Flyback converter problems (UC3845 + MOSFET)

[Taswar]

Hi all,
I'm trying to make an offline flyback converter. Since I can't get readily wound transformers, I need to wind my one. I got a few EE25 cores and bobbins and am aiming to make a 14V 2A flyback converter. I've been reading a lot on flyback and I think I got the idea. Last few times I tried (about a few months back), there was output for a few seconds and then BANG!!!

So I've been reading on flyback. I carried out calculations and stuff and arrived at this:
Primary: 147 turns
Secondary: 9 turns
Bias: 10 turns
I used thick paper and glue to get a gap of about 1mm (measured with slide caliper) between all three legs.

Before anything blew up again, I set up a circuit with UC3845 running at around 90kHz at 48% (max) duty cycle and pulsed an IRF3205 off a 12v line. The IRF3205 drove the secondary of the transformer and I set up a diode, cap and a load resistor on the bias just to check the drain waveform. The IRF3205 got EXTREMELY HOT and burnt. So, I placed an IRF840, which also gets EXTREMELY HOT, but didn't burn out.

This is the drain waveform with no snubber:

This is the drain waveform with an RCD snubber (diode - BYV26E, cap - 4.7u 100v, res - 4.7R 2W):

I'm doubting one of two things here:
1) It is due to poor winding or any such thing relevant which caused excessive leakage inductance.
2) My snubber values are completely wrong.

Can someone help identify the problem? Are the transformer calculations and airgap right?
I think if I make my offline PSU, the MOSFET will go kaboom! So, what can I do?

By the way, I didn't measure the secondary and bias windings, but the primary winding (147 turns) was about 1.47mH. Without the air gap, it was about 16mH.

Thanks in advance.

 

[mister_rf]

Read here:
**broken link removed**
**broken link removed**
Electronic Device And Electronic Circuit: Design the MOSFET RCD Snubber Circuit
**broken link removed**

 

[Orson Cart]

It appears your flyback Tx is saturating as you can see from the ON drop of the fet that the current is climbing rapidly during the ON time, the cores may be too small in this instance, also your duty cycle is way too high - you need a way of controlling it from 0% to 30% say for testing! In the finished article you would have a volt feedback loop that would limit the duty cycle to near zero for no load, and increase it as the load increases to maintain ouput regulation, - Regards Orson Cart.

 

[Taswar]

It appears your flyback Tx is saturating as you can see from the ON drop of the fet that the current is climbing rapidly during the ON time, the cores may be too small in this instance, also your duty cycle is way too high - you need a way of controlling it from 0% to 30% say for testing! In the finished article you would have a volt feedback loop that would limit the duty cycle to near zero for no load, and increase it as the load increases to maintain ouput regulation, - Regards Orson Cart.

Okay, I'll try with a feedback loop to control the duty cycle. I doubt the transformer is saturating because the cores are too small, because there's only a 240 ohm resistor on the bias winding, of course, it is rectified as required in flyback converters. Is my line of thinking right? Could there be an air gap problem?

---------- Post added at 08:21 ---------- Previous post was at 07:27 ----------

Is my transformer calculation right?

 

[Orson Cart]

from the scope shot it appears the Fet resistance coupled with the wdg resistance & possibly the power supply too) is the only thing limiting the current (notice how it tapers over in a L/R time constant fashion) - the current should be a dead straight line (V/L=di/dt). The key here is to limit the duty cycle and so produce the desired o/p volts to the load (a big zener might be a good idea). For a fixed duty cycle if the load is removed the o/p volts will just go higher and higher until some other point in the system limits it by breaking down. Regards, Orson Cart.

 

[Taswar]

Hi,
The scope is showing voltage, right? How did you find the current from this? Why does the voltage shoot up so much? I'll control the duty cycle. However, could there be a problem with the transformer winding and calculations. Are the calculations ok?

 

[Orson Cart]

current x resistance = voltage, as seen on the scope, once you get your duty cycle under control - you will get much better results from your ckt and will then be able to see for your self if the Tx core or wdgs get too hot and need re-design, use the minimum RCD catch snubber, power integrations website shows classic snubbing for flybacks, what is the peak power you are after on the o/p? V/I? Orson

 

[Taswar]

This is supposed to be an offline project, but I just wanted to test everything at 12v first. The load for test could be 12v 0.5 to 1 amp on the output. But for the offline version, it's going to be 14v 2a output.

 

[Orson Cart]

OK, so only 28watts, what freq? I suggest 2 x BYV26C in series, 10nF 250v cap and 100k 1W for the turn off snubber, your flyback voltage is 228 volts on the primary - so the fet has to withstand peak mains, 375V say + 228volt flyback + a bit of overshoot = an 700/800v fet in practice

with 1.47mH for full power (assuming 70kHz) the peak pri current each cycle needs to be 0.824 amp (assuming 80% efficiency overall)

to get to 0.824 amp in 1/3 of the 14us avail, you need (V/L=di/dt) 260VDC min on the HVDC (should be easy with rectified 230Vac)

If you have 1.47mH of primary inductance using EE25 cores (6.35x6.35mm centre gap) and 147 turns then the gap (total) must be closer to 0.775mm, i.e. 0.388 mm between each leg, this gives a Bmax of 0.19 Tesla for 0.8 amps peak in the pri winding - which is OK for freq's up to 100kHz - so your transformer design is pretty good.
The key to a good flyback is to keep the Tx leakage down (close coupling) and tight tracking of the pri switching and the o/p rectifying loop - Regards, Orson Cart.