Project Journals > Oztules

home brew 5kw simple inverter..... 20kw surge ( really), and seems bullet proof

(1/4) > >>

oztules:
Well it is time to actually do a how to build your own power inverter, as it is now mature and very very well field tested by now in as many as 16 houses on a remote location. They run small houses, big houses, islands running reverse osmosis salt water units, and shearing shed, while also running hot water, air conditioning, fridges and freezers.. all at once if necessary..... these are not toys.

Nor are they expensive. If you can get hold of big transformers, the box you put it in will likely cost as much as the electronics.

It is my experience that they cannot be bested as an inverter even by top of the line units like the selectronics ( one has replaced one recently, the selectronics is on standby duty to start the big diesel only , as the selectronics does not seem to play with all the home appliances properly.)... the other thing they can do easily is start 5hp induction motors... that takes 30hp to do... and they do it easily ( 5-6 times the motor rating as a general rule). When you see well over 80 amps@ 240v on your clamp meter at motor start up... you know there was a lot going on.... I actually use that motor as the reference for the current sense shut down trim pot.

Anyway, I 'm not sure how to start this off, or how much detail to go into, so I'll just wing it and see how it goes. I will start with a ebay bought inspire inverter, which were cheap as chips and available everywhere for a long time until I did these things, and now they seem to have disappeared from sight.... but I am told there are still lots of Aerosharps in the solar industry warehouses some where.... but any big Grid Tie that is galvanically isolated will have big transformers of the torroid style so there the thing to look out for.

There are other folks out there who built these too, but even bigger, so they are running 8kw for long periods..... they handle it easily.
OK.

The heart of the unit is a small chip from the chinese. It is an EG8010, and can be bought for a few dollars or less from Aliexpress like this


This lousy little chip is just incredible as an inverter. It seems to handle everything thrown at it and then some... truly amazing piece of electronics, and blitzes everything else I have played with.... just seems near idiot proof ( I'm proof of that).

Suppose I should put up a pic of what we are trying to  finish up with. Here is one of my many units.... it is simple and bland.

If you want thousands of things to change and play with this is not for you... get a selectronics.... if you just want an inverter that works... then this will be the ticket.


and


So thats what we want to end up with.

I start with two of these inspires or two aerosharps, but if you can get decent housing for your project, then really the GTI's are not necessary, but the transformers will be very much more problematic to procure, and probably expensive. If you can source good welder transformers, then these can do too... be some rewinding, but so will the big torroids. The idle power will be about double for the E-I transformers, as they are not as tight magnetically ( more leakage), and heat and weight will be about double as well.... but you will still have a very good inverter.

Still in the box


Open the top and we get this... nicely built, but a dog of an inverter actually, seems prone to high frequency oscillations that appear from nowhere really. Any way, we toss all this out, and everything underneath, except for the tranny, and the noise interference units, we can use then later.


in the bottom of it we find the tranny etc....



There is the tranny and the big noise filter, and some other stuff.... all goes out to get a clean box/s



Now we need to remove the huge heat sinks... and I use a 50 ton brake press , but a hammer and chisel, and screw driver and any other prying devices you can get your hands on will do ...... this can be frustrating with simple tools, but doable... and you get these.



Next thing to do is make holes for your fans. I use 125mm fans. I buy the fan grills as well from ebay too....
I used to carefully grind and mess about making nice clean round holes.... but that gets old fast, so now just profile it like this..



You can see where I put them on the finished pic above. Two on each side of the two boxes... so you have four to do.
A word about these boxes. They are quality units, and you drills will have to be very good and sharp. I found the cheap Chinese cobalt ones from ebay are terrific... the Aussie frost, and P&N, and viper etc are not much good really, expensive but not up to this stainless steel.... but the cheap chinese ones are. You can get 40 holes with the chinese ones, or maybe 5 if your lucky with the top of the line Viper bits.... go figure... you milage may vary, but thats my experience anyway.


Well thats it for the first bit. as we are at the 9 pics  for the post..... so I will leave it here while you drill your holes and cut out the fans bits, and get the heat sinks off..... we then just join the two boxes together via the heat sink bolt holes. They are mirror images of each other, so bolt directly together with 5mm bolts and nuts.

