 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 : Random Unexplained Inverter Failures
  Page 6 of 11   |
|||||
| Author | Message | ||||
| Warpspeed Guru  Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
I never had any blowups, but then I only built and tested this EGS002 inverter design very briefly. My main day to day running inverter is still a very slow switching multistep design, with a much larger multistep design to very shortly replace the original. I feel in my bones that you are quite correct in your hypothesis about slow switching saving many of the currently working PWM inverters. I also have some suspicions that there may be some different EG chip and software versions in play here, some of which have the shoot through problem and others that may not. Its very difficult to convince people of these effects, as few Forum members have a digital storage oscilloscope good enough to catch these once only and very infrequent events. I have never seen this occasional shoot through effect, but then I prefer to use my analog oscilloscope for day to day use which would not have picked it up anyway. My digital storage CRO only ever comes out for very special occasions, for very slow motion events, or to capture very fast once only transient happenings such as this. I really don't like my ancient slow sampling rate digital oscilloscope it too frequently aliases and tells me lies. One day I will get myself a really fast sampling digital, but right now something really worth getting is way out of my budget. Cheers, Tony. |
||||
Clockmanfr Guru Joined: 23/10/2015 Location: FrancePosts: 429 |
Gaspo, wiseguy, For clarity, and for others following this thread but may not fully understand. Regards whats on the 8010 sub board pin out, here is the OzControl board rev12bb. This board originates from Oztules original simple design layout 2015/16 period. Since then there may be changes around the peripheries of the board, but the 8010 sub board pin out has not really changed. I believe there are a couple of hundred boards out there with this pin layout. For clarity...... (Photos are limited to 1.5m)    Everything is possible, just give me time. 3 HughP's 3.7m Wind T's (14 years). 5kW PV on 3 Trackers, (10 yrs). 21kW PV AC coupled SH GTI's. OzInverter created Grid. 1300ah 48v. |
||||
| gaspo Regular Member  Joined: 25/06/2018 Location: AustraliaPosts: 65 |
Wiseguy, what I intend to do was not put permanent tracks from pins 10,11 on 8010 to connectors but to place jumper pads on the board. So instead of soldering wire links from 8010 pins to connector one would just solder jumper pads together with a drop of tin. A lot simpler then to mess around with soldering tiny wires. |
||||
wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 1156 |
Gaspo I was responding to what you wrote which was: Tinker, I should be able to connect tracks from the crystal pins on the 8010 to the corresponding connector pins. This sounded permanent, If it is now planned to be done with solder pads and not a permanent link I think it is a good solution - keep up the good work! If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| gaspo Regular Member Joined: 25/06/2018 Location: AustraliaPosts: 65 |
Wiseguy, yes the jumper pads will be used, sorry for the confusion. |
||||
| theoldwizard1 Newbie  Joined: 25/09/2018 Location: United StatesPosts: 5 |
Newbie - late to the discussion ... You can now buy almost complete inverters based on the EGD002 board from China for under $100US. Add a transformer, housing and fan and a few other items and you are done. What is "different" (?) about their design is that it use a LARGE 50Hz/60Hz to boost the output to "mains" voltage. Almost all of the previously designed inexpensive inverters on eBay that never seem to last very long take Vbatt and run it through a SMPS and boost the voltage up to at least peak-to-peak mains voltages and then pass that on to the final MOSFETs. What is the theory about why big transformer seems to be a more "robust" design ? |
||||
| wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 1156 |
Welcome to the forum. You have a lot of catch up reading to do. Briefly, why is the 50/60 transformer version more robust? It has tremendous surge ability. It has relatively safe & simple low voltage circuitry and parts for the majority. Voltage conversion and the PWM sinewave modulation is achieved in one stage. Creating a 4KW 48V to 340V SMPS Dc to Dc converter front end that can cope with 100/200% overload for minutes is not something back-yarders really want to get into - and neither do I. Is there an EGD002? The EGS002 has some issues and most here do not like them or trust them. The whole concept of this forum is to keep it simple, without being too simple, and be able to reproduce them with reasonable cost & reliability & resources & a modest skill set. If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| Clockmanfr Guru Joined: 23/10/2015 Location: FrancePosts: 429 |
