I made a PCB for an ESP32-based solenoid controller that is powered by a 24V DC Jack. I'm using an AP63205 Buck Converter to step that input down to 5V DC, and then a linear 3V3 LDO to power the ESP32. I have assembled multiple such identical PCBs, and they all need to be powered by 24V DC.

The voltage rating of the Buck Converter is up to 32 VDC, and all components around the converter are rated for at least 32V. When I plug 12V into the DC Jack, everything works and the power LED I have on board lights up.

However when I plug 24V, sometimes the boards works normally, but sometimes the buck converter IC will fry itself (only upon this initial plug-in of of the DC jack), shorting 5V to GND (verified that IC is damaged by removing it, and short from 5V to GND on PCB disappears). When this happens, the power LED turns off.

If I power the PCB with a DC Power supply starting at 12V, and slowly cranking up the voltage to 24V, the board always works as expected without frying. I suspect the issue to be related to plugging in the 24V DC Jack, but I'm not certain what could be causing the inconsistent deaths.

Even when the Buck IC dies, the 3V3 LDO still functions and USB powers the board properly.

My buck converter schematic is below:

Buck Converter Schematic

EDIT: The components I used from DigiKey:

The 24V DC Power Supply I'm using is the following: 24VDC Power Supply

Is it possible that the inductor is not being powered properly upon the initial plug-in of power, causing a large back EMF that fries the IC? Is there any protection against this?

I noticed upon plugging the 24V DC Jack into other devices that sometimes there is a small spark. I'm not sure if that's an issue with the 24V Jack or if that's common, but that's very probably the culprit. I would still like to protect against this.

  • 4
    \$\begingroup\$ Use a TVS diode on the input that can stand off 24V but clamps before the 32V max input voltage. Use a fuse before the TVS so that in case of fault the fuse will blow before the TVS. \$\endgroup\$
    – John D
    Aug 20, 2019 at 1:32
  • 2
    \$\begingroup\$ Can you please edit to include the actual part numbers of: the Input Capacitor (C7), Output Capacitors (C8,C9), and inductor (L1). See AP36205 datasheet page 13. Input capacitor must meet inrush current rating (RMS current); Output capacitor must be low ESR; inductor must have sufficient LI^2 energy storage (saturation current). Each of these capacitors/inductors will have its own datasheet. I assume you also followed the suggested PCB layout on page 15? \$\endgroup\$
    – MarkU
    Aug 20, 2019 at 2:20
  • \$\begingroup\$ ti.com/lit/an/slva670a/slva670a.pdf inrush current limiting ideas via a load switch or a discrete solution \$\endgroup\$
    – User323693
    Aug 20, 2019 at 8:00
  • \$\begingroup\$ MarkU I have added the component links. I used only components I had laying around from an old BOM, so I replaced the input capacitor with a 4.7uF. I didn't think that would make a difference, but perhaps using the actual rated 10uF would reduce this inrush current? \$\endgroup\$ Aug 20, 2019 at 20:21
  • \$\begingroup\$ I had the exact same thing happen to my project! I made an ESP-32 based 24V RGB LED controller. The power supply design I used to power the ESP32 was basically the same as your design. Only difference was the type of converter IC used (EUP3458, micro-bridge.com/data/Eutech/EUP3458.pdf) which is rated to 30V, and the fact that literally EVERYTHING was fried except for the MOSFETs (to drive the LEDs). I am currently redesigning the board to use a higher voltage rated«buck ic (LM22675MRX-5.0) IC, together with a polyfuse and a TVS diode. I hope that will resolve the problems... \$\endgroup\$ Nov 15, 2020 at 12:27

3 Answers 3


This is due to inrush current. To get a good feel for what you are actually dealing with a simulation is always a good help! (LTspice is worth looking into)

enter image description here

As you can in the simulation the capacitor takes in (approximately...) 20 A peak at 24 V. To limit a high rate of change in current, an inductor can be used. But this will induce voltage oscillations at the input. This is not completely avoidable but you should try and dampen them as much as possible.

The simulation I did is NOT a complete solution. You should play with it to see what works. A TVS diode might help you out, but a proper input filter should be able to deal with the problem.

(also on page 10 of the datasheet, there is mention of some soft-start circuit with the EN(able)-pin which will alleviate things more.)

  • \$\begingroup\$ (I'm not super familiar with inrush current, I'm reading up on it) wouldn't inrush current damage the input capacitor and not the buck IC? I tested the input capacitors and they're fully functional, only the IC is fried. \$\endgroup\$ Aug 21, 2019 at 22:28
  • \$\begingroup\$ Yes I agree. But regulator IC's are complex things and it might be due to switch-mode control oscillations. Your input supply is limited at ~1.5A (probably higher in peak) but this willl make the supply cut-off also. 2 control algorithms fighting each other is something to avoid. Dampen the current flanks (di/dt) and everything might "fix itself". \$\endgroup\$ Aug 22, 2019 at 5:52

I'd guess a voltage surge at turnon the most likely but not certain cause.
A small series input resistor followed by a say 27V zener would clip startup spikes. Adding a capacitor across the zener would reduce zener dissipation.
Size series resistor to drop as little as possible and as much as necessary under maximum load current, Maybe a volt or 2 drop.


What if your 24VDC power wires had significant inductance causing a voltage spike of almost 48V at turn on when C7 has zero initial volts. This is why it was accepted old school practice to use 60 volt parts for good reliability. Now that you have committed to the 32 volt chip you must adress the peak volts across C7 by whatever means.


Your Answer

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

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