230V AC to 12V DC @90 AMP

[madammim]

Hi everyone.

I am afraid i have started something i cant finish.

I am trying to make a heavy duty powersuply using a NE555 timer and a set of 25 IRF740 mosfet. The timer uses a 12 V external supply. I plan to pull out about 90 A at 12-15 V. But i run into some problems.

The transistors are running at 25 Khz, and the duty sycles is controlled from 50-75%. All of this is tested and working ok.

I am useing a homemade transformer made of 1 mm steel plates (50 pieces) with walls of 50x50 mm. On the primary side i use 280 turns of 1.4 mm wire and on the secondary side i use 23 turns of 2 mm wire, 6 of them in paralell.

I am useing a 10 A 700 V bridge and a 2200 uF 400 V cap to make DC. The negative side is feed through the transistors.

Ok. So the problem. When I start up this supply, some of my transistors break down, and I get 50Hz on the primary side, makeing my fuses blow. I dont know what happens first, 50 HZ or transistor breakdown. I have a theory that the startup current is so high it makes the transistors brake, feeding my transformer with DC, until the cap gives up, feeding the system with 50Hz. If so, how can i prevent this from happen? Any other ideas about whats wrong?

Thankful for all answers.

[MP]

You should post a copy of your schematic diagram so that members of the forum can comment on what might be changed.

MP

[madammim]

I am sorry.

Here are a sketch.

I can post a complete one later, but I need to draw it first. My original one have been changed some since I started this project...

When I first testet this I used a 60W bulb lamp in series with the transformer. I had 25 Khz out from the transistors, and was able to adjust the duty sycles. Then I removed the lamp and started it with just the transformer. I heard the 50 Hz humming for about 0.5-1 sec, and then the fuses blowed.

[audioguru]

Hi Madam,
Your circuit applies 325VDC to the primary of the transformer. If the mosfet conducts enough current then the transformer would be saturated and operate like a piece of wire.
Each time the Mosfet turns-off at 25kHz it has thousands of positive volts across it due to the inductive flyback voltage.

[madammim]

Thanks Audioguru.

You say that this setup wont work cause a high voltage flyback that kills the mosfets?

Is there any thing I can do to prevent this from happening?

Can I rearange it in any way?

I am not so experienced with such conections, but I think it shuld work in theory at least...

[audioguru]

You say that this setup wont work cause a high voltage flyback that kills the mosfets?

Is there any thing I can do to prevent this from happening?
Transistors that drive an inductive load have a reversed clamp diode across the load.
It would need to be a pretty fast high voltage diode.
If the transformer saturates due to having DC on its primary then it is a dead short to the Mosfets, so they would blow-up anyway.

quote]I think it shuld work in theory at least...

No way, man. Transformers work with AC, not DC.

[madammim]

Ok. But I am a little surprised.

I make 325V DC, cut it up to 25 Khz, feed this "almost AC" to the transformer.

I have calculated the N1_min to be about 22 turns at 20Khz.

I dont know where I got this wrong, but I am not so experienced. I found a scetch for a SMP supply on the net, rearanged and recalculated it, and started building it "my way".

Hope tou can take your time to explain a bit more, and maybe someone can give me a tip for a good WORKING SMP.

Thanks

[audioguru]

Does your DIY transformer even work at a frequency as high as 25kHz? An ordinary power transformer doesn't. It is a special art to make an audio transformer that works well up to only 15kHz. High frequency. high power transformers use a ferrite core.

If you don't understand about the flyback voltage of an inductor whose current is turned-off then maybe you should learn about it the hard way:

1) Hold both terminals of a new 9V alkaline battery with one hand. You won't feel anything.
2) Connect a small relay's coil across the battery. If they match, the relay will operate.
3) With the battery connected to the relay's coil, hold one hand across the terminals of the relay's coil. You won't feel anything.
4) Disconnect the relay's coil from the battery while still holding terminals of the relay's coil. OUCH!

If a reversed clamp diode isn't across the coil and it is driven by a transistor, then GoodBye transistor.

Did you calculate the switching time of the Mosfets which is lengthened by the limited output current of the 555 to charge and discharge the 40nF total gate capacitance of 25 Mosfets?
A 555 has an output current rating of only 200mA.
It takes 200mA about 1.2us to charge or discharge 40nF.
Half a 25kHz cycle is 20us so the Mosfets will be over-heating for 5% of the time.   

[madammim]

Ok. Thanks for taking your time, Audioguru.

You know I did a lot of research before I started this project. (Maybe I shuld have asked first here  :'()

I found that recomended maximum frequency for a normal steel transformer was about 25 Khz. (...use a frequench of 20-25 Khz due to no noise...)

But still even if I reduce the frequency to 10Khz (N1_min 43 turns) I will have the problem with the flyback voltage.

Do you think this is not do-able at all?

By the way, here is a data-sheet for the mosfets: http://www.elfa.se/pdf/71/07115835.pdf

[audioguru]

Hi Madam (really?),
You could easily make AC in the high frequency side of the tramsformer if it is center-tapped, and each end is driven with push-pull Mosfets. Then the transformer is never turned off so the flyback voltage is reduced significantly. The voltage peak is doubled across the Mosfets with a center-tapped transformer.
Then since the Mosfets on one side will need to have their drive inverted, you could use a SMPS IC as the oscillator, inverter and high current Mosfet driver. The IC will have built-in circuitry to avoid the moments when the Mosfets on both sides will be conducting at the same time.

Look at the high input capacitance of the Mosfet on its datasheet. It takes a lot of current to charge and discharge it quickly, especially with 25 of them in parallel.

