 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 : 8 KW Inverter Build
  Page 11 of 22   |
|||||
| Author | Message | ||||
Madness Guru  Joined: 08/10/2011 Location: AustraliaPosts: 2498 |
Gidday Warp, It is happening regardless of load which is part of why I keep thinking the driver is the issue as 3 FETs on each leg of the bridge is fine but 6 is bad. 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 |
The IR gate driver is definitely weak, and something with a lot more grunt would definitely be a step in the right direction. The other problem when running a large numbers of small mosfets in parallel is getting them all to switch simultaneously and share equal currents. Its not as easy as it appears, and matching Rdson is only a part of a very complex problem. Wile mini TO3 packages are nice and slim and compact and low cost, TO247 packages offer better thermal performance. SOT-227 package would be even better with a bloody great thick aluminium plate bolted across all the sources and the drains. But that would be expensive. Third issue is using really cheeeeep Chinese mosfets........You get what you pay for, Ah so. The other imponderable is the magical mystery super low cost EG8010. Nobody really knows what oriental mysteries lurk within. Too many unknowns. Cheers, Tony. |
||||
| Madness Guru Joined: 08/10/2011 Location: AustraliaPosts: 2498 |
I am using TO247 IRFP4110's and HY4008W I have 5 boards made up and have checked RDS on and they are all very close in particular the HY4008's are all the same. All the PCB's give me the same result, who knows what goes on in the EG8010 but it can produce a very clean waveform. Where I have trouble is when there is more than 3 FET's per leg. The IR2110's should isolate the number of FETs issue from the EG8010 and as you mentioned previously Warp the IR2110 does not have the balls to drive the gates real hard. 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 |
Where you have multiple mosfets, the first one that turns on, and the last one that turns off sees ALL of the load for an instant. The more mosfets you have the more likely it is there will be a matching problem. Also, the slower the gate drive, the worse the potential problem, especially if the gate thresholds or die temperatures diverge to have become very different. Using fewer or larger mosfets definitely helps here. The "big fella" mosfet power blocks usually have multiple mosfet dies internally, but the bonding wires are very short, the dies much better matched, and so are the individual die temperatures are better matched. So its much closer to the ideal for simultaneous switching rather than a long spaced out string of smaller individual mosfets. I have no idea what the problem is, just playing Devil's advocate, and suggesting some potential problems. I don't like high frequency PWM for really high power low voltage mosfet inverters, for all of the above reasons. Although it does work well at lower power levels where we don't have to use so many multiple parallel mosfets. High frequency PWM works much better at higher voltages as used in grid tie, especially using single individual IGBTs which are far more rugged than mosfets anyway. IXYS make some pretty big mosfets. I have not really researched this in depth, but when I was considering building a high powered dc to dc 20Khz full bridge forward converter, I was looking at the big IXYS mosfets. The 320N10 is the baby of the group and lowest cost for around ten dollars US each + postage. I was thinking of getting four to test as a prototype. http://ixapps.ixys.com/Datasheet/DS99833A(IXTN320N10T).pdf You can get them at that price from Vertical in the US, and they are the real thing, not a Chinese knock off copy. Even larger ones are available, but they get to be rather more spendy. https://www.verical.com/pd/ixys-fet-mosfet-ixtn320n10t-553444 Some sellers are asking up to fifty dollars a piece for these which is ridiculous. I have not used these myself yet, so cannot speak from personal experience. A recent change of plan here means I will now be using several lower powered 1Kw flyback converters run in parallel, rather than a single full bridge multi kilowatt forward converter. Cheers, Tony. |
||||
| Madness Guru Joined: 08/10/2011 Location: AustraliaPosts: 2498 |
Those MOSFETs are interesting, they are good value compared to banks of smaller ones to do the same job. The mystery for me is why Oztules has his working and so easily, must be something to do with living on an Island. There are only 10 types of people in the world: those who understand binary, and those who don't. |
||||
Grogster Admin Group  Joined: 31/12/2012 Location: New ZealandPosts: 9308 |
The IXTN320N10T seems to be no longer active, according to the IXYS website, but they recommend the IXFN360N10T as a replacement part, which has even higher current rating, and lower Rds-ON. Would it not be sensible to use four of these instead of paralleling up lots of wee MOSFETS? You only have four devices to worry about then, and being the beasts they are, should be more difficult to kill, yes? Four of these would allow you to build one single output stage to drive the toroid directly, and should also restore a perfect waveform output over multiple parallel stages, yes? I am seriously looking at doing it this way, as I have not got around to building my inverter yet, although, I have been following along with the threads whenever I can. Much more expensive on a per-MOSFET basis, but would work out about the same as buying $50 worth of TO220 MOSFETS and then having to mount 24 or so of the bloody things, to have one blow and take out the rest. The big beast MOSFETS mean you only have to replace four - worst case - and that is a damn sight quicker to do then 24 220's...... I'm interested in anyone's thoughts on that, but particularly warpspeed and oztules. Smoke makes things work. When the smoke gets out, it stops! |
