Building a 6kw pure sine wave inverter using power jack boards part1 transformer

[Bruce S]

ClockmanFRA;
Will do what we can, FL was unstable at best this past weekend, don't want to push it just yet.
Deleting the post would be a HAZ! so I may merely single it out , split it off, then quarantine that single post.
BTW: Since I'm posting on this thread instead of lurking,,,
LET me also say CONGRATS!! on the build.
I've said it in the past OZ has schooled me more than once, and this lesson was awesome!!
I'll 2nd the motion about the primary acting like heat pipes, so some air flow around the outsides of them should be sufficient.

Cheers;
Bruce
I'll work on the other post will having my lunch :_)

[DamonHD]

I just removed the photo attachment link that I thought was the offending one.  Did I do the wrong thing?  %-P

Rgds

Damon

[clockmanFRA]

Thanks Bruce S.

Thanks Damon that's spot on that post.

The other photo with my wrong numbers is in post 93.

Thanks!.

I have just copied post 92 and 93 and could re-write them both with appropriate amendments if you want to send both to quarantine.?
Damon its snuk back in......

[Bruce S]

I really need a 5th monitor on my work computer!!
I didn't see Damon's work until I split and moved  .
Can be moved beck if necessary.
Sorry!!

[clockmanFRA]

Bruce S that's good, please remove the other wrong numbers post as well and then I can re-post both.

Thanks

[clockmanFRA]

This post is slightly later than it should be, as I got the Pin count the wrong way round.

Hi mab,

I have done some detective work on my strip for parts 10,000w PJ Inverter.

The below picture of the PJ Power Board now shows the solution, and yes there is a ON and OFF switch on a multi pin connector and its ribbon cable to a little PCB at the rear.

So to switch on and off the Power Board, connect Pin 11 to Pin 13 , counting the pins as per my photo.


 
I note that according to PJ, there is also a sleep mode, so Pin 11 will want to be connected to pin 12..


 
Good old Chinese Instructions, Not.

This is the On Off Switch/Mode.

 

And that small pesky PCB on the 10,000w PJ Inverter.
 

[clockmanFRA]

First posted on 22 August but incorrect PIN OUT NUMBERS, now all corrected.

Oztules, forum members, It LIVES........... It LIVES................................

That Pin 11 connect to Pin 13 and she switches ON, disconnect and she Turns Off.  Brilliant.......
oztules, I trust that's correct?.

Okay, I was running out of time and light this evening, but I put 4.8kw of load on for 35 minutes or so.
Photo shows Inverter running after 30 minutes, and yes the PJ Fan came on at 42 c. Temp meters showing the warmest bits on the PJ Board, and with the Toroid its the Primary Turns/big wires that get warm.
The meter is showing battery Voltage under load.



HZ at 50.



AC voltage with 4.8kW load.



Ac voltage No load.



Yes oztules you were right, the E core/choke has a gentle hum, and yes I super glued it together. In fact the e core/choke makes more noise than the toroid.

I am a happy man...............................
I have a suitable cabinet to put everything In, so I will get busy, but I will work out an Air flow for that PJ fan and my other 2 Independent Fans.

Oztules, do you cool the toroid in any particular way?.
 Me thinks it would be nice to have some air coming up the middle ?.

Thanks again.....     
 

[clockmanFRA]

Further thoughts before final mounting in its steel case, 600mm by 400mm and 300mm deep, enclosure.

Cooling is important as it gives those extra kW range.
I will leave the PJ fan and its connection as is and well alone and connected to the PJ control board.

I can not seem to get good Beecroft ball bearing 5 inch fans here, so gone for a couple of these 5 incher Startechs.
http://www.ebay.co.uk/itm/380665507976?_trksid=p2057872.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT



I have a separate 24v supply in the enclosure so will use this to set temp switching of these two fans using this.
http://www.ebay.co.uk/itm/261735347690?_trksid=p2057872.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT



There seems to be a fair bit of noise about, so as oztules suggestion I will fit 4 off these EMI Filters on the AC 230v Output, that should give 40amp.
http://www.ebay.co.uk/itm/321820973359?_trksid=p2057872.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT



Oztules, sorry to be dim again, if I am back feeding through the 230vac when AC Coupling and charging the batteries, will there be any issues with these EMI Filters.? Can they handle it?

