Power supply for the ESP32

[Andreas]

Hi there,

@clemens just reminded me that I promised to drop some words about issues coming from insufficient or unstable power supply to ESP32-based modules and how to mitigate them. We researched this issue together with @tonke the other day, thanks!

Hint: If you are encountering problems like described here and you will not be able to make it work just by swapping the USB cable, a little amount of soldering might be involved.

Reference

@clemens

Hier gibt es ähnliche Berichte zu Strom-Problemen beim ESP32 nur oder “schon” im WLAN-Betrieb.

https://arduino-hannover.de/2018/07/25/die-tuecken-der-esp32-stromversorgung/

Auch hier wird die Lösung @Andreas freuen (running gag) mit dickeren bzw. kürzeren Kabeln konnte man das Problem auch hier lösen!

With kind regards,
Andreas.

[Andreas]

Introduction

To get started, the official documentation from Espressif has an appropriate troubleshooting section in its esptool documentation. Please read this carefully, especially when it comes to debugging issues revolving “stable power supply” after experiencing connectivity or responsiveness problems with your board.

TLDR;

Things might go south if power circuit fails to meet the large spikes of ESP32 power demand when using WiFi and/or when attached peripherals draw too much power.

• Check your USB cable.
• Check your power supply.
• Apply the cap trick by adding a missing capacitor to your board.

Power supply

The official documentation already has a detailed section about insufficient power supply, see troubleshooting insufficient power.

Insufficient Power

The 3.3V power supply for the ESP8266 and ESP32 has to supply large amounts of current (up to 70mA continuous, 200-300mA peak, slightly higher for ESP32). You also need sufficient capacitance on the power circuit to meet large spikes of power demand.

Insufficient Capacitance

If you’re using a pre-made development board or module then the built-in power regulator & capacitors are usually good enough, provided the input power supply is adequate.

This is not true for some very simple pin breakout modules - similar to this. These breakouts do not integrate enough capacitance to work reliably without additional components.

Surface mount OEM modules like ESP-WROOM02 and ESP-WROOM32 require an external bulk capacitor on the PCB to be reliable, consult the module datasheet.

Power Supply Rating

It is possible to have a power supply that supplies enough current for the serial bootloader stage with esptool, but not enough for normal firmware operation. You may see the 3.3V VCC voltage droop down if you measure it with a multimeter, but you can have problems even if this isn’t happening.

Try swapping in a 3.3V supply with a higher current rating, add capacitors to the power line, and/or shorten any 3.3V power wires.

The 3.3V output from FTDI FT232R chips/adapters or Arduino boards do not supply sufficient current to power an ESP8266 or ESP32 (it may seem to work sometimes, but it won’t work reliably). Other USB TTL/serial adapters may also be marginal.

[Andreas]

Issue collection

You might want to read from others having similar issues with ESP32-based development boards from different vendors.

Faulty USB cable

Motto: Check your cables and wiring. In doubt, swap them.

Try just change your USB cable (I got some flash problems with a cheap cable).
– A fatal error occurred: Failed to connect to ESP32: Timed out waiting for packet header · Issue #333 · espressif/arduino-esp32 · GitHub

For me it was a bad usb cable.
– A fatal error occurred: Failed to connect to ESP32: Timed out waiting for packet header · Issue #333 · espressif/arduino-esp32 · GitHub

Power supply

For me it was a power issue, I am using a EzSBC esp32 dev board. Interesting it happens when I upload a sketch that requires a lot of pin usage. The way I “fixed” the situation was to apply 3v 2amp external supply to the 3v pin on esp32, plus an external ftdi to rx/tx, with common ground. Applying 5v 2amps to Vin on module did not solve the problem.

– https://github.com/espressif/esptool/issues/226#issuecomment-367594576

Figured out the issue. Looks like the FiPy was consuming more current that my USB was able to deliver […]. I connected a LiPo battery to it and didn’t have an issue since.

– <SOLVED> FiPy not detected on Win10 (using PySense) | Pycom user forum ganz unten.

Missing cap

Fixing this issue already goes by “applying the cap trick” ;]. Enjoy.