||||
Tinker Guru Joined: 07/11/2007 Location: AustraliaPosts: 1904 |
So I took this bait  and ordered 4 x IXTN32010T from Verical who sells them.Keep in mind that the cheapest shipping option is nearly as much as the 4 Mosfets. But what the heck, I like to try them out. At that price the FedEx delivery is much quicker than that stuff from China so I should see them by late next week in my letter box. What I particularly like is the devices can be bolted directly to the back of the big Aerosharp heat sink and using my tinned copper strips for the power tracks. A little PCB can then accommodate the gate resistors & diode. Klaus |
||||
| Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406 |
I do not know if IXYS have an agent in Australia, or what sort of deal we could get from them if they do. The prices at Vertical are good even with a high US dollar, and they accept pay-pal too, but shipping is cruel. It should make for a much more easy to assemble (and repair) power stage, but it will still need four hairy chested gate drivers with around ten amps of gate drive capacity through a one ohm gate resistors to get sufficiently fast switching. I did build up a bridge of four SOT-123 devices on a large heat sink, but used SCRs of the same type of package as dummies to get all the mounting holes positioned. I never bought the big IXYS mosfets to try out. My digital camera is FUBAR, need to now go out and buy another camera so I cannot show you what I did with that. What I was planning to do was build around a 4Kw dc to dc converter to convert 30v from a lithium battery to drive my high voltage inverter at night. The problem with that is my refrigerator draws a start up surge of almost 4Kw, the average load on the inverter at night is usually only about 50w to 150w. A 4Kw forward converter is just not efficient at such low average power, and it would not have been happy running at such light load. So I am now going to flyback converters which are wonderful at very light load, but much less efficient at flat out full power. That will be fine, because I only draw full power for about on second while the refrigerator actually starts up. Multiple flybacks also work well paralleled, so its easier to build four 1Kw flybacks than one 4Kw forward converter. So I no longer need to use those big IXYS mosfets. All this is way of topic, but just trying to explain where I am at with all this big single mosfet stuff. Cheers, Tony. |
||||
| Madness Guru Joined: 08/10/2011 Location: AustraliaPosts: 2498 |
Is anyone game to put a figure on what maximum power these could handle for short bursts and continuous in an Inverter? 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 |
Its a matter of peak current and junction temperature. Without making actual measurements and knowing real heat sink capacity, there is no way of guessing. But at some point it will definitely go bang. Cheers, Tony. |
||||
| Madness Guru Joined: 08/10/2011 Location: AustraliaPosts: 2498 |
If there is an aussie agent the markup would probably be far more than the shipping. The Midnite charge controllers I have cost about $800 AUD includung shipping from EBay seller in USA but to buy locally are over $1,400 plus freight. What sort of fridge needs 4 KW to startup? is the start capacitor RS? I have a digital camera and never bother to use it anymore as my phone does such a good job. Even taking pictures of my CRO screen the camera needs all sorts of fiddling to get it right and all I do with my phone is turn off the flash and it is perfect every time. Would that Totem pole driver circuit back on page 14 be up to the job of driving those big MOSFETs? 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 |
Its a fairly recent model refrigerator with nameplate consumption of 112 watts running and 275 watts in defrost mode. I hooked it up to my digital storage oscilloscope through a current transformer and set it to trigger and capture the start up inrush current. The first half cycle is a stunning 17 amps ! And it does that every time it starts. My little home made 1800 watt sine wave inverter handles that 4Kw+ surge with ease, only a slight flicker of the house lights, hardly noticeable. But if I completely lose grid power, and want to run completely off a battery, that 4Kw surge is a problem. It all has to come from the battery through a dc to dc converter to step up the voltage. Cheers, Tony. |
||||
oztules Guru Joined: 26/07/2007 Location: AustraliaPosts: 1686 |
