I'm using LM3478 converter to convert 12V to 30V in standard circuit for 456kHz. Stepup circuit with LM3478

Circuit was working perfectly but when I was measuring something I demaged probably LM3478. I replaced it with new IC but it did not help. I can see after power cycle full current to R4 sensing resitor because Q1 transistor is opened via DR pin. But IC keeps it opened forever. Seems as oscillator does not work or so. I can see 2V at COMP pin (which is output of err comparator but thats all). Q1 seems correct, when I force 0V at gate it closes transistor. D1 seems correct too. FB voltage is supposed too. L1 is 6A inductor, it's heating but I don't think it's broken.

Rarely I can see that oscillator starts (I cannot repeat it) but any load causes it falls back to error state above.

Does anybody have an idea what might be wrong or what focus on? Replacing blindly IC over and over trail&error is rather strange approach.

  • \$\begingroup\$ How long in mm is the output current loop ? Q1,D1,C4,C10 and how much does the feedback loop differ? what is Vfb value? \$\endgroup\$ Jun 6, 2021 at 18:12
  • \$\begingroup\$ Make sure FA/SD is <1.3V to operate clock, BTW 0V on gate Opens it not closes \$\endgroup\$ Jun 6, 2021 at 18:20

2 Answers 2


In order of vulnerability: -

  • The chip
  • The MOSFET
  • The diode
  • Electrolytics
  • The current limiting resistor
  • The inductor
  • The pot

You could argue the diode is the most vulnerable and of course if the design originated from a flaky area then question the power ratings of some of the other resistors. The MOSFET could be damaged and exhibit "nearly proper" working characteristics.

Try measuring, with a meter the gate-source resistance - it should appear to be infinite on the meter - anything less and I'd be suspicious of the MOSFET.


Seems the first IC was good. Problem was probably too tight current limit from external power supply. The circuit could not reach steady condition after power cycle.

The second IC was wrong. There was broken FA/SD input (short circuit to AGND). The IC was not able to start oscillations. The DR input was forced to 5V, i.e. full current forever L1, Q1, R4.


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.