[madammim]

Thanks again.

I am a huge Disney fan (as you?), and choose to identify my self with a lookalike...  ;D

If I have to follow your suggestion, witch I probably shuld, I will have to re-wire my transformer, buy new transistors, and make the occilator from scratch again... The time is not a broblem, but I am allmost at the budget limit as it is.

I have been searching for clamp diodes, and found some from my supplier:
http://www.elfa.se/pdf/70/07040900.pdf
http://www.elfa.se/pdf/70/07012081.pdf
Is there any chance I can use some of these or any similare?

[audioguru]

Hi Mr. Madam,
HotWaterWizard in drag? ;D ???

The tiny surface-mount diode would probably vaporize, and the transorb is a zener, not a diode. Use high voltage fast diodes with about a 5A continuous current rating. I don't work with kW stuff so I don't know any.

[madammim]

I have now ordered some diodes.

Will come back with the test results.

Thanks a lot for now.  :)

BTW I normally dont dress up in womans cloths, but if I ever shuld, I would brobably look just like her... :P

[Staigen]

Hey madame

Have you checked out how they do it in computer PSU:s( AT, ATX ) ?
BTW, steel plates for the core? Will that work? And dont it seems to be a very big core?


//Staigen

[indulis]

What your trying to do is possible, but not easy. The transformer type is unimportant as long as you remain within its limitations. Oh..... I forgot to mention, what your trying to build here is formally called a single ended forward converter. Since I don't see any active reset circuitry for the transformer, I'll assume it'll be resonant reset. In which case you'll need a snubber from the MOSFET drains to the input cap, or ground. It is more efficient to connect the snubber to the input cap as you recapture that energy instead of shunting it to ground. This is what's probably killing your FET's. Next, there is NO WAY your going to drive 75 FET's with a 555!!!!! Contrary to popular belief, it's not the gate capacitance that you need to worry about (although, it is part of it) when switching MOSFET's, its total gate charge that is of interest.

[madammim]

Thank you for responding Indulis and Staigen.

I think you got me wrong, it is "only" 25 mosfets... But I am not that experienced in calculating this so I asked people with this skills, and they said 25 mosfets on a 555 shuld work fine.
I think the wiering of my transformer is ok, but please check the drawing.
I am not sure what you mean when saying a "snubber". Please explain for me. (My english and skills...)

I have used B<0.3 in my calculations for the transformer, and the frequency is set to 25 Khz due to advices about the noice. But I can easily change it to 10 or 15 Khz.

I have put in the diode on the drawing for the circute. Will this work ok?

[audioguru]

At work, I sent Request For Quotes to transformer manufacturers for a 400W audio output  transformer with a -3dB response from 50Hz to 15kHz. A few samples turned out to be just power transformers with lousy response. One nearly made it and had a good price but the manufacturer didn't have an amplifier powerful enough to drive it. I took my amp there and talked to the designer of the transformer about how its 15kHz response was a little low. He said that he interleaved layers of the primary and secondary windings but will try bifilar winding where the wires are side-by-side. The result was excellent with a -3dB response at 25kHz. ;D

[Staigen]

Hi Madame

I think you got me wrong, it is "only" 25 mosfets

A 555 can NOT feed 25 mosfets, it can not deliver that much current needed to charge/discharge the gate capacitance!

I asked people with this skills, and they said 25 mosfets on a 555 shuld work fine

They did not have enogh skills! ;D

I think the wiering of my transformer is ok

The coils should be at the same leg

I dont think it will work! You must do something to the gate drive first! Then test, you will only loose 25 Fets at max! ;D ;D

//Staigen

[madammim]

Ok.

I am a little bit on thin ice here, but total gate charge for this mosfet is 63 nC.
Using this converter: http://www.translatorscafe.com/cafe/units-converter/charge/calculator/coulomb-to-ampere-second/
I find that the current is much lower than the max rating (200mA) for the 555.

I have also been running the 25 mosfets in my test setup, and it seemd to be running fine (messuring the frequency between  "D" and "+").

Can anyone please underline where I have got this wrong by doing the calculations showing the 25 mosfets are too heavy load for the 555.  ???

[Staigen]

Hi

I am a little bit on thin ice here, but total gate charge for this mosfet is 63 nC

I checked on the net, the gate capacitance is 1259 pF/fet, thats a plenty of pF when multiplied with 25! The 555 cant feed that much current needed for a fast switch!

//Staigen

[audioguru]

63nC converts to 63ns at 1A. That is 315ns at 200mA from the 555.
25 Mosfets in parallel will take 7.9us to charge and another 7.9us to discharge.
A full cycle of 25kHz is 40us. So the Mosfets will have roughly half the load's power heating them for 40% of the time. Only 8.6W each, no problem except the wasted heat is terribly inefficient. ;D

[madammim]

Thanks both.

You know your theory well enough to ruin my project  :'(

I will change my frequency to 10Khz. It obviously will give me a better chance for success both with the curciut and the transformer. I just hope it will not give me an annoying 10Khz tone.

If I still need to wire the two transformer circiuts on the same leg, will it then be any different if i put the 325V outside the 12V?

[Staigen]

Hi again Madame

You know your theory well enough to ruin my project
I will change my frequency to 10Khz

No, increase the current drive instead, with a pair of transistors! :)


//Staigen

[madammim]

Thanks a lot Staigen!

I have been thinking about somethings like that, but did not understand how to do it (did not think as far as using two transistors).

So there is no problem feeding my 2. transistor with "-" on the "gate" and "+" on the "source"?

[Staigen]

Hi

One way of doing it is like this: