Magic Smoke Seen on Some ESP8266 Modules

While playing with my ESPToy today, I discovered a rather bizarre issue. The symptom is that some ESP8266 (the ESP-01 version) modules became very hot to touch. In fact, a couple of them started smoking, and the red indicator LED started burning away! Shocking. What’s even more shocking is that these modules continue to function, as if the chip isn’t bothered by the magic smoke at all…

Not all of them have this mysterious problem. So what the heck is going on here? After some digging, I found that on the problematic ones, the tiny resistor next to the red power LED measures only 4.7 ohm?! This can’t be right. This resistor is a clearly a current limiting resistor for the LED — the LED and the resistor are connected in series between VCC and GND. A 4.7 ohm resistor means the current flowing through the LED would be somewhere around (3.3V – 2.1V) / 4.7 ohm = 255 mA! (2.1V is the typical forward drop voltage of red LED). No wonder why the LED (or resistor, or both) is smoking! Phew, that really scared me. To double check, I also measured this resistor on the good ones — they all measure 4.7 kilo-ohm. That’s about right. So clearly the manufacturer has made a mistake and this is really unfortunate. I’ve shot a video of the smoking ESP8266. See below.

esp8266_power_fail

Anyway, because some of these defective modules are shipped with the ESPToy, if you’ve bought ESPToy the past few days, you may end up getting some of these units. I sincerely apologize for this. It’s not anything I’ve anticipated. There are several options to solve the issue:

  • Option 1: let it burn (~~~). Once the LED is burned out and the magic smoke is gone, it will become open circuit and the problem is gone!
  • Option 2: if it’s scary to look at the magic smoke, or if the LED refuses to burn out, you can use a hot air gun or a soldering iron to remove either the red LED or the resistor next to it, or both.
  • Option 3: if neither option is ideal, send an email to [email protected] and we can arrange to send you replacements.

Well, this is just one of those headaches to deal with when buying goods from China. To be fair, we buy only from reputable sellers, even so, issues can happen, but we do gain more experience each time a lesson is learned 🙂

10 thoughts on “Magic Smoke Seen on Some ESP8266 Modules

  • January 7, 2015 at 9:42 am
    Permalink

    Hey Ray, may be a test jig for testing these babies… and sharing those designs 😛

    • January 9, 2015 at 1:05 pm
      Permalink

      Well, these modules came pre-packaged in a anti-static shielding bag, so we don’t usually test each one of them (and that would also be impractical), but we do sample test them.

  • Pingback: Faulty ESP8266s Release Smoke, Then Keep Working? | Hackaday

  • January 11, 2015 at 5:08 pm
    Permalink

    Hi,
    Did you measured the resistance removing it from the PCB? or just with the resistor soldered in place? If you do the measurement with the resistor in place, the measurement will be cheated by the rest of the circuit… Also 4.7kOhm for a led resistor is too high
    I suggest to remove the fake resistor, power again and see what happens, also measure out of the board… and measure out of the PCB…
    Regards

    • January 12, 2015 at 12:24 pm
      Permalink

      Yup, I measured the resistors after removing them from the PCB.

  • June 15, 2015 at 10:03 am
    Permalink

    Have you had any issues with ESP8266-1 modules getting hot like lava that had the correct resistors fitted?

    I have a couple of them that, when run in “CIPMODE=3”, pull a constant 200mA, no matter if data goes in our out & are “crash happy” and need a hardware reset all the time.

    I could get them to a more or less usable state by using “CIPMODE=1”, but I still have a feeling that something is not quite right (current down to about 150mA, less heat).

    A representative of espressif suggested I had either an antenna mismatch problem or power supply issues. I don’t think the latter is the case (regulated supply is pretty much a flat line). As for the antenna mismatch, that might be the case, but to me all those modules look like clones to me, so I guess either all of them are bad, or none – as far as the layout goes.

    How hot or warm do your modules get during sending / idle? Mine get so hot, that placing a finger onto the chip is very painful.

    Trying to decide if moving to a different ESP8266 module would be the way to go. I’d try one with an on-board chip-antenna.

  • July 23, 2015 at 2:01 am
    Permalink

    I have a ESP8266-1 , which stopped working and I cant install a boot loader in it. It also seems to get hot snf I could not communicate with it. I did some tests & found that the RX line is shorted to Ground somewhere under the ESP8266/MCU (not under the EEPROM which I could remove OK.
    Removing the 8266/MCU is just too hard, so in the bin it goes.
    FWIIW , as far as the ballast resistors on the LEDS, the RED led has a 2.2 K & the BLUE has a 4.7K
    Lee

  • May 15, 2016 at 11:27 am
    Permalink

    Hi,

    I have managed to burn 4 ESP8266 on a dev. board (https://github.com/nodemcu/nodemcu-devkit-v1.0), and the reason is still unknown for me.

    The history is like this:
    After I have burned the:
    – 1st one I removed GPIO10 from my code an also let if floating,
    – 2nd I have reduced the current for driving MOC3041 from 15 to 7.6mA,
    – 3rd I removed the ADC In setting (analogRead) from my code, even if I don`t use it for anything, yet.

    Now I have the following circuit:
    220V AC -> Power transformer from 220 to 6V (AC) -> Rectifier bridge (Vout ~8.6V DC) -> 2 capacitors (470uF + 100nF) -> LM7805 -> Power the ESPshield to Pin 15 (+) and Pin 7 (-) -> use Pin8/GPIO14 as an output to power an opto-triac MOC3041 (current was limited to 7.6mA using a 270ohm resistor) ->a triac to drive a load of 21W (7 x AC LED lamp) -> Pin1/GPIO16 used as an input (if pin state is changed than also the output is changed).

    After running the application ~15 minutes (Playing on/off via web requests) I have played also a little bit with the input pin and everything worked as expected. Then I have decided to let lamps ON (input pin was pulled to GND – Red led from shield lit), but after ~5 -10 minute the lamps turned off (Red led from shield remained on) and now the 4th ESP is dead.

    I have checked the value of current limiting resistor for the Red Led (470 ohm, value was obtained after I have removed the resistor from the shield) and from my calculations the current ~3.2mA is small enough not to damage the ESP, and now I have no idea what I`m doing wrong.
    If you think that is relevant I can provide also the code (written using Arduino IDE)

    Have anyone an idea that can help me with this issue?

Leave a Reply