milika:
I just wire gnd to usb shield and now works perfectly from external power.
Seems there is a component between board gnd and usb gnd...
Hello milika greetings from Mexico, what pins or how did you connect GND to the USB port? There are also some errors in the letters it seems to me, of some pins like instead of saying GND it should say CMD. When I put 5 to 7 V to vin, the led blinks, I have to put up to 9v for it to stabilize. This card is a generic Dev kit v4. I also have one with an IPX connector, which is more stable at 7v, but it fails when I input 5v. Connected both cards to the USB port they work fine. Thanks for your answers.
It is strange that this issue has not been located by now. I have been using this device for a couple of years now to operate a string of lights that are in a remote location. I control to power through a sonoff switch and used the OTA system to change the sequences. About two months ago,(AUG2020), I discovered that I was unable to communicate with the device through the OTA. I can not ping the device either. Due to the location, I decided to setup a new one if I was going to have to access the device. I set up a breadboard and soon found that five different devices were all having this issue. They run fine with the usb cable connected , including the wifi, but as soon as the usb cable is removed and the bench supply installed they are dead. The bench supply shows they are still drawing current but I have no idea what they are doing. I have tried all the above "fixes" but no luck. The OTA code that I am using came from the esp32 course from RNTlabs. The device had proven very reliable until now and far more stable and reliable than any of the ESP8266 OTA things that I had tried. I really thought this was a keeper.
I wonder if there has been some library change that is causing this. I just don't have the background to root out a problem in a library but it seems funny that it effects more than one manufacturer. I have boards that use the Espressif device and others that use the WROOM device.
I am using Arduino 8.1.13 on a Win10Pro machine updated.
I went to Espressif forum website and found a workaround from a poster regarding the same subject. I don't think the problem is caused by the chip maker as much I think it may be caused by the myriad of board assembly people. The answer that was supplied was to connect the power to the device with the usb cable. I loaded my basic OTA code to the device from the PC, then removed the cable from the PC and connected it to my phone charging hub. I also had to make a connection to the 5vdc line on the led string. Everything works, leds light up, I can change the sequence with the wifi OTA and that's where I was a year ago before all this mess started. Hopefully someone from Espressis will take note of that in their forums.
YMMV
KentM
This topic was begun more than two years ago, and still remains a problem. I use an ESP32-Dev kit, and when powered via USB bus, it all works fine, always.
But as I am preparing for a mobile device, I tried to use a Li-Ion (nominally 3.7V, fully charged 4.2V) via a Buck-Boost board set to 3.3V and feed the power into the 3.3V pin of the ESP.
This consistently fails with a brownout at the moment of the WiFi.begin command. I tried various Buck-Boost boards including ones delivering 1.5A or even 3A acc to their specs. All fail. Even putting a 3300µF cap across the 3.3V line did never help.
On the oscilloscope I can barely see a voltage drop on DC coupling. On AC coupling with higher amplification I see a drop that at least time-wise seems to be related to the brownout, but its magnitude is only ~100mV on a slow sweep. Supposedly the ESP can handle voltage from 2.3V up to 3.6V. The observed drops seem to be too small to explain the brownout, or not?
When I use the same batteries as above and first use a boost-up converter to 5V, and then feed this to the ESP via the USB port, it also works fine. Which makes me wondering: when feeding power through the 3.3V pin, am I feeding something quasi in "reverse mode" into the circuitry, which consumes a lot of power?
I have found that the ESP32 does not like getting switched 3.3V.
If I power the ESP32 from 5V I use a switched P.S. and put a polarized cap on the 3.3V pin to gnd. When I power the ESP32 with 3.3V, I always use a MCP1700 LDO 3.3V regulator with proper input/output filtering.
@Idahoewalker: the MCP1700 LDO 3.3V is rated at 250mA output, which is only half of the ESP32 Datasheet recommendation of 500mA:
Notes on power supply:
• The operating voltage of ESP32 ranges from 2.3 V to 3.6 V. When using a single-power supply, the
recommended voltage of the power supply is 3.3 V, and its recommended output current is 500 mA or
more.
You are running your ESP with this sole supply, including WiFi?
I am running 5 different ESP32's from a MCP1700 with WiFi, 24/7.
5 of the MPC1700's are being fed 5V, switched, The output of the MCP1700 is filtered with a ceramic cap and a 47uF Tantalum camp.
2 of the ESP32's are ran by solar power. When I did the solar power project, I noticed that if I supplied the ESP32's with 3.3V, using a MCP1700, battery life was longer by about 5+ hours.
One of the 2 ESP32's that is solar powered sends 11 MQTT publications per second, 24/7. I plan on slowing that down at some point. I had tried using regular LiPo's but found they were not hands off. I installed a PWM Controller and a 12V 16Ah LiFePo4 battery. The battery has been working since Oct. 2020. I was curious what performance I could get from the battery. I took out the deep sleep programing, upped the MQTT transfer rates, and waited to see if the battery would fail/drain during the winter. So far the battery has not needed any attention from me and the ESP32 transmitting 11 times a second being supplied by a MCP1700 did not fail.
Well, that is some convincing evidence! Looks like Espressif uses some good padding in the power specs.
But when 250mA is sufficient, isn't it even more surprising that these BuckBoost devices cannot deliver the power?
Basically I am doing the same thing as you do when using my ESP32-Dev module from a battery: from a LiIon (nominal 3.7V) boosting to 5V (in things marketed as "18650 shields"), fed into the USB connector, fed to the AMS1117, fed to the 3.3V line. It always works reliably.
With the MCP1700 having a much lower dropout voltage, there should be a benefit on batt life. I'll try.
It is getting stranger: I took the Li-batt setup as just described (post #29) but instead of feeding the boosted-up 5V power to the ESP32-Dev-Module via a USB cable into its USB port, I connected the 5V to the module's 5V pin, and left the USB port open. Result: perpetual brownouts!
This can't be a power effect, because the very same power source worked perfectly well via the USB port!
I went one step further and fed the 5V power into an external AMS1117 (same chip as used on the Dev Module) to create 3.3V, and fed this to the module's 3.3V pin. Same result: perpetual brownouts!
Seemingly the only thing different is the extra CP2102 USB-to-Serial chip on the module. Is it possible that these "brownouts" have nothing to do with a lack of current, but with some unidentified interference from the CP2101?
@Idahowalker: are you using ESP32-Dev-modules, or are you using the plain ESP32 packages? with/out any USB-to-Serial?
hi , I am facing this problem too, the ESP8266 keep disconnecting from the WiFi on external power supply but not on the microUSB from the board itself. Who else encounter this problem too ?
