smps....what am i doing wrong? :(

[52613 sfw356645eh5ah6j75]

well still doing the smps part of my fan controller i encountered a little problem.....dont know if it's the transformer, that's why i'm here , but something make my mosfet heat a lot and that means bad efficiency, aka sub 50% with little loads.
i used a smps transformer from a 200w atx psu, just the opposite way, so i switch the 12V input, and use the 200v input as my output, flyback alike.

pictures:
**broken link removed**
**broken link removed**
**broken link removed**

and the oscilloscope screenshots (gate voltage)
**broken link removed**
**broken link removed**
**broken link removed**

the sg3525 drives the mosfet directly, as you see about 130ns rise and fall, the grounds are super reinforced, from "main power" (12V 2A regulated psu) goes to LC and then to drain, source to ground, then secundary is rectified and referenced to ground....
the thing is, as said above, the mosfet gets hot, with a 12V 0,55A load (cant even handle 6W u.u), it drains about 1,34A from the 12V psu, the transformer doesnt get hot at all.
Tried varing the switching speed from 50k to 150, with very little change in efficiency, with less load it reahes about 74% (150mA 12V output)
is it all switching losses or am i doing something wrong, cause the atx psu i took the transformer from didn't had problems at all delivering 12V 5A. or is it because it doesn't work to rise the voltage?
i'm really lost here, even conecting it directly to a 12V 12Ah battery didnt change a thing, only increacing the frecuency to about 80khz seemed to make it output more power but even with the batery i coundln't get more output than with the 2A psu.....
the only thing i think could be a problem are the little bumps after opening the mosfet.....but is tit because of that?

 

[crutschow]

Post a schematic with part values. Otherwise we are flying blind.

 

[52613 sfw356645eh5ah6j75]

i dont have the squematics, made it on the fly, and it's not like i can make it either, the transformer is unknown in every aspect, and i dont have a lcr meter , then everything is very simple, the SG3525 pwm controler controls the mosfet as shown in the screenshots (45A 0.017ohm) , that controls the transformer and the rest is the mur820(25ns 8A) diode in the secundary, the 200V 220uf cap and the load
made this, hope it's enought http://sketchtag.com/KSFsGuKIRk
the sg3525 has a 100k pot so i can make it go from <20khz to ~200hz
i really dont have much more, the circuit is pretty much a generic isolated flyback converter, and the transformer originally worked at 50khz converting fro 311dc to something above 12V for the psu

 

[ronv]

From your schematic it looks like your FET is backwards. If that is just a typo try putting a 0.1 ohm resistor between the source and ground so you can measure the current. It could be the inductor is saturating.

 

[Diver300]

The transformer secondary should be the other way round.

**broken link removed**

(Well you can reverse either the primary or the secondary, but it's wrong as you have drawn it)

 

[bountyhunter]

You have to have a plot that shows the FET current as well as voltage, that way you can calculate the total power losses. There are only two kinds of power: conduction (ON state) and switching losses (turn on/turn off transistions) and the turn off power loss is always higher power than turn on.

You can inset a small resistance in series with the FET or use a current loop and current probe.

 

[52613 sfw356645eh5ah6j75]

yes that schematic was horrible, xD
the mosfet was in the wrong way, but the transformer was right, both points are in the same place so it's in phase or something like that, as far as i have read about that "point convention" in wikipedia. if i connect the secundary the other way, i get horrible spikes and the mosfet gets really hot. and the output power is really low. (even worse)

i'll see if i can make a 0,1ohm shunt to meassure the current.
i made a toroid transformer with a toroid from a filter or something, no idea what material is it, 8:25, with 2 fine wires for each winding, 60khz, the secundary in anti phase, seems to work way better than the other, about 66% efficience. could stand about 14W just getting almost warm, so it seems it was the tranformer the problem.

There are only two kinds of power: conduction (ON state) and switching losses (turn on/turn off transistions) and the turn off power loss is always higher power than turn on.
werent hysteresis, eddy currents and magnetostriction also losses?

Ill do the shunt and keep readding

 

[bountyhunter]

My answer referred to the OP's statement:

the thing is, as said above, the mosfet gets hot,

In a switcher, the FET has only two kinds of power: on state conduction, and switching losses.

There are certainly other power losses in other components.

 

[TechnoGyppo]

As mentioned above, you need to make a shunt to measure primary current. You need to be able to adjust the duty cycle of your FET on time as well as the frequency. To calculate the maximum power available you need to be able to observe at what current your transformer starts to saturate.

From the knowledge of your supply voltage and time taken to reach (just before + a small margin) saturation, and the value of the current at that time, you can calculate the inductance. Using the inductance and current you can then calculate the stored energy per pulse. The maximum frequency will be determined by the time the transformer takes to reach the peak current. If you calculate the frequency from this time, and then half it, and multiply by the energy per pulse, you should have a comfortable level of power available.

If you look here: https://dos4ever.com/flyback/flyback.html you will get a good insight into what I'm trying to say.

I'm not sure what voltage out you are going for, but with the diode you are using (200v) I would say that the maximum output voltage would be a fair bit lower than this before the voltage rating of the diode is exceeded, the exact value being determined by the (unknown) turns ratio of your transformer.

 

[52613 sfw356645eh5ah6j75]

well back from vacations, sorry for the big delay.
i did the shunt, used a CC batery charger i made to calibrate a piece or utp wire to 0,100 ohm.
conected it betwen the mosfet source and ground, and it seems to me (no idea) that the problem with the flyback + unknown tranformer is the mosfet voltage is not enough to withstand the spikes, as i get constant reverse current peaks above 20A (2V+ with 0,1ohm is 20A isn't it) for some ns.
also the transformer is nowhere near saturation, it start to saturate below 12khz.
Tried adding a 600V UF (50ns) schottky diode in series (no idea if this works like this) with the mosfet (before the drain) but the reverse current didnt go, it just helped a little after the end of the peak.(but worsen the efficiency obviously)

the shunt
**broken link removed**

saturacion point
**broken link removed**

the reverse current peak / bigger
**broken link removed** **broken link removed**

no idea if it's the voltage if the mosfet or what, i still have to make myself something to meassure the inductance of the transformer and stuff, goona read that about flybacks as seemed better than what i've found and also gonna see what transistor the original driver used althogh the original was from 220 to 12(not sure if 12) but anyway gonna take a look
also, look at the ch2 meassuring, it says 920mV rms, i don't believe my oscilloscope to be inaccurate but isnt 0,92/0.1 about 10A? my supply delivers only 2,5A before the SC protection kicks in.

ps: the idea is 100W at 40V output, the diode has 200Vr (but i reached about 150V before it started leaking)

edit: yes, the original transistor that drove that transformer were all above 700V npn power transistors (cheap chinese atx psu x_x)