Author Topic: Building a 6kw pure sine wave inverter using power jack boards part2 the guts..  (Read 32363 times)

0 Members and 1 Guest are viewing this topic.

Offline MarNet

  • Full Member
  • ***
  • Posts: 60
  • Karma: +0/-0
I mean bobbin*

Offline oztules

  • Moderator
  • Hero Member
  • *****
  • Posts: 1177
  • Karma: +105/-8
  • Village idiot
The 8:1 thing is a product of head room and the properties of a sine wave. Our 240vAC is really 320v with .7071duty cycle... to get our 240v rms ac.

So to make 240v ac from a DC source we need to handle the 320v part of the curve as well...... so our DC side needs to be .7071 of the expected AC wave generated in it or for 48v about 48X.7=33v.... so we need at least as low as a 33v primary to get to the peaks needed to make the sine wave....... now with losses in the switch, wiring, sags to cover high power surges etc, we need another 15% safety margin so we are reasonably assured of getting our 240vrms at all times.

So now we can see why we are using about 8:1 for our transformer. We are really winding a 48Vdc primary (30 odd volts AC rms sort of thing) to make  320v peak ac or 240v AC rms. ( the rms is the root mean square... which is equivalent to  the DC heating value..... so the power in 320vpeak ac sine is equal to 240vDC in raw heating power ( no power factor)... or 240vac as we call it for the mains. If you rectify and filter... you will measure 320vdc.

Power in the PJ. 4110 will easily deliver 8kw if you have 24 of them @ 48v system..... it will run 4kw loads all day if you supply your own transformer, and cooling... and way beyond 10kw for shorter runs. 8kw seems to be ok for 20-30mins ON MINE..... mine has it's own cooling circuits, and big transformers.

As an example, it was very hot here last few days... 33C, and yet the inverter transformer ran at 60C and the heat sinks at 38c while running the hot water (2500w) for a few hours. No fan on the tranny, and 5 inch push pull on the heat sinks, and ran the 1.5kw water pump for an hour or so at the same time. surprisingly low heat sink temp.

I think because it is mounted vertically in a tall enclosure ( with venting top and bottom), so the natural convection is enhanced by the fans.... but why ever it is, it is better than the one I built with horizontal board fixture.... that would probably be in the higher 40's with the same environment.

The boards are good... pj figures and design.... not so much

They are cheap for a reason.... thankfully easy to rectify.


..........oztules
Flinders Island...... Australia

Offline MarNet

  • Full Member
  • ***
  • Posts: 60
  • Karma: +0/-0
Thanks for the answer oz
My system is 24v, I'm thinking to get the 6kw or 8kw board.
I only need a 3-4 kw pure sine with a peak of 8k let's say.  I use it for water pumps. If the pump is 1kw, i need a safe 7-8 kw to start it.  And if i spend the money, let make it counts.
I'm thinking about 40-45cm square yoroodal core with secondary of 3x1.8mm wire. 1.8mm wire can deliver 10.5A, at 230v is 2,3kw so i can have a spoke of 7-8 kw right? And a decent 2-3kw(maybe 4) running.
What do you think?
Thanks again for the answer. 

Offline MarNet

  • Full Member
  • ***
  • Posts: 60
  • Karma: +0/-0
So you have 24 mofsets x 370w dissipation power in a 15kw board. That means 8.8kw maximum capacity.  Where is the rest of 6.2kw? Pj warrantee 15kw right? :) And they warantee 60kw peak. Wooow :)
I see why you get 15k boards for 6k inverter.

Offline oztules

  • Moderator
  • Hero Member
  • *****
  • Posts: 1177
  • Karma: +105/-8
  • Village idiot
No.......
In theory, a single 4110 can carry 150amps@100v... so a single one can actually do 15kw on it's own

Practice is different, those skinny legs can't handle 100 amps continuous, apparently the silicon can, as it has a resistance of only .003 ohms.... so at 100 amps it will drop E=IR or 100x .003 or .3v losses@100amps.. so the dissipation will be W=ExI or 100 x .3... or only 30 watts.. so that leaves over 340 watts of dissipation spare.

Can you see that the dissipation figures are not the power handling figures, the 370w is the figure it can safely dissipate in losses..... not throughput.

It's losses is  the Rds on in ohms  (.003-.006 for 4110) x the current flowing through it... these losses are dissipated in the mosfet and heat sink.

Switching losses are added to the IR losses,( and in anything I build usually outweigh the resistive losses hands down). Getting the switching losses down is a science unto itself... black science almost, add in parasitic oscillations and lots of other gremlins, and we have our work cut out to keep the losses less than the 370 watts. If we fail, it goes up in smoke.

For the most part, if we keep the switching frequency low, stick to the design rules regarding layout etc etc etc... we do pretty darn well.

