Failure in buck switching voltage regulator (LM2574)

[alek]

Hey guys and gals,

I was prototyping a buck regulator circuit (per reference circuit shown in a datasheet for an LM2574-3.3 http://www.ti.com/lit/ds/symlink/lm2574.pdf) and had it fail after several turn on cycles. I did diverge from the ref. schematic by using a larger output cap (1000u instead of 220u - had them lying around) and by using an 1N4148 instead of the suggested 11DQ06 (once again, what I had lying around and I am only drawing 100mA of current).

While I was playing around looking at its turn on transients, the LM2574 failed - so now I'm trying to figure out what was the culprit. For your reference the input is unregulated 35VDC (rectified and filtered 24VAC) and the circuit was being tested unloaded. I think I actually captured the turn on cycle at which it failed. I had the scope hooked up to LM2574 output pin on single shot. That time instead of many pulses, I just had it saturate to somewhere around 20-30 V over several milliseconds. Cycled the power again and then there was no pulses coming out of the LM2574. Output voltage was then 0.5 V. I checked the diode and it's still functional. Checked the output cap and it's still good. I even hooked up a 3.3V supply to the output to see if anything would fail catastrophically - nothing, PS was outputting 3.3V at 1mA.

Any ideas? Any help would be greatly appreciated - I'm somewhat lost here.

[Niklas]

It could be due to too low switching speed of the diode. It is fast for low currents, but slower with higher current. When the regulator turns off the switch transistor, and the diode is not fast enough to turn on, then the inductor can   create a voltage transient high enough to breakdown the switch transistor. The input voltage is already close to max, so your safety margin is low.
Google "diode induced failure linear". Linear has both an appnote and a video clip illustrating this issue.

[alek]

After replacing the diode with an appropriate one and blowing yet another regulator I found my rookie mistake of the week: over-voltage due to inrush current in a hefty 24V transformer (attached).



Looks like I need a bigger cap + TVS on input.