0
\$\begingroup\$

I have been working on building a DC/DC converter for an ESP32 using an LM2596.

With a few prototypes out in the wild I'm seeing them reboot randomly.

Having the circuit back on my desk with a scope I'm seeing the output of the LM2596 spike more then I would expect. I have designed the circuit within specifications for the LM2596 with an assumed current draw of between 500mA to 1A.The scope readings were done without the ESP32 connected.

LM2596 scope view

As you can see, with a desktop power supply I'm seeing spikes of up to 3.74V and as low as 3.0V. It seems like this could be causing the issue with the ESP32 as I'm aware that they can be sensitive to input voltage.

I assume my design is causing the issue as when I use an off the shelf LM2596 power board like this one, I'm seeing a much smoother output

enter image description here

I'm considering adding an LC filter to the output of the LM2596 but I'm trying to understand why my design would be causing this much noise in the output from the regulator.

  • What could I change to address this issue and where is my understanding gone wrong?
  • Would an LC filter fix the problem or is there a design flaw with the regulator circuit as is?

enter image description here

The board layout looks like this, I omitted the other components in the schematic as I didn't feel it was relevant.

enter image description here

\$\endgroup\$
5
  • 2
    \$\begingroup\$ Try to test it with some load. \$\endgroup\$
    – user263983
    Commented May 12, 2021 at 12:38
  • 1
    \$\begingroup\$ Maybe your layout (or your measurement technique) is bad. \$\endgroup\$
    – Spehro 'speff' Pefhany
    Commented May 12, 2021 at 12:43
  • 1
    \$\begingroup\$ Show us: your initial schematic and pcb layout of the LM2596, the way you are measuring (how you place your probe when measuring). As @SpehroPefhany said, your meas. technique might be bad \$\endgroup\$
    – Christianidis Vasilis
    Commented May 12, 2021 at 12:45
  • 1
    \$\begingroup\$ Zoom in on the noise, is it 150kHz or something else? Doing PCB layout for these switching regulators isn't trivial. Very easy to get lots of noise if you have a bad layout. \$\endgroup\$
    – Lundin
    Commented May 12, 2021 at 13:23
  • \$\begingroup\$ I have added the PCB design, it was designed with not much consideration other then the inductor not overlapping the feedback loop \$\endgroup\$
    – Ezzy C
    Commented May 12, 2021 at 13:53

3 Answers 3

Reset to default
2
\$\begingroup\$

I advise you to reconsider the design and consider the following: Etch away as little copper as possible. Reduce the impedance of the lines on the circuit board. Define a single ground point.

Use the two layers properly. Do not let the inputs, especially on logic components, float.

When it comes to switching power supplies, the quality of the capacitors used is very important. POSSIBLY. Ceramic capacitors can be placed very close to the source of interference.

\$\endgroup\$
1
\$\begingroup\$

enter image description here

Page 21 of the datasheet. Please make sure you use all the recommended values and components. Also from my experience with ESP32 it also required a 10uF electrolytic capacitor very close to its power pin.

\$\endgroup\$
4
  • \$\begingroup\$ I have based the design on the datasheet recommendations for the other components bit must have missed this about the ripple. The ESP32 currently is just the dev board, which includes the 10uf cap on the board itself. The variants in the voltage is about 25% which seams well out of specification. Would an ESR of 50mOhms be considered too low for the output capacitor \$\endgroup\$
    – Ezzy C
    Commented May 12, 2021 at 13:58
  • \$\begingroup\$ Considering the graph (Figure 9-2) Capacitor ESR versus Capacitor Voltage Rating. I believe the ESR is will below the recommended specification. It looks to call for something in the realm of 150mOhms \$\endgroup\$
    – Ezzy C
    Commented May 12, 2021 at 14:02
  • \$\begingroup\$ Yeah, you can try to switch the capacitor. I would also try adding a resistance to act as a continuous load as someone else suggested in a previous comment. \$\endgroup\$
    – andrew
    Commented May 12, 2021 at 14:07
  • \$\begingroup\$ I will definitely give that a go once I get back to the bench, I might add the ESP back to the board and test with the real world load on it and see if that will provide any insight. \$\endgroup\$
    – Ezzy C
    Commented May 12, 2021 at 14:12
1
\$\begingroup\$

Please don’t tell me you feed mains voltage into this board. It looks like a capacitive dropper psu with no protection. This is a serious safety issue.

Your pcb layout is poor as the gnd track meanders and forms loops. The regulator will not like this and is most likely unstable. The pcb layout for buck converters is critical.

I’d suggest you use a power supply module for safety and learn more about pcb layout.

\$\endgroup\$
3
  • \$\begingroup\$ It appears to be. Complete with fuse. \$\endgroup\$
    – StainlessSteelRat
    Commented May 13, 2021 at 1:03
  • \$\begingroup\$ It is not powered off mains but a 24VAC transformer outside my control. There is a fuse on the input of the board as well as a MOV to ensure most of the transients is dealt with before the bridge rectifier and regulator. Good point regarding the ground meandering, I just connected it to fit without much consideration about PCB layout. \$\endgroup\$
    – Ezzy C
    Commented May 13, 2021 at 1:21
  • \$\begingroup\$ However, during the testing on the scope, I'm powering it directly for a DC 24V power supply without the bridge rectifier or anything before the first Cap to isolate the regulator circuit first \$\endgroup\$
    – Ezzy C
    Commented May 13, 2021 at 1:22

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.