I've made a simple AC to DC chopping/switching power supply circuit (I'm not sure if it's called that) as shown below. in the left part of circuit (which is the gating/chopping part) when the rectified input voltage gets higher than 20 volts, the Zener diode conducts, transistor turns on and the MOSFET turns off consequently, therefore the maximum voltage on 100uF C3 cap remains at about 20V. the right side of the circuit is a buck converter based on MC34063 based on its typical application.

the circuit works fine when tested on 25v AC input (with no load). I get 5.2volts output capable of providing up to 300mA; but when I connect it to 220v AC line the 33ohm R4 resistor burns and the PCB track along that resistor snaps off immediately, breaking the circuit. any Idea about that??

and here are some information and test results :

  • the MOSFET Vds-max is 600v, the 100uF C3 Cap is 25v tolerant.
  • R4 resistor is 1W, Zener diode is 20v 0.5W, the other parts are 1206/0805
  • in 25v AC input test, the voltage across the C3 remains at 20.5v which is fine. no part got hot
  • the parts are cheap Chinese ones. but the problem remained through various tests, various PCBs

  • 1
    \$\begingroup\$ Switching scheme is your own development or did you take something as a basis? \$\endgroup\$
    – AltAir
    Mar 25, 2019 at 20:04
  • \$\begingroup\$ this is an exact implementation of a circuit I saw in elector mag a few years ago. just changed the buck circuit and the mosfet. @AltAir \$\endgroup\$ Mar 25, 2019 at 20:21

2 Answers 2


Consider what happens when you increase \$Vin \cdot10\$. The voltage slew rate also increases \$dV/dt=I/C\$ just when the Zener conducts. But increasing current x10 also increases R4 power \$(I^2R)\cdot100\$, so the designer forgot to include this in calculations.

This topology has poor characteristics for line voltage and should be avoided.

Replacing R4 with an ICL can prevent inrush currents but is still very inefficient.

An SCR regulator or better a buck forward or flyback regulator is best.

Even better is ZVS offline regulator with active PFC!

  • \$\begingroup\$ The key is the value and wattage of R4. It should be the flame-proof type. \$\endgroup\$
    – user105652
    Mar 25, 2019 at 19:49
  • \$\begingroup\$ @ Sparky256 Increasing the resistance and power of the resistor will increase the loss of heating. The switching circuit loses its meaning. \$\endgroup\$
    – AltAir
    Mar 25, 2019 at 20:03
  • \$\begingroup\$ @AltAir Back off some. It was just an observation, not my design. Yes it is a poor design in that the wrong PWM controller is being used in terms of the power source. If R4 keeps burning up, maybe the OP will get a clue. \$\endgroup\$
    – user105652
    Mar 25, 2019 at 22:21

You can insert a zener diode between the C and E of the transistor Q2. Another useful thing is to move R3 between the emitter of Q2 and the cathode of Q1. This way you can keep the charging current of C3 limited by the mosfet. Some math is needed to check if the current pulse can supply your power needs and blows nothing.

Something to think seriously before you continue:

You have a power supply with galvanic contact to mains AC. Think the day you or some other connects this to a device which has internal connection between 5VDC and some freely touchable wires or case. You have a substitute for electric chair. If you cannot quarantee the 5VDC stay behind proper mains insulation, you danger lives and the house you live.

ADD: the questioner wrote that isolation is no problem.

So do a simulation where a constant current, say 75mA is discharged from C3. you should get peak powers, peak currents and average dissipations.

  • \$\begingroup\$ the efficiency or isolation is not a concern. but I'm wondering how a Zener in there should be placed alongside other components \$\endgroup\$ Mar 25, 2019 at 20:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.