[oztules]

The EMI filters will have no effect on theback feeding... they only impede hf waves, not 50hz waves. To 50hz waveforms these are invisible.
I prefer two stages...ie caps torroid, caps torroid,caps.... output... so two of those smaller ones in series and then paralleled for power

There will be a few in your grid tie for the same reasons.

As you have found, they are pretty good up to 4-5kw with very little real cooling on the bench test. If you expect to run in the 8kw range for long periods ( my batts would be very disappointed doing that for long periods) then cooling will become very important... make sure you can ventilate the box well.... you don't want heat building up in that or all your efforts will be in vain for high power long periods.

Keep the original temp probe in it's original position as a back stop emergency shut down device.

................oztules

[oztules]

For example this is real time now 10.30am.

This is a pic of the inverter in it's current guise before I upgrade it.
It shows the blue screen as watts input from a single grid tie..... the hot water is turned on for 2400 watts... the grid tie is providing the first 1120 watts, the inverter providing the rest plus the house loads for another 1.4kw so about 2.5kw going out ot the moment.



Here is the noise filter running at the present time.. it is from 2 x 3kw aero sharp output filters in parallel.... I did not use noise filters... until I decided to build a single valve regeneration reciever to listen to the local ham net ( covering about 3000 miles)... thats when I found out how much RFI these things had.... and thats the quickie solution.



Current solar coming in is shown here in my 100 amp solar controller.



So you can see it is in the thick of things handling the grid tie, and house requirements, while th solar is left to push in a few amps too ( nearly 40 amps to battery after using a few to drive the hot water in conjunction with the grid tie.


.....................oztules

[clockmanFRA]

Real Nice oztules, shows what's happening, excellent.

I wont be able to show in real time what my Grid tie is kicking into the Oz Inverter created AC 230v grid, (AC Coupling) as the GTI Install and PV is about 200 meters away. And I have Promised her indoors that "thy shalt not dig any more trenches".

Thanks for the Filter info.

Coating the PJ PCB boards.......

This Oz Inverter will live down at the Power station building, in the Inverter shed, and replace that SMA Sunny Island 6, that can not DC Couple, cope with the batteries being separately charged, and AC Couple at the same time, with out sending the HZ stupidly to 60HZ. SMA say "but you should buy all our special expensive equipment". Nope, my pockets are not deep and I certainly won't become a SMA slave.




The building has lots of air through it....... So most stuff sits in its own cabinets. The Oz Inverter will get a weather resistant cupboard, and then be in its own Steel enclosure.

My other cabinets.....
PV Solar Tracker box....

[clockmanFRA]

Continued........Coating the PJ PCB boards.......


I tried to get some of your lanolin stuff , but its not sold here in Europe, and the supplier in OZ said they can not send aerosols any more.
I have these can I go mad with them?



That excess flux needs to come off first?

[joestue]

The flux should come off with isopropyl alcohol and use a tooth brush to scrub the board.

I have not been paying attention but you need to read the directions on the conformal coat to make sure it will stick to the board regardless of the type of conformal coat already applied to the board.

(i'm pretty sure there are a list of incompatibilities)

but if there is none then use whatever you can get. just follow the directions and bake the board if you have to, at the correct temperatures.

[Neil]

Hi Clockman, I haven't started mine yet, I need a split phase board and not sure if these pj boards will accept two inputs of 110 v. I noticed your board only has the one input for your 220. I'm waiting to here back from the man in China lol


Thanks for posting all your work

[clockmanFRA]

Thank you 'joestue'.

Good luck Neil, and have fun. For me it was the uncertainty of getting those 50mm/2 Primary wires on properly as they needed guided brute force without damaging anything. phew.........

[oztules]

Neil,
I strongly suspect the split phase is just a normal 220v 60hz conrtol card, with cenetr tapped output transformer.