Hi theoldwizard1, You ask a question that I answer a lot, so here is the History. 'wiseguy' has answered a bit of your question and on other RE forums around the World you will find other answers from the likes of 'Oztules' and 'Warpspeed' our delightful Professor on toroids. In truth the original concept of the OzInverter was to have a real Inverter that could from 48vdc give me and 'Oztules' and a few others. a real 6kW 240vac running capacity, and stand surges up in the 50kW regions, and adequately power a large normal domestic dwelling. ie, we wanted a real Off Grid Inverter that was Simple, Robust, and very importantly Very Cost Effective. In 2015 .... I had purchased a very expensive SMA Inverter, it was rubbish, it would only do half its rated value running continually before it would overheat, and only 10 minutes at its so called full rated value. It certainly could not handle backcharging, AC coupling and DC coupling all at once. All SMA could say was that I needed lots of their expensive supplementary equipment to get my RE generation to work properly. The SMA machine went back to Germany and I got my money back. But I had to use the EU Trading Standards, stating that SMA do not inform the purchaser that SMA ancillary equipment is necessary. So with 'Oztules' help we moved forward with getting an Inverter that handle Backcharging through it,..... AC Coupling(Using GTI's to feed back into the OzInverter created mini grid),...... DC Coupling, (the batteries can run the OzInverter and also be charged by a DC charging source, I have 3off Hugh Piggot design 3.7m 12footers dia, Wind Turbines) and very importantly have a Low running/woking power loss, ie, 35watts. I have had 4kW EI type Inverters that have a loss of nearly 200 watts!!!! that's a waste of your batteries at nightime. At first the Power Jack boards were used, and with a decent size and especially wound toroid it would work very well. But the Chinese boards were bad on QC, and getting them became an issue. With the P J boards on the OzInverter when AC coupled, the AC voltage would rise when the batteries were reaching full capacity and pushing back, and this knocks off the GTI's because of the GTI internal settings. The 8010....... I had noticed the EGS002 board but the response's from folk were not good. Yes it worked well on normal EI transformers, but only up to about 3kW with toroid transformers. Some preliminary stuff here...... http://www.anotherpower.com/board/index.php?PHPSESSID=h7jd5l4edljl6k4ls18pgfj4i1&topic=1116.0 'Oztules' further investigated, and Like me we really wanted to get away from relying on a particular manufacturers products. The 8010 chip was removed from the EGS002 board and the OzInverter evolved with mostly 'Oztules' endeavours. “Looking good.. now with just off the shelf components, we can finally make a real inverter.. that is easy to fix as well. utopia indeed for remote living I don't think any other unit on the market can say that.... usually you have to replace propriety boards... if you can get them. . ‘oztules.’ Now the OzInverter original design has evolved with folk like 'Madness', 'Tinker' and others going their own particular way with their inverters using the 8010 chip. 'Oztules' and then Madness have added a GTI control circuit as the 8010 chip holds a rock steady constant AC output whatever the AC load is doing. Because of this rock steady AC, a normal GTI without control will backcharge to the batteries and over charge them, and that's not good.! Madness now has some of his MadInverter boards that have a lot of ancillary control and data output visual stuff using MEGA's and his specialized written software. I still use a simple Voltage comparators to shut down the GTI's as the battery voltage rises. I have documented the OzInverters progress and have published a book, on 'How to make a 6kW OzInverter' at present I am doing a second edition that will fully explain and document the up-todate progress. And with permission from 'wiseguys' and 'gaspo's' also add in their latest findings and works. We are dealing with an output of at least 6kW with this OzInverter, so its important that certain protocols must be adhered to. Its Power Electrics, its Power Electronics, its Mechanical Engineering, so any poor quality build skills, straying or shortcuts from the build road will lead to the OzInverter failing to work correctly. OzInverter Mantra ...... KEEP IT SIMPLE, MAKE IT ROBUST, KEEP IT COST EFFECTIVE. Everything is possible, just give me time. 3 HughP's 3.7m Wind T's (14 years). 5kW PV on 3 Trackers, (10 yrs). 21kW PV AC coupled SH GTI's. OzInverter created Grid. 1300ah 48v. |
||||
herbnz Senior Member  Joined: 18/02/2007 Location: New ZealandPosts: 258 |
I have had a brief look at the building of inverters and would like to address the failure issues from a different direction. Years ago in my work I spent a lot of time researching the effects of switching off AC circuits. This subject is not well documented but without going in to great detail very high DC levels can be reached with high frequency AC superimposed depending on the amount of reactive components, time in cycle the switching occurs and the load present. From what I see the common failure seems to occur after a switching event. I suspect depending on the part of the cycle the switching occurs the fets can not handle the high voltage. This of course gives a random nature to the failures and is more likely to occur when Max energy is stored IE no load. The switching can be the protection circuits or simply the on off switch. I have always chosen inverters that have a current limit feature rather than straight shutdown or as in one case of a cheap Chinese unit I