Very weird blown capacitor on multiple ESP32-PoE-ISO boards

[spatialeffects]

I have this problem on 3 ESP32-PoE-ISO boards:



What could have caused this? I think the boards are revision C or D.

[LubOlimex]

This is not typical. Probably some overvoltage situation occurred, this is a 10V capacitor that is located after the Si3402 chip, so maybe the Si3402 also died?

How do you power the boards? Do you use real PoE or some sort of passive injection?

Also please give us the exact revisions of the boards that got damaged. It is written at the bottom side near the blown capacitor.

[spatialeffects]

The boards still work, and they're powered over PoE.

The power comes from TP-Link T1600G-52PS JetStream Smart PoE Switches. I would say it's about 50 meters of CAT6 from the switch to the ESP32.

All three boards that are affected are revision B

We've found a capacitor casing, the silver tip didn't break away. It looks like it was heatshrunk.

My theory is: the DC-DC converter made so much heat that it shrunk the heatshrink sleeve of the capacitor over time. The boards are in the orientation you see on the photo, with not a whole lot of airflow. So the hot air from the DC-DC would flow right on this capacitor.

[LubOlimex]

Oh well, that is quite interesting. Thanks for the feedback. First time see something like this and we've manufactured quite a few of these boards. These are 4-layer boards heat dissipation should be quite good. A lot of customers used it in harsh conditions and never reported such a problem, it is either they didn't report it or something in your design is harsher. Good news is that revision B is quite old revision. Currently manufactured boards would behave quite differently since that part of the board was redesigned with different DC-DC. I would suggest latest revision of the board would be better for heat dissipation.

[slacks33]

Oh well, that is quite interesting. Thanks for the feedback. First time see something like this and we've manufactured quite a few of these boards. These are 4-layer boards heat dissipation should be quite good. A lot of customers used it in harsh conditions and never reported such a problem, it is either they didn't report it or something in your design is harsher. Good news is that revision B is quite old revision. Currently manufactured boards would behave quite differently since that part of the board was redesigned with different DC-DC. I would suggest latest revision of the board would be better for heat dissipation.

Hey there, don't mean to necropost, but I ran across this thread having the exact same issue on a ESP32-POE-ISO-EA-IND.

Bought this board in January of 2024, it has been running in a climate controlled room since shortly after that date. It has been largely untouched since then, I pulled it out to put it in a new 3D Printed case, just to find this:


This is Revision L, so I'm surprised that this is still happening. Reached out to support a few minutes ago, just wanted to also add to my findings here. Hoping this can get resolved, spent the extra money on the IND version to ensure I would have great longevity with this product.

[LubOlimex]

Thanks for the feedback. We always improve on our designs, they are not set in stone, sadly we can't emulate scenarios and all problems here - sometimes we need customer feedback before we can spot a potential issue. We will bring some changes to the hardware design that should further reduce the chances of such a capacitor pop in the next hardware revision of ESP32-POE-ISO, e.g. hardware revision N.