Problem with Single Transistor Forward Converter

[galib.tan]

Hi everybody

Hope you all are fine.
I'm working on a forward converter which is giving me a lot of hassle. The required output voltage and current ratings are 16V and 2.5A. The switching frequency is 100 kHz with a duty cycle of 35%. This PWM is generated by a microcontroller and fed to the gate via an IR2110 and a gate drive transformer.


At first I used an 8 ohm NTC and 1A fuse and it worked though the voltage was greater than expected because I mistakenly added more turns on the secondary. After rewinding the circuit worked for several minutes then suddenly the fuse blew off. A number of fuses blew. The problem had gone using a 20 ohm NTC. The fuse was also changed to 5A. Then the mosfet got damaged (not burnt but G, D and S all short), the PWM was turned off in that case.

Then the mosfet was replaced and powered up again with PWM on, this time the NTC exploded and the reset winding diode got short. This diode got short several times before.

I'm new to power electronics. I don't understand if I'm doing anything wrong or missing something.
Please do help. Thanks in advance.

 

[BradtheRad]

1.

L1 is in a position to generate high voltage spikes. I believe you need to put some kind of snubbing network across it.

2.

While you are in the development stage, I suggest you put a full diode bridge where you have D6 and D7. This will permit current flow in both directions.

Later, after you determine which direction provides the greatest current flow, you can reduce it to a single diode.

 

[Orson Cart]

OK, do not put a full bridge to replace D6,7. You do need a snubber across the mosfet say 150pF 1kV, and 47 ohm for starters (2W), also a string of zeners across the mosfet, 750 volts worth, say 5 x 150V 1.3W in series, and snubbers across D6,7 say, 1nF 500V, and 10 ohm for starters (2W).
The 1N5399 is too slow, you need 2 x 600V U-fast in series (<35nS Trr).
This should get you going for starters, the two pri side windings on the Tx should be bi-filar for good coupling. You should repeat them on each side of the sec wdgs to get the leakage down to reduce the snubber requirements.

 

[BradtheRad]

OK, do not put a full bridge to replace D6,7.

Hi pal - Because output will be weak if the secondary winding is not oriented correctly, I thought a full diode bridge is a good idea, while in the development stage.

 

[Orson Cart]

Unfortunately a full diode bridge will tend to short out the sec wdg when the primary is trying to reset, with high volts, stopping this reset action will lead to saturation of the Tx after not many switching cycles, then big currents in the mosfet, then bang...

checking the phasing of the Tx wdgs - good idea...!

 

[galib.tan]

Hi BradtheRad and Orson Cart, thank you very much for your thoughtful replies.

I checked the phases of the transformer windings and they were ok. At first, I wound the primary and the reset windings as bifilar then I separated them to check if it works. :-D
Anyway, I'll definitely follow your suggestions. I've never designed a snubber circuit before. I found some app notes on snubbers and I'm gonna have a look at them. If you have good resources please share.

I destroyed the last mosfet I had unfortunately. The Vdss rating of the BUZ80 is 800V. Such high voltage mosfet is not so available in my locality. Do you think an IGBT would be a better choice?

 

[BradtheRad]

Unfortunately a full diode bridge will tend to short out the sec wdg when the primary is trying to reset, with high volts, stopping this reset action will lead to saturation of the Tx after not many switching cycles, then big currents in the mosfet, then bang...

checking the phasing of the Tx wdgs - good idea...!

I think your knowledge is greater than mine. The schematic of the OP is similar to a typical forward converter. The dots on the primary are at the same end as the secondary. This implies it is important they be in a particular phase.

It is the opposite with a flyback. The dots are at opposite ends. If I did not look carefully I would miss it. It is instructive to place a full diode bridge across the secondary, to see how current flow is distributed between pairs of diodes (in simulation).

I don't know that this is necessarily creating the problem presented by the OP.

 

[Anna Conda]

A fast (ultra fast) igbt will work but will need heatsinking (as will the mosfet) as there are much greater turn off losses in even an ultra fast igbt.

 

[Orson Cart]

Dear galib.tan, I assume you did have a heatsink on the original mosfet?

 

[mrinalmani]

I think the diode D5 (1N5399) is causing problems.
The recovery time of the diode is 2us typical. At 100kHz, with a duty cycle of 40%, the switch on time of the transformer is only 2us. This diode seems to be forward biased during a large portion or perhaps even the entire ON period. The bifilar diode should conduct only during reset period and not during the 'set' period. But due to the large recovery time of D5, it conducts for a long time even in the set period, thereby effectively shorting the primary. Fuse blow-off is expected!

Also the MOSFET you are using has an internal resistance of 3ohm. At full load of 2.5A, this will dissipate around 20W. This will require an excessively large heat-sink or a fan based cooling.
The current rating of the MOSFET is 3.6A. But this is the silicon limited rating at 25 degree celcius case temperature.
The junction-ambient thermal resistance of the MOSFET is 75K/W. Which means every 1W of heat dissipated wiil rise the temperature by 75 degree celcius. So without a heatsink this MOSFET will not handle say more than 700mA.
Use a MOSFET with Rds(on) in the range of 0.5 ohm or so.

 

[galib.tan]

Hi Orson Cart, yeah, I used a reasonable heatsink with the mosfet. The mosfet wasn't destroyed due to excessive heat, I think. It was fairly cool. All the three pins (G, D and S) became short. Excessive current would raise the temperature and the mosfet would burn. That didn't happen.


Hello mrinalmani, thanks for your nice explanation.
I agree with you about the 1N5399. It's too slow. Orson Cart also pointed that out. I'll replace it with an ultra fast one soon.
As for the second part of your reply, 2.5A is the secondary current. The mosfet will carry about half an amp total. So, the temperature will not be unmanageable, I think.