5
\$\begingroup\$

I've posted here before (this) and this could be considered a followup post.

I've taken all the feedback into consideration which would make me belief to have a working version.

However, after about 10-20 minutes of use, the thing just dies. It straight up dies. No output anymore over UART, it won't even boot. There is still ~3.3v on the 3v3 and chip_enable pins, as well as on GPIO0 (boot mode) which has an internal pullup.

Here is a screenshot of the schematic, and here is a link to the pdf, where everything connected to J1 was left unpopulated. None of those transistors were populated. All the unused pins on the ESP32 were left unconnected/floating.

Here is also a link to the pcb layout.

Here is the datasheet of the LEDs that were connected, I bought them as a generic WS1815 ledstrip, I cannot find the manufacturer of them so these might not be the exact LED.

What could be the issue?

The only thing I could think of is that, somehow, 5V shoots back into the ESP32, from the LED strip (the WS2815 have 5V logic level.)

In the two boards that I have tested, I tested a ledstrip connected to J3.

I'm at a loss, any help is appreciated.

To the remarks about ESD, it's not that I'm not at all concerned about ESD, it's more that over the past years I've worked with many custom ESP32 board that did not have any kind of ESD protection. I have never had any issues with those boards.

To the remark about the power supply, I have tried connecting the power supply to another board with the same power cirtcuit, which worked like a charm, the baord stayed alive for the couple hours that I left it on.

\$\endgroup\$
13
  • 1
    \$\begingroup\$ Where do you have the LED strips connected? \$\endgroup\$
    – JRE
    Commented Jan 12, 2021 at 21:37
  • 1
    \$\begingroup\$ Permanently destroyed after a certain time? Likely overheating. \$\endgroup\$ Commented Jan 12, 2021 at 21:38
  • \$\begingroup\$ @JRE I had them connected to J5. \$\endgroup\$
    – Cyborgium
    Commented Jan 12, 2021 at 22:01
  • 1
    \$\begingroup\$ I'd be looking at a power problem or out-of-limits voltages being applied to I/O pins. I've never seen an ESP32 even get warm! \$\endgroup\$
    – jwh20
    Commented Jan 12, 2021 at 22:08
  • 1
    \$\begingroup\$ Its not a wise idea to have unprotected port pins go out into the real world. Even for onboard ws2812 i use a level translator. Is the voltage regulator shutting down? Is it oscillating? Pcb layout is crucial here. \$\endgroup\$
    – Kartman
    Commented Jan 13, 2021 at 0:40

2 Answers 2

6
\$\begingroup\$

I had the same issue, with the SMD version of the ESP, during prototyping. Since I started using ESD mats, and ESD gloves, I dont see my ESP's dying :-)

What I observed as Stage 1 was random serial characters, and then within few hours, esp would be dead. No Flashing either! Obviously I was touching them with my bare hands, on a carpeted floor. I took 2-3 days to research and learn ESD protection circuits.

Additionally, the schematics you posted are very blurry, but since it works and dies off, I would presume, a power failure/transient coming from ESD most probably. I didnt see any protection on Tx and Rx line of your esp32.

Esp32, I believe is only rated for 500V, and human body can at a minimum, even to feel a ESD discharge requires 3kV, which makes ESD protection more or less mandatory on esp32. Look at the dev board schematic and they have it on there, on power, D+ and D- lines.

enter image description here enter image description here

\$\endgroup\$
5
  • \$\begingroup\$ Thank you for pointing out the CM1293A-04SO, that seems easy enough to implement in my design! \$\endgroup\$
    – Cyborgium
    Commented Jan 13, 2021 at 12:56
  • \$\begingroup\$ Would you have a link to that schematic? \$\endgroup\$
    – Cyborgium
    Commented Jan 13, 2021 at 13:05
  • \$\begingroup\$ here is the link for ESP32 devkit on the web . On the "micro usb and Uart " section, the protection is colored Blue(dark maybe) and naming is "ESDxxx" and for power they have a "BATxxx" . Diodes D3, D4, D5 and D6 @ dl.espressif.com/dl/schematics/esp32_devkitc_v4-sch.pdf \$\endgroup\$
    – Amogh Jain
    Commented Jan 13, 2021 at 19:47
  • \$\begingroup\$ Instead of the array I mentioned earlier, you can also use,dedicated/standalone TVS's that do not require any power source to operate 1. AU3321D5: for Digital IO Pins (Tinyyy SMD Package, ~$0.03 @ LCSC part # C393856) @ 80 Watts Power Dissipation 2. MMBZ5V6ALT1 : for Power and Ground Pins (SOT package ~ $0.03 @ LCSC, Part # C123251 and maybe x3 at Digikey) @ 24watts Power Dissipation. \$\endgroup\$
    – Amogh Jain
    Commented Jan 13, 2021 at 19:47
  • 1
    \$\begingroup\$ Thanks for the detailed response! \$\endgroup\$
    – Cyborgium
    Commented Jan 14, 2021 at 10:13
-1
\$\begingroup\$

Have you tried a more consistent voltage regulation like a LM2576 module instead of linearly regulated power source?

\$\endgroup\$
3
  • 3
    \$\begingroup\$ Linear regulators as a rule have a cleaner and more stable output than switching regulators like the LM2576. \$\endgroup\$
    – JRE
    Commented Jan 13, 2021 at 16:19
  • \$\begingroup\$ When used in audio applications like ESP32 wifi, linear regulation can lead to extra capacitor noise over time undesired for the cpu itself. Instead switched power sources combat this rapid load variations by autonomously regulating the same exact output voltage under a variety of load conditions. \$\endgroup\$
    – pateitos
    Commented Jan 13, 2021 at 16:27
  • 2
    \$\begingroup\$ Nope. A linear regulator with appropriate capacitors will be cleaner than a switching power supply. A linear regulator reacts just as well as (if not better than) a switching regulator to sudden load changes. Additionally, an ESP32 and WiFi can't by any reasonable means be classified as "audio." \$\endgroup\$
    – JRE
    Commented Jan 13, 2021 at 16:45

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.