........oztules

oztules:
Well there will be those that want to jump ahead I guess, and they want to see the boards. These have been designed for simple remote living, and can recover from a total disaster by simply unplugging three chips and replacing the fets... and your good to go from dumping it in the ocean and blowing it to pieces... thats remote living at it's best. No more waiting for replacement stuff from the manufacturer.... but for $30, you can rebuild the entire thing. This is why I developed it instead of the power jack boards.... so here are the three gerbers I got  printed at a place in China called pcbway. Beautiful work for peanuts... in fact they were doing 10 4x4 boards for only $5... crazy, I can't do one in the bath tub for that. test to see if I can post zips


Anyway here is the power board gerber. You can send this to them or any other manufacturer and get your bards professionally made. They are double sided and double thickness copper for the power handling aspect for the power board.

 W21223ASJ13_CAM%20for%205000AW.zip (56.2 kB - downloaded 452 times.)



It looks like this in my pcb program



Download the gerber files and get a gerber reader, and look at them in that.

Here is a picture of it in real life. On this one I used a copper bar 5mm thick as the negative power input soldered to the board, others use the holes I provided and used thick cables to get the juice to the board.

and

Here is the control card gerber files

 W21223ASJ14_CAM%20for%20CONTROL.zip (35.83 kB - downloaded 382 times.)


And it looks like this:



or this


The final board is just a fan control card ... tiny, but requires a seperate buck converter to drive the fans. The tip35c's on the control card are only for the control functions, and won't drive a fan, much less two of them, or four if your really keen to cool.
 W21223ASJ8_tempfan2.PCB.zip (3.93 kB - downloaded 347 times.)


So you can get them printed in China or wherever, while you build the case, and transformer.


Here is a finished boards set up...... just so you get the picture of how simple these things really are... yes thats the complete electronics to make a very robust powerful inverter.... seriously....




Will have to stop here as we have run out of attachments to be allowed to do.

.......oztules

oztules:
A gentleman called tinyt did the circuit diagram for everyone, ( I don't do diagrams/schematics, I just design in the pcb program... so I'm useless like that )..... but everyone apparently needs them, whereas the boards do all the work for you... who knew...

Powerboard


Control board


IMPORTANT!! The diode D6 is listed as a zenner of 13v... it is in fact 18v. If you use 13v you will lack enough drive for the fets... USE 18v for this not 13v

The pin 6 of the 8010..... When you put the 8010 on the daughter board, make pin 6 stick up not down. We will use this for the take off point for the on/off switch ( labelled as spwmen. Do not put the pin on this pin down or you won't be able to turn it off. I originally used the temp pins for on/off, so this pin was permanently 5v, by putting the pin up we can use a lead to the switch, and spwmen then goes to ground for off or + 5v for on.

or pdf here
 2018-02-12_151253_OZINVERTER_POWER_BOARD_SCHEMATIC5000AW.PCB(1).pdf (24.58 kB - downloaded 460 times.)
and here
 2018-05-04_074827_Oztules_Control_PCB_and_Power_Board_Schematics.pdf (49.63 kB - downloaded 487 times.)

So there are the circuits for the two main boards, the fan control is best shown here perhaps.... no circuit, but self explanatory.


I'm too lazy to do the BOM.... all the component values are on the boards.... use the gerber viewer to see the top overlays for the parts or the schematics.

The schematic depicts 6 big caps, it actually uses four on this unit, and the two center caps is actually one single 4uf ceramic cap for high frequency suppression on the bulk caps... but that's the only ,mistake on the schematic I can see at this point in time.

I now use the Chinese HY4008 fets . They are very very very good... much better than the irf4110, and come in a decent sized case too. They easily ( 4 of them per leg) run the 20kw start up currents for the big induction motors and everything else. I used to use 6x4110 per leg, but the HY4008 is a better unit I think. On Aliexpress they are only a dollar or so USD..... simply brilliant value. I test them on arrival, and all fall within specs... and look at those specs....


Just a single fet per leg will do 2kw all day long... darn impressive...