found the output voltage control and feed a feedback to it from a CT on the output. After 20 years of living off grid and abusing inverter I have touch wood never blown one. My feelings here are a lot of failures would be avoided if all shutdowns were by first winding down the voltage then shut off. I see the trend is to give soft start ups but we need more importantly soft shutdowns. I do want to experiment with building a inverter but neither have the time or facilities I have bought in one of the control boards and also one of the built up units referred to in a previous message but so far have not livened up. On writing this I have a thought, instead of a on off switch use a volume type pot with a switch included. Feed a control back from this to the over ride the voltage setting. I have not got time to play but maybe someone out there may. I must also congratulate Tony on sticking in there,as you no doubt recall I got into serious trouble here many years ago resulting in my with drawing from the forum. I was surprised I could still log on. |
||||
| wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 1156 |
Clockman I consider this an open forum - you are welcome to use any of my contributions circuits PCB's ideas etc. My permission is not required but I appreciate the gesture. You are also welcome to mention my name with whatever you use - but more from the point of view that if it causes any unintended issues or grief the right person gets blamed for it. I am currently still investigating the EG8010 at power up, especially after the reset (crowbar) switch is pressed. This results in all the SPWM outputs going high simultaneously for a couple of milliseconds after the switch is released but before the IC control takes over. The anti cross conduction transistors should fix that up as well. Some odd things also happen on one of the SPWM outputs, which is suppressed for ~5 of the 50Hz cycles and then appears normal again after that. I am not convinced it is to do with a high side pre-charge but will get to the bottom of it soon. I am also looking into the simplest way to incorporate a simple staged startup (safety) regime that will not enable the FET drivers until a short time after the start up sequence has finished (all supply rails settled and correct etc) & then release the soft start. This is for my implementation of my control board. I plan to use some small DC/DC converters for 12V, 5V and isolated supplies for the FET drivers. These will have different start up times to the current control board so without the right safeguards I would probably be inviting trouble for myself. Herb, From what I have read here inverters have had the main 20A+ circuit breakers trip repeatedly without disaster from overloads. I do not dispute though the harsh nature of a sudden disconnect to the Power stage, but I don't intend to pursue this further at this time. I do see room for improvement with the start up & restart regimes and the odd failure that occurs then with either no, low or little load. Most failures by my reckoning do not seem to be overload related but more disturbance related. If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Suddenly switching off an ac circuit, particularly an inductive circuit can certainly produce enough voltage to spectacularly destroy things. The way around that is to clamp the voltage spike somehow to some safe maximum. With a bridge type of inverter, we can place inverse diodes across the switching devices. In fact mosfets already always have these diodes as part of their normal internal structure. Any inductive back EMF due to inductive turn off will be very safely clamped, and the resulting energy fed back into the big electrolytic that should always be there right at the dc feed to the bridge. In fact if you look at a bridge type of inverter, just the four internal mosfet diodes look and work exactly like a bridge rectifier to anything coming back the other way. Any ac or dc energy kicked back into the inverter output from the load or the inverter primary circuit series choke, can flow back onto the dc supply rail making the inverter bi directional (as long as the four protective diodes are present). Some very effective built in turn off protection comes for free with a mosfet bridge. Cheers, Tony. |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Have to agree, it shows all the signs of sloppy software programming where the program internally jumps or branches or receives an interrupt request, and there is sometimes very momentary evil data sent to an output port. There may also be vulnerability to noise. Any random noise glitch on an input that can initiate an interrupt, is not likely to be very helpful. Cheers, Tony. |
||||
| wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 1156 |
Why is the AC over-current input used at all with the control board and why is it latched with the SCR? I dont want to state my understanding at all as it adds to confusion if not correct. I have seen posts that have described an overload tripping out a mech. circuit breaker & requiring it to be reset. Given the high overload capability of the inverter is it safe to dispense with the current transformer totally or not? If the (AC) current sense fast latched shutdown was related to DC Current input and the AC overload was via a mechanical circuit breaker would that approach be more beneficial ? The control board currently has the current transformer as part of a latched shutdown & as I am looking into modifying the start up circuit & I am questioning what needs to be where & why. I would be grateful for some feedback/comments. If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