With 6 fets per leg, we get roughly 150X6=900 amps of capacity X the voltage we switch... 48X900=43000w or more.. my bank is usually 56v so 56X900=50kw. ( so it should easily survive transients of this level)

Now these figures are all theoretical, and have no real place in the real world... except for fractions of a second, and there they count.

We switch at high speed........ over 20 thousand times per second... so we don't need to carry these currents for long, and our duty cycle is less than 100% (each leg is turned on only at a max of 40% of the time I suspect).

So in fractional times, 60kw is not out of the question... and surviving it is possible too.

Remember of your 24 fets, you can only calculate for 6, as each leg only has 6fets..... not 24.

I use the 15kw boards because the fet spacing is greater, and so cooling is better..... mostly not of interest, but more surface is always better than less.
Electrically they are virtually identical power boards. Physically they are better spaced in the 15kw version.... ( and  there are two more filter caps  on the bigger board... helps with transients)

3-4kw cont  in 24v systems is pretty heavy going. Your current will be around 40amps/kw on the primary.... or about 170amps for your 4kw..... thats going to mean very thick primary. I don't like 24v for this very reason, 90mmsq wire would be useful at these levels if you want 4kw cont. 3x1.8mmsq  is fine for the secondary

Transformers are only as good as their cooling. Thats what defines their power rating really. The little PJ transformers can do  very impressive start ups, and short runs at very high power, but not for long. They will do the output they claim... but not for long. They limit the big figures to about 12 seconds ( so it does not vaporise),peak for much less, and if you read the figures on the unit itself regarding what you can run,  it is quite realistic... but way less than the headline figures.

The size of the transformer is a fair indication of it's power handling. I found 2800mmsq of steel is about 1 turn/volt ( from memory). We only use bigger steel because we need bigger wire to keep the copper losses down... so we need to be physically bigger. A big core can also help with cooling, and  evens out the transient temps as a heat bank.

Have fun with it, I did.



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



Flinders Island...... Australia

Offline MarNet

  • Full Member
  • ***
  • Posts: 60
  • Karma: +0/-0
Much clearer now.
Regarding this toroidal core:
http://www.airlinktransformers.com/toroidal_cores/ros217
Will be ok for a decent 3-4kw transformer?
Thank you

Offline oztules

  • Moderator
  • Hero Member
  • *****
  • Posts: 1177
  • Karma: +105/-8
  • Village idiot
Not sure of the measurements.. is that a 100mm hole or 70mm...... You will need (want or hope for) a 100mm hole... easier to work with. There is enough steel there to pull some out of the center and still have a decent core if necessary.

If you can fit the wire, then fine....plenty of mass there.


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


Flinders Island...... Australia

Offline MarNet

  • Full Member
  • ***
  • Posts: 60
  • Karma: +0/-0
Basically the measurements are like this:
External diameter,  internal diameter, and height.
I read somewhere about the way you calculate the size of toroidal core. So should be the ???of the transformer rated power.  For 6kw transformer sould be ???6000, 77cmsq. With a core of 77cmsq you can have a 6kw transformer without expecting any extra heating or any surprise. Ofcourse here depends about the quality of the materials you use as well...


Offline oztules

  • Moderator
  • Hero Member
  • *****
  • Posts: 1177
  • Karma: +105/-8
  • Village idiot
So thats about 4500msq or by my rough reckoning about 1.6v/turn.

Should get the wire in there. about 150 turns on the secondary x 3 of 1.8mmsq wire ( 7mmsq)...and then 9 - 10 turns of 90mmsq cable.
112mmsq is the right size for the primary if we wanted to balance the two windings.... copper wise..... but you will not be running full power all the time, so 90mmsq will be fine.

Even 50mmsq on the primary would work fine 90% of the time. ( don't mean duty cycle... just normal use)

These are just my rule of thumb guesses, someone who knows the real equations will probably get a better handle on it, but those figures will work satisfactorily from my previous experiences.

"I read somewhere about the way you calculate the size of toroidal core. So should be the ???of the transformer rated power."..... That seems pretty fair for  100% duty cycle @ 50hz.....
Now I just need to remember it....





...............oztules
Flinders Island...... Australia

Offline MarNet

  • Full Member
  • ***
  • Posts: 60
  • Karma: +0/-0
I'll call the company monday. I'll ask them what power/cmsq they're toroidal cores can handle.  Its just so confusing.  Some companies selling 21cmsq toroidal core and advertise them as being 3000va.
By the way, the PJ power/control board from China didn't come as combined postage. I have to pay 30 usd postage for each one....
Any other merchant for the boards?
Thank you

Offline oztules

  • Moderator
  • Hero Member
  • *****
  • Posts: 1177
  • Karma: +105/-8
  • Village idiot
"I'll call the company monday. I'll ask them what power/cmsq they're toroidal cores can handle.  Its just so confusing.  Some companies selling 21cmsq toroidal core and advertise them as being 3000va. "

The core size is not really a case of power:size..... if we had copper of zero ohms per foot of whatever size we had in our hand... then very high power transformers could be had very very small.

