Flyback Converter

[whatty]

Hi, There
I am trying to build a flyback converter i am prototyping it in bread board the converter works fine as i am getting the exact output but after running for quite the mosfet explodes after researching the explodes and gate resistor is burned along with it the flyback converter ic uc3845 also burned. I am prototyping it on a bread board does it causes a large parasitic resistance on gate and high voltage is build up....? does building the circuit on perf board solves my issue...?

[Alti]

What is the peak voltage across the switch w.r.t. datasheet ratings?
Or rather, what is input voltage (peak), output voltage (peak) and ratio of turns?

[Someone]

I am prototyping it on a bread board does it causes a large parasitic resistance on gate and high voltage is build up....? does building the circuit on perf board solves my issue...?

Tough issue to solve when you dont know what is causing it. General solution is to run the converter at some fraction (say 1/4) of the intended load current and voltages. Measure every single node and path as best you can to try and find where the voltages and current do not match your expectations.

[TopQuark]

Show us the schematics, a photo of the physical layout and the way the flyback coupled inductor is constructed and specified. Show us waveforms if you have any. I hope you are not hooking mains up to your breadboard circuit. 

Breadboard is a terrible way to prototype a flyback converter. The high dV/dt and dI/dt involved in the circuit, plus the high inductance of breadboard wires means you will get voltage spikes and ground bounce all over the place. Perfboard may improve things if you are careful with the construction and use good layout techniques. I usually just use copper clad board and carve out copper islands for various connections, while allowing me to have a nice solid low-impedance ground plane. Careful construction and operation is required to prevent electrocution  . See a demonstration below.

If the MOSFET is the first to blow, check that you do have an adequate RCD clamp at the MOSFET drain. If it runs for a while before blowing up, it is possible that avalanche operation was just barely keeping Vds in check, before overheating and thermal destruction.

[BadeBhaiya]

If the circuit was working fine for a while and only failed during prolonged operation, its not unreasonable to think it has to do with heat buildup. The FET may have failed, the drain and the gate could have shorted and burned the gate resistor and the switcher IC in its path.

But of course, as other posters pointed out we would need more information.

[Picuino]

Test the circuit with low supply voltage and low output current.
Study all the important parameters to see if they are the same as the parameters given by the simulations (computer or paper ones).

Currents and voltages in power components are particularly relevant. You should also look at the overall performance, to see if you are having more power losses than expected.

Gradually increase the voltage and power to study how the circuit behaves and if it does as expected.

[whatty]

Hi there Thanks for your replay

I am sharing you the schematic and all the exact values used in the circuit.

Note: I laid the circuit as per the schematic on the bread board. I obtained the exact output voltage of 12.5V haven't loaded using any load (just the feedback resistors acts as load). After running for few minutes (like for1minute) the mosfet explodes.
After I checked the components burned and found the current sense resistor, IC uc3845, Gate resistor and mosfet these parts are burned.

Does this is due to high parasitic inductance in bread board so gate voltage is build up high eventually this breaks the mosfet thus it creates short circuit in mosfet ..?

Does laying the exact circuit in perf board may solve problem...?   

[temperance]

Did you try this circuit on a breadboard?

The UC3843 current input is very sensitive and should not be used on a breadboard unless you really know what you're doing.

The circuit requires a star ground layout to work properly.

[TopQuark]

A few things.

1. If this is your first time building a flyback converter, start with a DC to DC design involving 48v or lower, this is how I first got started. This gives you confidence to probe around comfortably without worrying about electrocution and the whole thing violently exploding. I have had mains flyback converters explode in my face before, and I know it can be quite the shocker (emotionally, not physically hopefully).

When you are comfortable with flyback converter design, you can attempt to do mains input again. Though you really should be powering your converter design with an isolation transformer and fuse in series. This will reduce the risk of killing yourself and your test equipment.

2. If your transformer has 130uH of leakage when the primary inductance is 800uH, you are going to have to dissipate around ten watts in your RCD clamp and use a clamp resistor in the 1-10k ohm range. The RCD clamp power dissipation is proportional to the leakage inductance, and so usually designers aim to have leakage inductance be 1-2% of the primary inductance.

The consequence of not dissipating enough power with a low enough RCD clamp resistor is your MOSFET will blow up due Vds over voltage breakdown. You are using a 900V rated MOSFET plus the MOSFET avalanche energy makes it easier to prevent that, but with that much leakage inductance and an inadequate RCD clamp, it is likely not enough to prevent the MOSFET from blowing up I suspect.