There has been quite a bit of discussion about overload protection for a very long time, and a few different schemes have been experimented with. The overall consensus is that there is a lot of stored energy in the electrolytics so its not feasible to protect the inverter at the dc input. A reasonably fast current transformer in the 230v output triggering an SCR has proved to be quite effective, and the latching action of the SCR prevents cycling. The overload circuit must be manually reset. A suitable non saturating choke in the primary of the transformer will limit the rate of initial current rise, at least up to some point. And hopefully the mosfets will be turned off quickly enough to live another day. Its not very scientific, but it has proven to be quite effective in practice, and you cannot really argue with success. Some of the schematics people have posted concerning the circuitry around the CT are woeful in that there is no low resistance burden resistor across the CT, so it cannot really ever work properly. But to point out glaring design errors in a schematic that many people have copied, and are completely satisfied with would be suicidal. The Inverter cowboys on the Forum would gun me down at high noon, so I only mention this now. However, a decently fast current transformer (with correct burden resistor!) a high gate sensitivity SCR that shuts down the enable pin of isolated gate driver chip directly does work rather well, and its what I used in my own experimental PWM inverter, although I never really tested the output protection to destruction. The SCR also shuts down the software but that is incidental. Its done only so that resetting the SCR also brings the soft startup into action. My own very odd multistep inverter is very different. There are no series chokes feeding the transformers, its four brutal square wave inverters with 40,000uF of low esr capacitance feeding the whole thing at +100v dc. I am using IGBT's that are much more robust than mosfets and the high power stage uses IGBTs that can handle a 1,000 amps of fault current for 10mS. My largest output transformer is only 2.25:1 so I am confident that I can protect the output with just a normal C curve 20 amp thermal and magnetic circuit breaker and nothing else. Previous experience with this type of inverter has convinced me that this will be quite sufficient overload protection. Everyone else here has a 9:1 transformer, 48v,runs PWM, and multiple mosfets which will be far more fragile than my big slow IGBTs, so some kind of fast and very direct hardware shutdown will be well worth providing. Cheers, Tony. |
||||
| theoldwizard1 Newbie Joined: 25/09/2018 Location: United StatesPosts: 5 |
Trying to absorb all of the information in this ONE thread !!!  I really like this design (KISS, although the isolated grounds and twisted pair might be overkill ...) Has anyone implemented this on an EGS002 ? It appears that many of you have abandoned the EGS002 (but not EG8010) and are implementing the functionality on you own PCB. Are there "better"/more popular ways of providing the same protection ? |
||||
| Solar Mike Guru Joined: 08/02/2015 Location: New ZealandPosts: 1138 |
I'm reading this blog with interest Wiseguy to see what improvements can be made concerning the use of the EG8010 chip, I certainly don't trust it in all respects; any engineered improvements would be good. Herbnz: I don't think soft shutdowns in an overload situation are going to work too well with the EG8010, it only measures the output voltage once each 50hz half cycle; by attempting to have the chip reduce its voltage output in response to current overload will be way too slow, by the time the thing starts a soft shutdown all the mosfets will be dead. My own analog based design is to latch shutdown all mosfets in the bridge by removing their drive at the same instant when a current overload occurs, whether it be DC over-current or AC sensed. The management cpu is too slow to detect an overload and attempt a soft shutdown that would protect the mosfets. Cheers Mike |
||||
| wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 1156 |
Thanks for the feedback Warp & added humor - makes reading more interesting. I accidentally discovered that I have suicidal tendencies - I obviously need to tone down a bit having been duly reprimanded. I'm afraid that I tend to play devils advocate a lot, its part of doing FMEA's (failure mode analysis) I also question "we've always done it this way" - not to insult anyone but more to make sure that the right solutions or approach have been chosen - or at least ones I can relate to. I say this more with regard to what I will do in my design by understanding what already exists & why. Just as an aside I will describe my fading memory of how an AC current transformer circuit works for those that haven't had the pleasure of playing with them - - for those that know skip the next paragraph. I am doing this without referencing any controller schematics published