The magnetising current is fixed for freq and voltage for a core..... you cannot saturate it unless you change the hertz or the voltage.... so core size does not dictate power handling.... sounds wrong I know

So what does?

It is the copper loss that causes us to use bigger cores... we need the real estate to use 3X1.8mm wire and 90mmsq wire of the turns needed to do the job.
The bigger the core, the less turns you need for the same frequency and voltage.... so for the same winding window we get more copper in there... we can make a bigger window and get more copper in there.... and we can see where this is going.... to get more and bigger wire in there, we need to use more core to get more real estate to play with... and better still .... less turns of it too.... so bigger wins.

If we had super conductor wire, then this would be  mute. We could use any core for any voltage and freq ( 50-60hz etc), as the wire could be thinner than hair, and carry a 1000 amps, so we could have our 2000 turns on a tiny core, and handle huge power.... remember saturation is not from current or over loading... so once the core is magnetised for that voltage and frequency, all extra amp turns involved with the transforming process do not direct their extra amp turns into saturating the core, but rather using their MMF against each other, and inducing current into the other...but not the core... so if you look at the equations for core saturation.... no mention of current... anywhere.

The little PJ torroids do the job, but their copper loss heats them up too fast to be useful for more than a kw or two, and their sag will be pronounced... so we use a bigger core to allow us to use sensible sized wire to keep the copper loss to a value we can handle heat wise.

Where does this get me???? well core size helps us get to our objective heat wise. If we used silver wire, our core could be smaller, as the R is less for the same power handling.... and losses are I^2R... we need R to be tiny for high currents.... or we burn up.

A big core handles no more power... but allows us to use much more copper to get the losses under control..... then the transformer as a whole handles more power for longer, or if we have enough cooling.... indefinitely.... ie oil cooled, forced air cooled etc.

I guess what I am trying to say is that core size is an indicator, but the copper is where the bulk of loss comes from, particularly in torroids, hysterisis loss is generally  small for size, and eddy currents are small, as they use very thin laminate compared to EI ones. They tend to use high grade steel for torroids, as they are very expensive to wind to start with.

Some EI transformers deliberately run in saturation... but these are special cases....particularly ferro resonant ones

PJ boards are just W7 boards in disguise. There are plenty of manufactures on alibaba with virtually identical boards... but getting them to part with them is the problem.
I had a few discussions with several of them, and all will sell you the control cards for very much less than PJ..... if you buy or have one of their inverters... so you can try, $45usd seems to be what they would charge  from the discussions I had with them. With PJ, they at least put them on Ebay... cheap at one time , but they have wised up.

Nothing stopping you from trying the others though...... all the LF  chinese ones seem to be the pretty much the same.

Some folks here are going to try the EGS002 cards too. 6-10 dollars.


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


Flinders Island...... Australia

Offline MarNet

  • Full Member
  • ***
  • Posts: 60
  • Karma: +0/-0
I have 2 1.5kw inverters. Probably Tuesday I'll try to fix them.
Going back to the toroidal core, you think that 220mm external diameter,140mm internal diameter and 50mm height can handle 3-4kw but you need more turns /volt correct?

Offline oztules

  • Moderator
  • Hero Member
  • *****
  • Posts: 1177
  • Karma: +105/-8
  • Village idiot
Is that about 2000mmsq?.. ie core cross section is ((220-140)/2)*50.....
If so, then about .7v/turn @50hz is my guess

So for 240v thats now 350 turns for the secondary, we wanted 3x1.8mm wire before, but your using more turns now.. so we want thicker again... we've gone from 150 turns to 350 turns,,, so need twice the sqmm for the same handling.

See how it gets away quickly as we go down in size. With super conductor, no problems with the extra turns, but with copper... it gets tight quickly... same with primary.. now 21 turns... need twice the cross section really.....

So no, the core would be fine, but the copper would be difficult/impossible for 4kw cont at 50hz... maybe you could do 1.5kw-2kw with that core.

It is not the core limiting things, it is the copper


..............oztules
Flinders Island...... Australia

Offline MarNet

  • Full Member
  • ***
  • Posts: 60
  • Karma: +0/-0
Hi
I have ordered 2 toroidal cores 220mm x 140mm x 50mm.
Is the only big cores what they have in stock now.
I have contacted few W7 inverter manufacturer's on alibaba and they refuse to sell the mainboard and control board only..
I'm thinking to dedign a mainboard myself...

Offline oztules

  • Moderator
  • Hero Member
  • *****
  • Posts: 1177
  • Karma: +105/-8
  • Village idiot
My guestimate is 1.4v/turn. Stacks of winding room, so many in hand is not a problem.   ie 260v would be 185 turns... but gee plenty of winding space with a 140mm hole.

The power card is a soda, it is the control card thats the key.

.....oztules
Flinders Island...... Australia