3. There are no compensation in the feedback loop at all, not on the primary side, nor the secondary side. I am almost certain the output is oscillating with an average output voltage of 12.5V. Hook an oscilloscope to the output and you will see.

Even without an oscilloscope, just by listening to the sound the flyback converter makes will tell you a lot. Your flyback operates at 65kHz, so in theory you should not hear anything. But if you hear coil whine, periodic noises, or even discontinuous pulsing, you know it is doing what it is not supposed to do, and is likely oscillating.

Compensation components are those little R and C placed near the TL431 on the secondary side, or near the FB pin on the primary side. You may not fully understand why they are there or how to calculate the values, but you absolutely need them, and choosing them is one of the most important stops in designing any power supply with feedback.

Oscillation can cause the duty cycle of your converter to be higher than it should and need to be in some cycles, risking component overheating, over current, and transformer saturation.

Read the excellent presentation by flyback expert Christophe Basso on the subject of flyback compensation. https://www.onsemi.com/pub/Collateral/TND381-D.PDF

4. The UC3843 is a bit of an old generic part, and lacks features like VDD overvoltage protection (Controller won't shut off if you have too many windings on the Aux winding and duty cycle is too high under heavy load, and can literally blow itself up instead of throttling down the output), internal slope compensation, current sense leading edge blanking and internal soft start etc.

You can definitely design a great flyback converter with a UC3843, but it is just more difficult, especially if this is your first flyback design.

TEA18363 is a nice part with all the modern comforts mentioned, makes it harder to fk up your design and everything blowing up. For something cheaper, HG2269 is also a decent part and a bit simpler. Both are available from LCSC.

5. There are a lot of excellent resources online that teaches flyback converter design properly, with more accuracy and details than I can type here.

First of all, page 11-33 of the UC3843 datasheet basically tells you how to design a flyback step by step. Don't attempt to skim through it and ignore the steps mentioned. It is already information dense and every bit of information there is essential.

The video lectures by Prof. Sam Ben Yaakov is an excellent resource for understanding theory of operation.
https://www.youtube.com/@sambenyaakov/search?query=flyback

Videos by long time practicing engineer Robert Bolanos is also great, he goes through the design process step by step in a hands on manner.
https://www.youtube.com/@RobertBolanos/search?query=flyback

[whatty]

Thank you TopQuark for your detailed answer yep this is my first design will study all articles and resources provided and will start design basic dc dc converter at low voltage.
 

[SuzyC]

You should post a pic of your breadboard.

I am quite sure your problem might be due to the inductance of the lead to the RCD clamp to the MOSFET drain to the Vin P/S.  Even a small inductance (just one or more inches of wire) to the clamp from the Drain can create unexpected high voltage spikes on the drain on MOSFET turnoff that are not clamped and will cause the MOSFET to fail.

A properly designed flyback convertor is very much like a high-frequency RF circuit, keep leads to common ground and from the drain  as short as possible!

Also the V++ 250V bridge output cap ground must be as short as possible to a common ground connection with the MOSFET source. It is not uncommon to have a small valued HV capacitor directly connected (as short as posssible) from the drain to source to quench these spikes if needed. A typical value for this cap is usually between 1 to 3nF at 1500Vor so, low ESR and must be rated for repetitive  high surge currents  ( polyprop cap).  Too high a  protective spike cap will cause failure due to the energy stored in this capacitor upon MOSFET turnon.

Breadboarding this circuit at 48V is only going to give you pipe dreams your circuit is working and convince you your design is working really great and reliable when the MOSFET it is not being stressed at the highest voltage it is intended to work on with 230VAC input. You are wasting your time with that approach, unless in the case your breadboarded connections are excessively sloppy in regard to
ground and lead length.   

Hint: you can create a low-inductance common ground and Vin HV with strips of PCB (even on a breadboard). Try cutting two thin strip(1/2-in wide one for V++, one for Gnd) of PCB stock with several flying wires connecting to the ground bus on the BB. The other strip cut is for power with another bypass cap bridging the two PCB strips(~4.7uF/450V). Connect a few nF ceramic cap in parallel with the added small uF bypass on this floating power/Gnd busses.

Both strips can be on one PCB by stripping off a gap between the two buss strips and both can be flying above the breadboard. The ground PCB strip will then be suspended vertically by several grounding wires as short as possible above the breadboard and the Vin HV strip bus just above it. You can then solder the RCD snubber wiring directly to these floating busses.