As such for grid tie you would either use 220v grid tie across the whole secondary (220v output), or 110v grid tie into either 110v outlet.... or 2 on the two 110v outlets.

So it would be treated as a simple 220v unit with tapped output.

Because of this setup, getting a fix on the output current is problematic for this topology, as it uses the AC active leg for current sense. As a single ended output transformer, this works fine to tell the controller how much power is being generated.... but when you split the output..... where and which leg can you put the CT to measure the whole output.... you cant. So unless you match the phases with loads, or just run 220v loads, the unit will never have a real idea of power drawn, as it can only measure one leg, and never see whats happening on the other

If you were to use a 220v secondary and drove another split/tapped transformer with that, then you could measure total power and have full control, as the 220v rail will see it all.

As they don't use the DC to measure power, I have no idea how they accurately measure the load... I think they use 1 phase and guess the rest....

Just what I think they do, never seen one nor will I over here..... but cant see how else it would work.

So yes, I think both 220v and 110v grid ties will work.... but only 220v charging will work....... but I've been wrong before too.



................oztules

[Neil]

Hi oztules,  I see what your saying makes sense, I think i'll just try to fix this Vanner I have (tsc24-4500d) hard to find parts for it though.

Thanks for all your input.

Neil

[clockmanFRA]

I have a decent wall mounting steel Enclosure case with hinged door, (it was a very good price), 24inches high x 16inches wide and 12inches deep.

My design for the contents, and the important COOLING flows.

I have 3 fans at present, two will blow in fresh air and one at the top will suck up and out. I have a vent at the top as well, but not sure if this should also have a Fan for sucking up and out.?



 

 

[oztules]

See anotherpower site re fans Clockman.
Also from that diagram, i would place a fan at the bottom of the heat sink as well as the top one you have shown Use push pull fans on the controller cards for much better results. Use the speed control boards for the fans not the bang bang  for very much better results.

The caps cop a lot of ripple, and 60000uf needs cooling under those conditions for long life. Life expectancy is shortened exponentially with heat for caps... very important for long service. Fets can take 150c before failure.. caps cant.

A gentle constant gentle breeze from the controlled speed fans will be very useful for the caps and fets to stay cold 90% of the time... better than heat up and bang bang cooling. They will only turn on full when they need to.



...............oztules

[boB]

See anotherpower site re fans Clockman.
Also from that diagram, i would place a fan at the bottom of the heat sink as well as the top one you have shown Use push pull fans on the controller cards for much better results. Use the speed control boards for the fans not the bang bang  for very much better results.

The caps cop a lot of ripple, and 60000uf needs cooling under those conditions for long life. Life expectancy is shortened exponentially with heat for caps... very important for long service. Fets can take 150c before failure.. caps cant.

A gentle constant gentle breeze from the controlled speed fans will be very useful for the caps and fets to stay cold 90% of the time... better than heat up and bang bang cooling. They will only turn on full when they need to.
(Attachment Link)


...............oztules

If you have a 60,000 uF cap getting very hot, (how hot ?), it is probably a high ESR capacitor.  Can't you find some lower ESR caps ?  Is it some old computer capacitor ?  Is it even a 105 degree C capacitor ?  They say that capacitor life is shortened around twice by every 10 degrees C rise.  I think the problem with air-cooling capacitors is that the center of the cap may be much hotter than the outside due to thermal resistance.  I should certainly help though to cool it.

boB

[oztules]

No boB,

At the current time, the usage is hovering around 700w.. the temp of the heat sinks is 24C and the caps the same. Fans are not turning, and won't until 29c, then they will start to circulate some air. They are 120mm.... ambient temp is about 16C

I have never seen the caps hot, and I don't want to.... and I figure the best way to make sure that is not likely, is to have the auto fans cool effectively..... and keep the fets and caps as cold as humanly possible. I see no advantage in letting the temp get up over 25-30c if I can silently stop that happening.... yes in hot weather I may have to change the set point......but we have very few days over 28c here.

Your comment regarding the life of electro caps is exactly why I have done as I have done, and seems like a very good reason to be a bit fanatical over their temps.

