Hi folks, i need help desperately. I've built a two-switch flyback, and gave it a 60w load 5A and 12V it worked fine for two minutes then the mosfets blew up and shorted the whole system.

What's wrong, could the transformer have saturated?

I'd really apreciate your help.

If any additional info is needed just say so.

Thanks.

Oh and i forgot to mention that all the parts seemed to be cold after the boom. (rectifer diodes blew up, too large fuse)

 

With this configuration, the duty cycle shall stay below 50% otherwise the a DC current is built-up until it exceeds the maximum MOSFET ratings. If the duty cycle is less than 50% then then current through the transformer when the MOSFETs are off has time to fall to zero. Thus, in each cycle the currents starts from zero and there is no DC built-up.

 

Thanks, i have an analog ammeter hooked up in series with the transformer primary winding. I wasn't looking at it during the accident... So it would have showed me if the current was climbing, or am i wrong?

Does the increasing current mean that the transformer is going to saturate?

 

Thanks, i have an analog ammeter hooked up in series with the transformer primary winding. I wasn't looking at it during the accident... So it would have showed me if the current was climbing, or am i wrong?

Does the increasing current mean that the transformer is going to saturate?

If the current was rising slowly you would be able to see it.

Yes, the transformer is going to saturate if the MOSFETs do not blow.

 

Yes, the transformer will saturate, and the MOSFETs will pop.

 

How do i know what's the duty cycle if i don't own an oscilloscope?

Why wouldn't i see the increse in average current after some time on the ammeter?

 

How do i know what's the duty cycle if i don't own an oscilloscope?

Why wouldn't i see the increse in average current after some time on the ammeter?

How do you generate the PWM?

You should be able to see the increase of the current.

 

Why don't you post a full schematic of your circuit? The schematic you posted earlier was just a conceptual flyback converter.

 

This is the full schematic, the pwm is generated by the 555, the potentiometer changes only the duty cycle, frequency remains the same.

Is something missing apart from feedback ?

Is the output stage configured correctly or do i need some additional parts?

If transformer saturation caused the breakdown would all the parts including the transformer be cold?

Thanks for the feedback.

 

What did you use for the MOSFETs (part number), and how far are the 680 Ohm resistors from the MOSFET gates?

680 Ohms is quite large for a gate resistor.

 

The mosfets are irf840.

By how far, do you mean distance?
If so their leads aren't and they're directly connected to the mosfets so i'd say about 2-3cm away.

Do you think the gate might not charge fast enough?
What problems could this cause?

 

If your IR2112 is that close, then reduce the gate resistors to 10R in case the RC delay is causing some cross conduction.

Did you run with the input supply at a low voltage, current limited level and check all the waveforms makes sense, and with no un-explainable current transients or voltage overshoots? You can check current waveforms by temporarily inserting a surface mount resistor of <1 ohm (aka the type used for current sensing).

Have you designed the transformer correctly (checked the inductances and leakage inductance), and observed a primary current ramp up to the max peak current level to see if there is any sign of impending core saturation and hence current spiking (slope of current ramp will change at onset of saturation for a constant voltage source).

Ciao, Tim

 

I believe what is being conveyed is the fact, it is almost impossible to trouble shoot one of these type circuits WITHOUT an oscilloscope. Meters will not show you transients. Analog SImpson meters can show you audio freq transients, but higher freq are to fast for any voltage meter.

If the circuit has problems with turn on/off times, that will be easily seen on a scope. I'm afraid doing without one may be more than just hard to do.

 

I'll the try to borrow a scope next week. I'll also try to recalculate the transformer and increase frequency, i think it was a little lower than in the calculations.

What interests me the most is that with a 12 w load it worked fine for hours.
After the test i touched all the parts and they were ice cold. Only the low side mosfet was almost unnoticeably warm.

Could the warmth mean that the turn on time for the two mosfets is a bit out of phase? What would this result in?

Do the problems caused by current transients and voltage overshoots increase with load?

Thanks.

 

If I am not mistaken, the 555 circuit you posted is able to go above 50% duty cycle, are you absolutly sure you dont exceed said 50%?

 

Yes, it can, the problem is i'm not sure if it didn't exceed 50%.

I also found some transformer miscalculations, Bmax at 50% duty cycle exceeds 5000 Gauss.

Is it true that the safe value for ferrite flux density (B) is 2000-3000 Gauss or is it lower?
Is there a rule of thumb in choosing this value?

When i get the oscilloscope i'll try to restrict the pwm circuit from going over 50%.

Ironically this power supply was ment to be used in a makeshift oscilloscope.

 

The flux density a core can handle depends upon the material the core is made of.

You'll need to look up the data for your particular core.

Your 555 circuit can indeed vary from ~3% to ~99% duty cycle in the output.

You might consider using a more traditional R1/R2/C1 circuit (pin 7 connects between R1/R2) with an NPN transistor on the output (pin 3) to invert the duty cycle. That way, your duty cycle will be limited to ~48% maximum.

 

The core material is unknown, i took it out from a power supply of some device.
Is there any way to know what material is used?

Could you draw a schematic of where the transistor is connected?

Thanks

 

Basically, this:

Duty cycle will always be less than 50% at the collector of the transistor Q1. Frequency will change somewhat, but will be fairly constant. I didn't bother to calculate the frequency.

[eta]
Add a 680pF cap across R5 so that Q1 turns on and off quickly.

As far as determining the core material, you'll have to run tests on it; and you'll need a 'scope.

Check out Ronald Dekker's page:
http://www.dos4ever.com/flyback/flyback.html

 

I already had a transistor hooked up i the same configuration on the original schematic.

I'll try to run some test on the scope.

Thank you all for the input