There is a new scope..... a German thing, but it has dual scan.  The top one is the 8010 running a few kw...... and the bottom one is the town grid  I am at 50hz, and they are drifting above that from the looks of it.Here is the unit s few weeks ago...  and now a day or so ago...  So we have gotten a few kwh's under it's belt...... The unit runs at 232v... so the voltage drift is very slight for a 5kw load.... still blows me away. Madness.... some times it just works...... actually if you blow that pic up, you will find some glitches in there somewhere, but it is much better than the main grid from the looks of it.... and stress free life for fets on the island too Warp.... this one has it's current cut off somewhere past 15kw, as it shows 15kw and more starting the 5hp lathe motor with a clamp meter.... it's sampling is not fast enough to see the real transients I suspect.... probably in the 20kw range. I set the current trip so that it does start the lathe Start up is not a problem. ...........oztules Village idiot...or... just another hack out of his depth |
||||
| Madness Guru Joined: 08/10/2011 Location: AustraliaPosts: 2498 |
Are you showing off OZ  I will get there one day, I don't know when but I will get there.  Even though mine is not right it has been running over a week in it's current configuration with 6 KW plus continuous loads while running in ambient temps close to 40 degrees. There are only 10 types of people in the world: those who understand binary, and those who don't. |
||||
| oztules Guru Joined: 26/07/2007 Location: AustraliaPosts: 1686 |
In a word.... yes After looking at the Hydro Tasmania wave form.... who wouldn't?... and theirs goes off all the time too  You push yours a lot harder than I do. My 8kw runs are for 5 minutes as a rule, much longer than that, and the 20 amp o/load would trip. I'm surprised the thing holds when the lathe starts, but it is so fast that it does not bother it. Your long runs are very impressive, particularly in that heat. If it holds together now, you have beaten me hands down. ........oztules Village idiot...or... just another hack out of his depth |
||||
| Tinker Guru Joined: 07/11/2007 Location: AustraliaPosts: 1904 |
Warpspeed, do you have some detailed info or parts # for these "hairy chested gate drivers"? I take it the IR2110 chips are not up to the task? Thanks Klaus |
||||
| Grogster Admin Group Joined: 31/12/2012 Location: New ZealandPosts: 9308 |
A MOSFET has an isolated gate, unlike a transistor base connection where there is a fraction of the collector current, flowing in the base, so how come you need big beefy gate drivers, when the gate should not be conducting any measurable current anyway? I might be missing something with the high-current MOSFET's over the dinky wee ones I usually use, but talk of needing 10A or so of gate-drive current does not compute based on what I know of MOSFET's. However, we are talking some serious S-D current in an inverter application, so perhaps this does result in a gate current of that kind, but it is still making me raise my eyebrows..... Normally, you only need a voltage potential on the gate to turn the MOSFET S-D path on or off - or somewhere in between with a mid-range gate voltage, but the gate current is usually immeasurable if it exists at all. Can one of you gentlemen explain what is going on with such serious gate-drivers needed? Smoke makes things work. When the smoke gets out, it stops! |
||||
| Solar Mike Guru Joined: 08/02/2015 Location: New ZealandPosts: 1138 |
The gate acts as a small capacitor approx 10,000 pf to source and another smaller value to the drain. At turn on the voltage rise charges it and discharges it at turn off, this requires energy, for nano second switching a lot of energy, ergo current pulses in a small period of time are required; the faster the switching speed the more average power required. There will be a formula for working out this power requirement in coulombs if you google it. So a beefy driver is required that can supply many amps of current during these transitions, but the average current and thus power is quite low only a few watts for many fets in parallel. If the driver cannot supply the required current then the switching transitions are not so defined and results in the strange waveforms that some are showing here. This is where a Totem pole buffer driver placed close to the switching devices may be used to good effect. Mike |
||||
| Madness Guru Joined: 08/10/2011 Location: AustraliaPosts: 2498 |
The Application Notes for the IR2110 has information about driving MOSFETs & IGBT's needing higher gate current. "7. HOW TO BOOST GATE DRIVE CURRENT TO DRIVE MODULES Modules and other paralleled MOS-gated power transistors at times require more current and lower gate drive impedance than what a typical MGD can provide." This is partly where I got the information to investigate adding a Totem Pole drive to the existing circuit I am using. I have the PCB's etched to do this, I will be trying it on the weekend using the circuit on page 14 of this thread. The application notes above show low side but as far as I can see not high to do high and low as we require for an inverter. There are only 10 types of people in the world: those who understand binary, and those who don't. |
||||
| Solar Mike Guru Joined: 08/02/2015 Location: New ZealandPosts: 1138 |
Madness, look at section 8, fig 15; it shows the high side buffer. We will all be interested on how you get on. |
||||
| Page 11 of 22 |
|||||
 Print this page |