Here is a pic of the fan control and the buck converter from ebay ( 10 amps@12v for about 8 dollars )



It is very very wise to use a few 68amp o/loads on the input from the battery to soften the blow if you should blow the thing up somehow. If you do like I did in testing, and you have the breakers ( NOT FUSE) in line, you will only damage a leg or two of fets with no board burn, and probably not even the drivers, no smoke and no tracks exploding. Without the breakers, the fets sequentially pop one after another until they and everything attatched is destroyed, and no path for the battery can any longer get through... this includes tracks. This is all stopped by the application of 5 dollar breakers.

If you question the manufacturers, they will tell you that you can use these AC breakers at voltages less than 60v with about a 10% derating DC breakers are expensive and will do no better.

There is a drawback ( there is always a drawback). When you attatch the battery and hit the breakers to energise the unit, it will likely weld the contacts together on some brands... it depends on the internal structure. It does not make a difference if it is DC or AC breakers, they will have to contend with over600 amps on battery connection..... the caps are that big, and the power is huge to charge them up instantly.... be warned.

So use a resistor of say 8 ohms, in series with the battery lead you connect or bridge the OPEN breakers with the resistor. It will charge the caps in a second or two with no pyrotechniks.... very sedate, then throw the breakers, and then turn on the on/off switch. The inrush currents are absolutely brutal.

Now this presents another problem... the unit will probably not run, and the indicator light may or may not light. It is controlled by the 8010, and it does not like soft slow ramping up of it's 5v... so it hangs. We have a reset switch for this very reason ( amongst others, like o/current events). Sowe now know that a slow ramp up will likely result in nothing happening.

Reset in this context is a press switch shorting out the 48v input AFTER the 120r resistor. This will result in the power to the 8010 dropping to zero, and then when you release the switch, it gets it's instant fix of 5v, and is happy thereafter. It is IMPORTANT to note that you press the switch with intent... a very fast contact will result in not all the voltage being shorted, and can confuse things , maybe even resulting if fet burn. So hold it for 1/4 second or so when you reset at any time. .. and all will be well. I did not ever notice this as I always used a press switch, but some time back a bloke simply stuck a screwdriver across the resistor to ground with a glancing touch. This blew the fets...... so we now know.

The torroid transfromer takes no prisoners, and will not tolerate a rapid change to the internal flux, and will resist totally. This means any rapid change to the frequency ( instant sort of thing) will result in the back emf blowing the fets. This only happens if you mess with the timing somehow. I found originally, the on board current reset on the 8010 did not take this into account, so the native current control, almost always blew the fets, as the 8010 changed gears, and the drivers did not do a soft change... bang...

This also caused the power jacks the same problem when they came off shore power battery charging, when the chip resumed the 50.0 hz from the say 51hz of the source, this change would be fine in a normal E-I transformer, as there is enough leakage to absorb the shock, but big torroids just blow everything up if you try that.

All that makes these things sound finnicky, but they are not, I'm just telling you what can go wrong. With 16 or so of them out in the field for years, there have been only two failures in probably 40 inverter years of use in the hands of novice users running households with everything going. The two failures were user initiated... letting the units run down in the 40v range in a high power instance ( yes the batteries were run flat).

So if you open the pulse width to max, and don't have the battery to support it, there is a chance of failure. Beyond that they have taken anything and everything these folks have done to them... they really are tough, but getting the battery hooked up is finnicky the first time, unless you have a very big switch that can handle the turn on current, or just splat the terminals ( that works fine, just scary.

So the take away is this....... do use breakers, even though that is inconvenient the first time connect. The chip may hang on the slow start, the reset sorts that out so no problem anyway.

Can we reset under high power....... yes if you want.

Can we start under high loads ( say 8kw load on line before we press start ) yes, no problem starting under full load. ( Once,after the front 20 amp AC o/load threw I just went out and reset it..... it restarted, and the read out showed why the 20a o/load went..... it was starting up under 8kw load. The 20a o/load can do that for a minute or two, then throws)

So we know the instant/surge loads can exceed well past 20kw, and we can sustain 8kw too. Your cooling and heat sink design will dictate how far you can push it. Some folks have huge heat sinks, and can run 5kw for 30mins, and the fans still don't come on..... so efficiency is pretty good.

