Please help for this SMPS circuit

[xyz9915]

Hi! The attached file is a schematic of 50KHz SMPS power supply which uses 325 VDC (after rectification of Mains 240 V). The required current is approx 15 Amperes. Please give comment if any modification in the circuit is required.

[areza]

half bridge, not push-pull, push pull is good for low voltage, irf840 will see 325 x 2 = 650v and will explode sg3525 will follow irf840, you should use atx/at psu they are good for 12v 15a, don’t bother built a power supply when they r widely available and cheap, if u really want to built one then again follow at/atx psu way, investigate on them

[switchmodepower]

Push pull is not the correct topology. Half bridge or full bridge would be better.

[Bootstrapper]

I wouldn't personally use three 5 A diodes in parallel, when the total current is 15 A. Since these diodes are individual, the current will not distribute evenly between the diodes and that may lead to failure. There are suitable diodes rated >15 A that meet your needs. You may also want to use a diode rated ~20-30 A to get some safety margin.

[switchmodepower]

Also, No output cap or choke?

[xyz9915]

I know that push pull topology is not a good idea for converting mains voltage to 12v (etc), but i want to use this method due to its simplicity and low components requirement. So is it feasible if I use 2KS792 power MOSFET (900V, 3 Amperes) instead of IRF840 (500V, 8.5 Amperes) for DC to DC conversion. Please reply me ASAP. Output voltage regulation is not essential.

Thanks

[Bootstrapper]

Note the on-state Rds when choosing MOSFET-switch, high resistance leads to higher losses. The maximum voltage stress should be noted as well, there is quite a huge difference at maximum voltages between these two switches mentioned.

Do not make hasty decisions

[golam]

For xyz9915,
Your configuration of push pull is completely alright. What is your output power requirement? 12v and 15A, that means less than 200 watt. Push-pull is more than alright. With so less power requirement, you could go for fly back also. for so small requirement, no need for half bridge or full bridge configuration. I have studied your diagram and find some points to be improved. You have grounded pin no.1, instead it should be connected to your output for feedback to keep your output stable and regulated. From your output, use highpower resistor for reducing current and then use a variable resistor to control the reference 5 volt(converting output12v to 5v), remove the 47k resistor from the supply of internal 5v from pin 16 to pin 2. connect pin 16 direct to pin 2. Now internal ref 5v from pin 16 will be compared with output 5v and pwm signal will be modified internally to stabilize your 12v output.You should add some mechanism for over current protection and should add 4x18v zener diodes at the gates of 4 mosfets to connect gate to ground for protection. Snubber circuits also to be included.
I will check your diagram in detail tomorrow and will try to provide you a detail one within short time.
Instead of irf 840, use such mosfets which is having voltage rating more than 800 v as push pull configuration requires 2xdc voltage rail +10-20% margin.You can use low wattage mosfets also as high frequency use is 3 to 5 times greater than the normal wattage capacity of the fets. If the mosfets are of 200 watt, in high frequency you can use that 600 wattage easily.
ETD 39 would have been adequate for you as with ETD 44, you can get almost 800 watt with 50 kHz frequency.

[Alexsch]

1. Choice of the output stage. Using diodes in parallel does not insure they will divide the current between themselves. The opposite may occur. The better of those diodes will conduct more current, heat up, and then, conduct even more current and heat up some more, until failure. This is called "thermal runavay".

2. ETD44 has Ae=1.72cm. Supposing there is an appreciable inductance following the diodes on the output, we see, that the following applies:
A. The operating DC of the power supply is 34.5%. The peak reverse voltage on the rectifiers will be 73V, which is too high for 12Vdc output at a narrow input range (you specified 325V+/-0V). A better ratio is, perhaps 89:4 which gives us a favorable DC of 86%. But the L-leakage will kill you anyway(more on that later).

B. The Bpeak of the core is 1465Gauss. So the core is actually some 120-80% underutilized at 50KHz.

3. The control circuit is Voltage-Mode, and, therefore does not support automatic flux balance on the core. You should use Current-mode for the control instead.

4. The transistors MUST fail as already been mentioned as they will see 2X325V. Using 900V transistors may help, theoretically but i do not belive that you can wind the transformer with such perfection so that leakage inductances will be negligible. The turn-off spike may easily go into 1-1.5Kv range killing the FETs immidiately, unless snubbers are used.
Also by using 8.9:1 turns ratio, the FETs dissipate ~4.4W each just on conduction, assuming 4.5ohm 900V devices. Somebody here already mentioned those points, i merely, adding some realistic numbers to the technical troubles you may expect.

5. One is not to use 350Vdc capacitors on 325V rails. A 450V caps are a much more safer choice.


Considering all, i would suggest bying a power supply, or investigating the entire concept of SMPS design + Magnetics to a greater depth before making an attempt on high-voltage designs. The results from design mistakes are always catastrophic, and you just left with a blakened pcb and a head to scratch just 0.1 seconds after plugging it in.....

[luka]

Hi

I have german paper of offline push pull, but it would be crazy for you to even try. You said you would do it do to simple design and cost. Where did you get that? are you kidding me?? simple?? low cost?? if you think that you are dreaming, coz everything will be more expensive just because of push pull + when you do and you will burn 10 sets of fets and other components you will go for half- or full- bridge, I just give you warning!

I can send you that paper if you still want to make it, it can e done, but it is pointless, like having 1000HP and front well drive in car, pointless

Best for you would be to get normal PC supply that gives 15A at 12v, or mod one to do that, trust me, you can't just put few components together and it will work, not if you don't have experience with this