( For other folks ( not boB) The anodes are metal ( Al ), and the cathode is electrolyte.. picked up by another metal foil(Al) so you can connect to the cathode electrolyte.... so keeping it cool starts to make sense... it is a bit messy when it gets outside the cap from temp)

 I have repaired too many pwm devices that have been blown up  or won't start because of cap failure... seem to be developing a fear of hot caps and pwm systems

You may have noticed two temp gauges measuring fets and tansformer on the front panel.... yep bit over the top, but we are still experimenting really. ( my excuse)

Transformer is currently 33C, and it can look after itself as far as I am concerned... it is big and brutal, and can stand high temp without compromising it's life span .... having said that I have never seen 60C even at 8kw for 10-15 mins during normal morning runs ( usually around the 3-4kw for a few hours between 10 and 12am.)

The caps are  -40c to 105c 10000 80v units. ( 6 units).

Never heard of the manufacturer Mson or something similar... but they seem to work fine....... low enough esr to not be causing any concern thus far.


.........oztules

[jan04]

Hey guys,

greetings from good old germany. I'm following this thread and your adventurous ideas for a while now and therefore, I got animated to do something,too.

 I just had a simple PV installation: 3KWpp PV, 1050Ah 24V Battery and a 2KW inverter from Outback. Now I extended this PV up to 7KWpp, but the Outback doesn't last very long. Thus, I needed a simple solution.

I connected a PJ Inverter 10KW 48V and it was fine.

Then, I tried to build a practical Inverter for continuous 3KW power from a new PJ ( I really got it for a few nuts, 3KW, 48V )

I already had a big toroid suitable for 4KW continuous power in my garage and it wasn't very hard to equip it with enough copper wire. So far, so good.

Yesterday, I started a run for the new 3KW PJ Inverter equiped with the new big toroid and tested it in some different modes.

Then, in charger mode it was able to be switched on without any implications , sent 58V into the batteries, everything seemed to be ok, but then – there was a sparkling flash and everything was down. I removed the circuit board and yeah, as I thought – three FETs nearly exploded.

On a second FET-circuit board  I detected two FETs, which weren't even soldered. ( see photo )

Thus, at this occasion I want to replace all weak IRF3710 FET with the more capable  IRFB410.

When I'm done with this, i hope I will be able to get the continuous 3KW, as I desired.

But that means I have to “drill” the circuit board up to 5KW. I think I will replace the ADJUST resistor, so that the controller thinks there will be a 5KW inverter. A jugglery.

I tried to measure the resistors on the 10KW and on the 3KW board. ( 18 Ohm and 30 Ohm), but I have the strong suspicion that these measurements can't be true.

 
Does anybody of you know the measurements for 5KW, or 8KW/10KW??

jan

[oztules]

The 8kw is 39R. It is difficult to measure when in circuit as they are in parallel with both the ct secondary and another resistor underneath.

I don't know what the other boards have, my 15kw is in use, and they tend to use two resistors stacked one upon another for all but the 8kw unit.
It is simpler and removes some stress from the CT on the smaller boards if you just bypass some current around it..

Use 6 fets per bridge , 4001 will be fine. remember to check every resistor in the gate lines.... they should have blown up... fair chance your control totem transistors and opto have gone too.

Battery charging is high risk if there is any chance of losing the AC input even for a fraction of a second.... I just won't do it any more.

Wondering if they didn't solder the legs.. ( unlikely really... even with their poor QC), or did they heat up redhot when the blast went through... looks a bit that way, what does it look like on the top copper layer... clean unused, or signs of hot solder, and are the legs a different color. If it all looks virgin board and legs, then yep, they missed them.

Not sure if that earth wire ( 4 in hand??) has enough copper in it for 3kw continuous. ( at least 60A).. prefer at least 30mmsq, and then air gap separated... tight wound like that is a bit scary unless they area lot  thicker than I think they are.... monitor the temp, and see how it goes I guess.