here so my comments can't be taken as criticism. A Current transformer can be easily saturated & so the whole scheme is to have relatively balanced currents flowing in the primary and secondary to avoid this. If the transformer is a 1:1000 then the current flowing in the secondary will be 1000 times less than the primary. If the primary current for a single turn is 25A the secondary will be 25mA for best balance of flux & linearity. In my experience usually voltages of around 1 to 4 volts are common for the secondary, so in the above example, if we want say 1.5V therefore the resistance will be 1.5/.025 or 60 ohms for the above example. Burden resistors should never be removed or be too high in value as ridiculous voltages can be present on the secondary for high primary currents. A suitable burden resistor can also be placed after a rectifier bridge on the secondary. I question the validity of "The overall consensus is that there is a lot of stored energy in the electrolytics so its not feasible to protect the inverter at the dc input" If we are talking about an AC overload, the FETs are passing higher currents cycle by cycle from the DC way before the AC sensor at the output sees it and passes that information back to the control board to shut down? If we were thinking it would be nice to stop the FETs self destructing by measuring the secondary current I think its probably too late to stop the primary meltdown underway. I will be investigating both DC RMS input current sensing, FET bridge cycle by cycle sensing and fast shutdown for the FETs. Lastly re voltage monitoring thoughts for 5V 12V and Gate drive supplies, I am toying with using some cheap optocouplers and a few other components. All the opto outputs will be summed to a common signal which as long as all the voltages are present & in range the signal will allow the release of the SPWMEN and the FET driver enables. Any voltage loss will either quickly shut down or/and refuse to start. A reset switch will be incorporated to disable the FET drivers & remove the EG8010 5V power (not all 5V) and re-power it again if & as required. I can see merit also in latched shutdown and will try to incorporate that as well. Solar Mike - you are wise to mistrust the EG8010, I considered a few times not using it, but I would prefer to add whatever "bells and whistles" circuitry it takes to better tame it as it appears to run well for the majority. All the crappy behaviour I have seen is really limited to startup or after a disturbance event. I just hope Warp is wrong and that the design is frozen as is & they dont/wont play with the code any further. If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 1156 |
I am not aware of anyone currently using Warps opto scheme - which would have solved the EG8010 overlap glitch. The EG8010 also has programmable dead times, so adding more is not truly needed, but given the consequences of getting things wrong there's nothing wrong with a belts and braces approach. Twisted pairs and isolated grounds are usually good things to have in high power & noisy circuitry. If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| theoldwizard1 Newbie Joined: 25/09/2018 Location: United StatesPosts: 5 |
In a previous life I did a bit of that ! I still want to see the FMEA on the 787 where they decided that the battery compartment did NOT have to be vented outside ! |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
Haha, FMEA's and safety critical equipment, and all that paperwork trail. Been there done that myself in the railway industry. It certainly raises one's awareness and thinking about possible unexpected failure modes and the consequences. The "Warp inverse opto scheme" solves quite a few potential problems with very little added complexity. Additional dead time is optional, but an EGS002 could be set for 150nS initial dead time, and a bit more added to that if required. The extra capacitance even if only a few hundred pF can only help with noise immunity. Anyhow, getting back to mosfet overload protection. We must not think about overcurrent as being a slowly rising event in the 50 Hz output sine wave, we must think about overcurrent at 23.5 Khz and the rate of current rise into a dead shorted inverter output with almost nothing to limit the fault current. Our big toroid is the ultimate low impedance fault current source. Without any dc choke in the primary, there is virtually nothing to limit the rate of current rise in the mosfets if the inverter output is catastrophically dead shorted. There may be something like 12.5 Joules of stored energy (10,000uF at 50v ?) probably more stored energy than that. That does not sound like much, 12.5 watt seconds, but its 12.5 Kw if it discharges that electrolytic in 1mS, and 125Kw in 100uS! More than enough to blast all the mosfets. If there is a non saturating choke in the primary that works up to perhaps >150 amps before saturating, that will slow down the rate of any fault current rise to give the overcurrent protection a chance of working. But shutdown still needs to be made very fast acting because the choke current will already be right up there, before the protection can kick in and remove the gate drive. Probably 1uS shutdown or less might be achievable even with opto isolators. Cheers, Tony. |
||||
| Page 6 of 11 |
|||||
 Print this page |
