 output current even more? |
||||
| wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 1156 |
The 0.5A base current would only flow for ~100nSecs whilst the gate capacitance charges. The IR2110 devices are rated to 2A & driving capacitance loads so that is only 25% of max any way - I don't see a problem there. If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Don't know about the drop out problem because my own inverter does not use PWM or any of those Chinese driver boards or Chinese controller chips or Chinese software. My own inverter is totally different in concept, but I still had to overcome many of the same kinds of basic problems others here are still being plagued with. My own philosophy is to build a self contained driver board that has all the required inputs and outputs, with its own ground system completely separate from the high power bridge, and only connecting to the -ve side of the main dc battery input at one specific point, which should also be the metalwork/chassis grounding point. That should keep the worst of the noise and any massive fault currents or transients right out of the control system. The second golden rule is to totally isolate all the mosfet gate drivers with opto isolated drivers that also have a built in undervoltage shutdown for each opto driver. My own personal favorite is the HCPL3120. These have an 11v shutdown, so need a +15v dc supply to work properly. I came up with a novel system for connecting the upper and lower opto isolators in inverse parallel to positively protect from cross conduction, even if the driver board or software goes completely spastic. By adding a single optional capacitor across the opto LEDs it will also add some extra fool proof dead time.  This is a very simple circuit, but its been well tested and is absolutely bullet proof to cross conduction failure due to driver misbehavior. Its impossible for both opto isolators to be on together, no matter what! And the transition always has to pass through ground below the LED threshold voltage where both opto isolators will be off. So there will always be some small dead time, which can be extended by adding capacitance after the LED current limiting resistors. Third golden rule power each gate driver individually from a fully isolated dc power source. Do not be tempted to use a common ground referenced dc supply for all the lower mosfet gate drivers. Make every gate driver identical and fully floating. Full dc isolation is essential for the upper drivers anyway. Make the lower drivers identical to the upper drivers. That will go a long way to eliminating noise in the gate circuits, and any cascading failures, where the +48v gets back onto any common +12v rail which could cause massive damage to spread throughout the whole system. If one bunch of mosfets blow up, the fault cannot escape back into the rest of the system, any damage will be localised. Preventing blow ups in the first place is only half of it. The other half is preventing the resulting damage from spreading further. A bit of thought and planning as to possible failure modes and the consequences (otherwise known as extreme fear and paranoia) may involve a few extra parts, but it can save a lot of tears later on if it does spit the dummy. Cheers, Tony. |
||||
| Madness Guru Joined: 08/10/2011 Location: AustraliaPosts: 2498 |
If the 80A breaker is tripped your not restarting without repairing the fault, that I can guarantee. We already have a variable setting for overload on the AC side. What intend to do is have a transfer switch that would change to a second Inverter if in a rare case there was a failure, I have not thought it through fully yet as to how to safeguard against a situation that would cause the second Inverter to fail. I had a faulty oven that shorted out the AC if liquid got to the bottom element, when this happened it just tripped the 20A AC breaker. Other things that I have done such as attempting to start a 5KW AC Compressor while in a stalled state also just tripped the AC breaker. Nothing else was effected apart from that circuit dropping out. As far as I am concerned the Inverter is very robust and reliable, it is only doing things that should not be done that will kill one of these Inverters when running correctly. Having said that though electronic components can fail due to many reasons, so it is impossible to guarantee 100% uptime no matter what. Situations like Tinker connecting the IFB (sorry Klaus) don't happen when the Inverter is running unattended so there is no reason to waste any time considering that type of case any further. The template is there to build one of these inverters that work, I can supply PCBs if you want to go down the same path as me or make your own. Once you have built one and have had it working for a while it may change your thoughts on needing to find a way to make it bomb proof. 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 |
Warp - You forgot to mention Chinese re-labelled parts ..... I gather you're not much of fan of most Chinese things.... Mad - I consider the "so there is no reason to waste any time considering that type of case any further" is a lost opportunity. I know we are dealing with immense amounts of power and the precision's etc needed to do it successfully. It is my inquiring mind - if we know why, we may be able to build the right safeguard. In my career I found it easy to build stuff to go bang, it was much harder but much more satisfying (& profitable) to find out why and fix it or re-design it not to. If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| Madness Guru Joined: 08/10/2011 Location: AustraliaPosts: 2498 |
Obviously, you have plenty of spare time your hands and a touch of Masochism, I have other things to do. If it fails due to my own stupidity then I should learn from it and don't do it again. Sorry but everytime I see your name I think of this. 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 |
I was well aware of its double meaning when I coined it - I grew up with that humour, no need to apologise - I invited it. I wasnt trying to imply the other "wise' part - that's still a work in progress. Dont worry I have been insulted by experts. Thanks for the laugh. Now what can I find about madness....haha If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
I still buy most of my bits from China, and I have also been stung fairly frequently. But most of it works, and the risk is worth the very large savings. But much of it is a cheap knock off, and sold as something it is not, so "caveat emptor". For very simple parts that can easily be tested and verified, Chinese is fine. But for more complex things like magnetic materials or software, I tend to be a bit more wary if failure can be dangerous or expensive. I also don't like things that have no published specifications or a circuit diagram. Cheers, Tony. |
||||
| Page 5 of 6 |
|||||
 Print this page |