..............oztules

[jan04]

tnks oztules for your conclusion.
you are right the 4001 will be good enough,  but i have about 50 pcs of 4110 here, why not using them?.
unsoldering legs.. i have looked your way.. indeed i was a bit to hasty in my investigation on the run. the legs are in different colors and underneath the FET-Board i can see little soot spots. it seems that the FETs have had a rough time :-)
yea, i gave up the charging-idea, but what now? How can i to support my batteries?
i didn't make good experiences with net switching on and off, because not all my home devices are capable to that. what is  the alternative? Unfortunately, i have to admit i didn't ever really understood your alternative with the grid tie, painfull for me..

my toroid primer  winding ist 2 in hand 4mm sq cupper wire and 5 in hand 4mm sq earth wire.
i will strictly investigate the temperature-development, and i will also install one of these proactive cooling systems, just like you said.

jan

[oztules]

Jan04,
Sorry for the 4001 thing... meant 4110. This site does not allow editing for very long after you post,  ( were are not adult enough to take ownership of our posts apparently ) so didn't change it because i couldn't when I looked at it... so irfp4110 3.7mo 100v 180a fets... you have the ones I would use right there... figured you knew ..... and suffered the same problem when you put
"Thus, at this occasion I want to replace all weak IRF3710 FET with the more capable  IRFB410." .... I figured you meant 4110.

If you use EI transformers, you can use the charging fine, but loose too much in the magnetising current for 24/7 operation. ... perhaps build a cheap charger from new boards and big EI transformers from old welders.... clunky but effective.

The grid tie is just using a grid tie across the inverter output. That charges the batteries very well, without the switch off problem emerging... but you have little control of the process unless your grid tie is reasonably sensitive to voltage pertubations... then you can use that to switch it off... it is a bit bang bang in operation, but effective. ( when batteries full, AC voltage will rise, and then at 250v or so will switch off... wait 3 mins and try again.

This is actually pretty good for flooded lead, but no good for most other technologies I suspect..... especially lithium.

I have a battery voltage switch that isolates the grid tie AC line when batteries reach 58v, ( 3v hysterisis ) it them reverts to the normal solar controller from the other arrays to finish off.

Do monitor the transformer voltage, there is an awful lot of plastic insulation holding the temp up, and stopping the core cooling very well.

None of my home devices net switch, and they never will. Time clock switching of the hot water is as technical as it gets for me.... see no reason for anything else to selectively operate here.

When things go wrong in the fets, it has to continue going wrong until all paths from batt+ to batt- are physically blown apart or otherwise made open circuit.... the batteries see to that very effectively and so I normally replace every fet and gate resistor.... sometimes the noise lasts a few seconds as the cascade goes through the fets.

This normally means that 50-60 volts or more ends up in the totem pole driver transistors, and they suffer the same problem... sometimes they burn the board underneath them as well.... the opto's cop the same thing too. The lm339 seems to always survive ( totem drivers)... and some times the pwm supply looses a diode or pwm chip.

Good luck with it

...... oztules

[clockmanFRA]

Hi Jano4,

I have grid, but I use the Cheap night tarrif only, and in winter months when the sun is poor and I have no wind, I use this 100amp charger, by ELTEK.

Its a used /second hand, but nice gear, the rest of control gear in the cabinet is for its automation, time switch, low battery switch on/start etc.



I have set it for float voltage charge at present....



And importantly I can also set the Amperage output to the batteries, and that's NICE!

[jan04]

oztules
hmm!? EI transformer?  ancient wisdom.. not to bad.. i have to reflect this..  btw i have one very very older welder to be lying around ;-)  will check the technical specification, tomorow.
magnetising current? it is no matter, because i would like to start the charger only at odd time, for giving the batteries a bit of assistence.  the arduino/raspberry system have to monitor voltage and power usage and make a dessision to start or not to start the charger.
ok, it is like you said, clunky but effective, i think so

tnks for detailed  explanation of your grid tie concept

jan

[jan04]

clockmanFRA

tnks for your information. i will have a sharp lookout for one suitable second hand charger. i don't know, maybe this is the one of easiest ways...

jan

[oztules]

Jan04
Here is a story i did using the W7 and EI transformers. You can see the size of them.
http://www.anotherpower.com/board/index.php/topic,780.0.html
Note the control and power cards... same as the PJ... part  for part for most of the boards.

