Flyback Transformer Meltdown in overload condition

Hi,

I'm working on a flyback power supply design. It runs fine at normal load but in overload conditions the transformer windings go into meltdown.
The output voltage drops and the duty cycle decreases as expected but that doesn't help the transformer.

I know I haven't exactly provided much information here but I was just after getting a general idea of if this is a common problem and what the mechanism is.

I assume that you have found this condition -I hope- during validation testing at not at a customer application.

What the failure essentially means, is that the product has no margin, or that abnormal conditions were not considered in the design phase.

Usually a design requirement should indicate that a unit must survive an abnormal condition, consisting of a load of yy amps, at an ambient temperature of xx degrees Celsius, for xx period of time, with an input voltage ranging from yy to zz volts.
This requirement is not optional on a power supply.

But back to your problem. A wire melts down because of the I2R heating exceeds the wire's capabilities at a certain surrounding temperature...which I may add, inside a PSU case may be many degrees hotter than the ambient itself.

First step: can you improve your current limit performance? Depending on how you accomplish this critical function, I've seen poorly designed PSU with a short circuit current 4 times as large as the rated current. This is specially true if you are only relying on a basic primary-side current limiting.

Second step: increase the wire cross section. This most certainly means that the turns will no longer fit inside the core window area, which will require a larger, heavier and more expensive transformer.

Yes its during validation, not originally my design either.

Its just relying on cycle by current limiting by the primary side I sense controller pin.
Looking at it in simulation (I'm not exactly a power supply expert) it appears that it would be the secondary side winding causing the problem as this ends up conducting most of the time a high current as the duty cycle decreases(?)

Having personally been in your exact same situation, I understand your pain and frustration.

Good electronics design is both a science and an art in which one must balance three conflicting requirements: Cost, performance and reliability.

If the company you are working for is similar to the companies I've worked for, cost is almost always the primary concern, it has a fixed maximum value one cannot exceed.
Therefore, one has to juggle performance and reliability. It takes significant ingenuity to achieve all goals. With margin.

Sometimes, if the reliability is really poor, one can wrestle with lots of difficulty the justification for a higher cost.

BACK TO YOUR PROBLEM
Sometimes I've seen that designs add a small series resistor with the secondary winding. The idea is that you transfer some of the dissipation outside the transformer. The series resistor has several drawbacks, but it is very low cost. If you still manage to pass the other performance requirements, hey! go for it.

is it an offline design?
Most offline flyback controllers have overload protection and either latch off or go into hiccup mode in case of output overload.
The controller simply checks the feedback pin and if it rails high (usually its high), then it knows there is an output overload and it switchs off.

So which controller are you using?

You can of course put output overload protection in yourself with external components.

When a flyback goes into heavy overload the vout falls well down and then you can get staircasing of the inductor current...
so yes you shouldnt really be rating your flyback to handle overload current...you shoudl get it to latch off in case of overload..or go into hiccup mode where by it starts up and shuts down repeatedly till the overload is removed...in hiccup mode its not on for long overall so theres overal low power dissipation.

Its 48V in. The controller is the LT3758 (doesn't have any hiccup type mechanism).