Grid tie inverters work well with these as well. We have run 8kw backwards through them to charge the batteries, and now a few Queenslanders do this all the time, as it allows for high voltage arrays at a distance, and the GTI can help run the house during the days with the air conditioning on all day, this takes some of the pressure off the batteries and the inverter.

They seem to run the airconditioners all day and night.  So the GTI runs them, and the inverter runs the excess( if any) back into the battery. It does this via the synchronous switching of the mosfets, and this happens to make the transformer into a boost converter in these instances, and is why the tranny which is really a 28v transformer for a 48v system, be able to charge the battery to beyond 62v or more with just the 240v input. The sychnonous switching in conjunction with the transformer is responsible for this behavior.

These inverters have a very very stable output. If you watch the lights, then switch in a jug or something else of a few kw, there is no flicker... at all.... remarkable really, but it stops the GTI from turning off from over AC voltage, which the powerjack did, as it let the AC rise if the battery was full, so it sort of controlled the charge ( brutally, but worked), but these things have such a stable output, that does not work, so we developed a new GTI controller for that aspect ( another article perhaps).

The use of an inductor will decrease the idle power consumption very considerably... around the 13-20uH seems to be enough to do it. ( three turns around a etd65 core is enough to do the job).

It does not matter if it saturates or not. You can design a biggie if you want to avoid saturation, but you will notice no improvement.  It is only used by me to quench the high speed current surge whilst driving the pulse width up. ie as the fets turns on, this inductor will impede the current just enough for the switch to turn on fully...... this means a 5kwh/day idle current turns into maybe .4kwh/day. It makes that much difference. It must also help the shock the silicon sees too.

Some folks have made beautiful inductors that do a little better too.

Thats enough waffle for this post.

.......oztules

oztules:
Ok, we'll come back to building the actual boards in detail later on.

Now the case needs to be progressed some more.
Here is the fan hole cutting using the slack persons approach.

I simply use the grill as a template, and sketch the holes and the rough outline  as below.


Note the lines lined up with the holes. This is the cut with a 1mm blade on a 125mm cutting wheel.If you stop at about where the curved lines from the template ( grill) was, then cut the corners, you will probably get there so it looks like this from inside.



There are two holes seen above the fan here, thats for the dc dc buck converter. The two too the left are for the breaker ( AC 230v side here). Also the hole for the ( neg in this case) main battery input. It is the neg for me here, and goes straight to the big copper strip on the power board, ( and the neg of the buck converter) You can also see the power switch ( blue thing) and the reset ( small black round thinggy) Inside the box we can also see the pos input terminal. This will go to the 126amps worth of breakers on that side of the box.

We can see better the two switches, an installed fan and an otherwise empty box...


Here is the fan for the tranny compartment, and the noise suppressors harvested from the inverters

Here we have two boxes joined via the heat sink holes from the original heat sinks. The board complex is sitting there screwed to two timber rods attached to the box bottom... rank but works well, and simple ( what I do best)



Here is how I mount the two 63a breakers. It uses a piece of din rail, cut, bent, and bolted to the side through the two screw holes mentioned earlier
The B+ will go from the stud sticking out to these breakers. A 25mmsq for each from the stud. I use 1/2" copper water pipe to contain the two 25mm ends, then bash them with a hammer, and drill a 10mm hole to take the stud. Proper lugs would be nice,.....


So wired thussly... also wires to the B+ heat sink from the breakers.


I make the studs out of 10mm 70mm long bolts. Cut them in the vise like so,


The easiest way to maintain the thread, is to grind them pointing like this. This always works simply and easily.


The I use the appropriate sized cable gland with the end truncated, as the insulating portion of the standoff.....

And goodness me, we have run out of attachment space.


.........oztules

Pete:
Nice to see that you have been sitting back relaxing on the island this winter OZ.
Thanks for the write up, it looks great.
I still have a PowerStar and a Power Jack running at my place but it looks like one of  your inverters may be worth putting together sometime.
From what I read it appears that if you put power back into the output of the inverter it will decide that it is a battery charger. Is this right?
Pete

Navigation

[0] Message Index

[#] Next page