The torroids are very stiff electrically, and the control card does not drift back to 50.0hz.. it snaps back.

The inrush currents of those big EI transfromers is not enough to bother the fets,... but  the inrush into the torroids however is enormous, and the fets fail on switch off of mains.. they drift on nicely ( you have seen this part), but they don't drift off mains nicely. Thats the only explanation I can offer as to why they don't survive with big torroids in the charging ... unless you can switch off the inverter, then switch off the mains, then switch inverter back on... that works... but livin on the edge.....


............oztules

[clockmanFRA]

Hi oztules, I converted your PCB layout to Bread board, last night..........

I have about 2/3rds of the bits, just awaiting the LM jobby.

I think my layout is okay?
But please shout, if I have made a mess.

 

[oztules]

Looks like it should run.

It is a very simple use of a quad comparator chip I have lots of. We only use 1 in this case, and I was surprised at how well it worked... so I built some more.

The comparator compares two voltages, and if the + input one is higher than the other the output transistor is turned on, if not then off.
So we have an open collector output to deal with....... which means two things.
1. we need to supply it with voltage on the collector ( via 10k )
2. when the +>- then the transistor is turned on and shorts the collector to ground.... so on is off and off is on.

To see how it works,
1.  we use a 10k and a 3v zener to establish a fixed reference voltage on pin 7 ( +input) This gives us a set point that we can use to compare the temp probe to.
2. We set up a voltage divider using the 47k, the 50k pot  and the temp probe all seriesed with Vcc to ground.... fed to pin 6 (-input)
3. When the NTC temp goes up, it's resistance comes down, pulling the output of the 50k pot lower and lower
4. When it gets lower than pin 7 voltage, then the comparator will switch off the transistor ( which was shorted to ground when on), and the 10k pull up resistor  ( Vcc to pin 1 output) will be able to deliver voltage to the now open circuit collector on pin 1... this voltage then fed to the fet to switch fan on.

Now thats as a comparator, it is on or off..... if we introduce negative feedback into the - input ( pin 6) then we alter the stage gain, and the more feedack we introduce, the lower the gain, and the sharp on/ off behaviour of the comparator is made wishy washy.... thats tech talk for it will now switch on and off before and after the set point, with a voltage slope rather than hard on or off..... this translates to the fet as more or less voltage to the fan/s.
The feedback is introduced at pin 6 via the 50k pot on the right. It takes some of the output through a 10k and then a 50k pot back to the input of pin 6.... so we can change how hard and fast the fans turn on.

I set mine so that as it approaches the set point, the fans wake up and start rotation, silent and slow, then as the set point gets closer, they speed up , if they can stabilise it at low speed then good, else it gets hotter, the fans spin faster until you can hear them... all the way up to full on at the set point... so feedback pot for slope rate.

So thats as simple a fan control as I can come up with, and still work nicely. It could be made a lot more flashy, but it does the job.

Each comparator looks like this. It may help to understand the output inversion......



................oztules

[clockmanFRA]

Thank you yet again 'oztules' for your fine description of whats going on in your circuit, excellent stuff.

I am waiting for the 'Facture', (postman) for a few components.
My Oz Inverter is being fitted into its enclosure.

Okay 'oztules' the million dollar question........

WITH THESE  Power Jack 15kW boards, whats going on,? and where ?.

We have some sort of quartz oscillator, ? that gives us 50HZ.
We Have the FET boards and capacitors, we have the control board.
Somewhere we have Totem poles. ?

Who is kicking who? and what is giving what?

Big ASK I know, but just a brief block diagram on stages about these PJ 15kW boards, would get my head around things.

Many Thanks.

[Neil]

Hi oztules    would you recommend using the same toroid clockman and your self made if using one of 15kw boards, I would need to make two one for the 220v  and one for the split phase as you described to me a few posts back. I'm suspecting the second one will need to be made a little different then the first.

thanks

Neil