Brown Out Detector Bug/issue

[urbanze]

Since yesterday the BOD of my esp32 is crazy, restarting randomly (apparently) when starting the WiFi.

Sometimes when I set the BOD to 2.6V, the ESP32 keeps restarting endlessly (but sometimes it restarts 5-10x and stops), however, when I change to 2.1V and return to 2.6V, it stops !!!! (sometimes this does not work).

I already checked the oscilloscope and added 40nF and 470uF capacitors between 3.3V and GND, there is nothing much beyond normal, power supply is good and not droping (3.3v has 3.3v and usb has ~4.6V)

What can it be? I'm using the NodeMCU32 DoIT.

*This only started to happen after the first change of the BOD settings. Before that there were never any problems like that.*

[ESP_Angus]

Hi urbanze,

What version of ESP-IDF are you using? There have been some significant changes to the BOD code and levels lately.

If it's crashing during WiFi calibration, updating to the latest ESP-IDF may help. There was a bug fixed which caused calibration to use more current than it should.

I already checked the oscilloscope

Can you give any more details about what you've checked? I'd suggest probing 3.3V as close to the ESP32 chip as possible, with oscilloscope ground connection also made as close as possible to the ESP32 chip's ground. Set the oscilloscope to trigger on a (say) 2.8V falling edge in Single triggerind mode, and then run the code which triggers BOD.

[urbanze]

Hi Angus,
I update ESP-IDF to new "3.0 pre-release rc1" and got new informations about BOD.

1-) Before the update, when it was with 1mF capacitor it continued to reset, however now with capacitor does not reset.
2-) Without the capacitor, reset.
3-) Note: This did not happen in the past, it only started to happen when I changed the BOD settings for the first time.
4-) This only happens when WiFi starting.
5-) BOD Level is set to 2.43V +- 0.05V (the smallest choice).
5-) I will only post the links to not pollute the post. Some pictures(oscilloscope):

ESP32 Deveploment Board used (DoIT): https://i.imgur.com/9qLuRyx.jpg
ESP32 Informations (dual core, revision, etc): https://i.imgur.com/QUyYnBt.png

Note1: Oscilloscope probe in 3.3V pin
Note2: 1V/Division
ESP32 Oscilloscope BOD RESETING without capacitor (min voltage is about ~2.6V): https://i.imgur.com/BFuqQAx.jpg
ESP32 Oscilloscope BOD NOT RESETING with 1mF capacitor on 3.3V/GND: https://i.imgur.com/6AI2nC6.jpg

[ESP_Angus]

Thanks for the update and the detailed scope shots, etc.

I think this is probably a hardware fault with the DOIT board not integrating enough capacitance on the board.

The fact you can see a pulse down to 2.6V on what looks like a fairly slow scope trace makes me think it probably is dipping lower at the ESP32 end for a short period of time, enough to trip at 2.43V (remember the scope probe and connections introduce some impedance, and the scope's bandwidth will filter the pulse as well. The actual time spent below 2.4V may only be a small fraction of this time).

You can also check (or try replacing) the USB cable and anything else in the setup (USB host port?) that may be limiting the current to the board, or introducing impedance.

How is the 1mF capacitor connected to the DOIT board? You may find you can use a smaller capacitor, depending on where it's connected to the board.

I actually have some DOIT boards coming soon so I can test a different reported issue, so I can try this out on those as well.

[urbanze]

Thanks for the update and the detailed scope shots, etc.

I think this is probably a hardware fault with the DOIT board not integrating enough capacitance on the board.

The fact you can see a pulse down to 2.6V on what looks like a fairly slow scope trace makes me think it probably is dipping lower at the ESP32 end for a short period of time, enough to trip at 2.43V (remember the scope probe and connections introduce some impedance, and the scope's bandwidth will filter the pulse as well. The actual time spent below 2.4V may only be a small fraction of this time).

You can also check (or try replacing) the USB cable and anything else in the setup (USB host port?) that may be limiting the current to the board, or introducing impedance.

How is the 1mF capacitor connected to the DOIT board? You may find you can use a smaller capacitor, depending on where it's connected to the board.

I actually have some DOIT boards coming soon so I can test a different reported issue, so I can try this out on those as well.

Hi Angus,

I had an old DoIT card (revision 1) and everything worked perfectly, including the automatic upload I am citing in this post.

https://esp32.com/viewtopic.php?f=2&t=3711

As you can read, adding the capacitor in the EN / GND or using "make erase_flash" allows the automatic upload of the code and this only occurs in this new board that I bought (revision 0).

If you have solutions to the problem of automatic upload, I would appreciate the help.

Another important note, I already tested with 2 cables (they are not the best but it is what I have) and sometimes the BOD does not restart the ESP, giving really impression that is the cable or connector. I'll get some better cable to test and I'll say it here. I will also be waiting for your test with the DoIT!

[FrankenApps]

I have the same DoIT Board and whenever I start WiFi, the brownout detector is triggered.
I have tried connecting a capacitor between 3.3V and GND, but that did not help anything.
It is a rev. 1 btw. and the ESP32 Chip mounted on it, is also rev. 1, I have checked this with that sketch:
https://github.com/SensorsIot/ESP32-Revision-level
Is there no solution to this?
If I can not use WiFi, the board is basically useless for me.

[FrankenApps]

Well there is a somewhat strange fix to this:
Simply connect a 3V3 source to the ESP32 3V pin and the corresponding GND to the ESP32 ground (I use an Arduino or FTDI breakout) while still powering the board via USB, this made it possible for me to use WiFi

[ESP_Sprite]

In that case: have you tried using a different USB cable? (thicker/shorter) Using a thin USB cable (or an unpowered hub, or a meh power source) can lead to symptoms like this.

[ajaybnl]

My esp32 Board was doing Brownouts then i attached a lm317 3.3v power supply. It was now working with wifi without problem (serial was ok).
Regulator was ok but resistance was bad.
Attached a 470Uf Cap directly to esp8266 Reg.

Solution:
Check if the traces going to esp32 vcc gnd pins have resistance or the regulator /cap is bad/ faulty.