Some third party ESP32 development boards use an automatic reset circuit for EN & GPIO pins, but don’t add a capacitor on the EN pin. This results in unreliable automatic reset, especially on Windows. Adding a 100nF (or higher) value capacitor between EN pin and GND may make automatic reset more reliable…

– Home · espressif/esptool Wiki · GitHub
– via: Establish Serial Connection with ESP32 - ESP32 - — ESP-IDF Programming Guide latest documentation

---

I have made this mod and it works every time on the faulty units now.

– A fatal error occurred: Failed to connect to ESP32: Timed out waiting for packet header · Issue #333 · espressif/arduino-esp32 · GitHub

---

I solved this problem by attaching a 100nF capacitor from EN to GND. I use a pure ESP-WROOM-32 with an FTDI board and the 2 transistors between DTR/RTS/EN/GPIO0.

– Failed to connect to ESP32: Timed out waiting for packet header (ESPTOOL-312) · Issue #226 · espressif/esptool · GitHub

---

The cap trick also worked for me! Keeping EN low while GPIO0 is toggled seems to do the trick. I just grabbed the first cap that came to hand - it’s a 22uF.

– Failed to connect to ESP32: Timed out waiting for packet header (ESPTOOL-312) · Issue #226 · espressif/esptool · GitHub

---

If you add a capacitor between the EN pin and GND (in the 1uF-10uF range) then automatic reset into flashing mode will probably become more reliable. Not all development boards have enough capacitance on this pin.

– Failed to connect to ESP32: Timed out waiting for packet header (ESPTOOL-312) · Issue #226 · espressif/esptool · GitHub

[Andreas]

@Andreas

Figured out the issue. Looks like the FiPy was consuming more current that my USB was able to deliver […]. I connected a LiPo battery to it and didn’t have an issue since.

Please be aware when running your device on battery: This is obviously very much prone to unstable behavior when running on low voltage, see also A fatal error occurred: Timed out waiting for packet content - problem after writing to SPIFFS memory of ESP32 (ESPTOOL-403) · Issue #263 · espressif/esptool · GitHub. So, please watch and check the charging state of the battery before wondering about erratic behavior.

[Andreas]

Closing the loop

by citing from the article @clemens originally just shared with us (thanks again!),

@clemens

Bei Die Tücken der ESP32-Stromversorgung | Arduino-Hannover gibt es ähnliche Berichte zu Strom-Problemen beim ESP32, gerade im WLAN-Betrieb.

Helmut gives very specific advice:

[de]

[Es gibt] zwei Möglichkeiten, die Stromversorgung weiter zu verbessern: man nehme einen 47-µF-Tantalkondensator, der direkt am ESP32-Modul zu den Pins VDD/GND parallel geschaltet wird. Dieser ist in der Lage, die geforderten Stromspitzen schneller als der LDO-Regler bzw. andere Kondensatoren zu liefern. Zusätzlich zum Tantalkondensator gibt es dann noch die Möglichkeit, einen großen Low-ESR-Elektrolytkondensator mit einer Kapazität von 470 µF bis 2000 µF zu verwenden. Der Elko ist dann in der Lage, den Tantal nachzuladen, womit die Spannungsversorgung weiter stabilisiert wird.

[en]

[There are] two ways to further improve the power supply: take a 47-µF tantalum capacitor, which is connected in parallel to the VDD/GND pins directly on the ESP32 module. This is able to deliver the required current peaks faster than the LDO controller or other capacitors. In addition to the tantalum capacitor, a large low-ESR electrolytic capacitor with a capacitance from 470 µF to 2000 µF can also be used. The electrolytic capacitor is then able to recharge the tantalum, further stabilizing the voltage supply.

Bottom line

[de]

Diese Maßnahmen (dicke Kabel und zusätzliche Kondensatoren) kosten nicht viel, lösen aber die Probleme!

[en]

These measures (thick cables and additional capacitors) do not cost much, but solve the problems!

[Andreas]

Haben wir die Stromversorgung des FiPy eigentlich ordentlich im Griff? Hier gibt es auch eine Diskussion dazu, um sich ggf. noch ein paar Anregungen abzuholen. robert-hh ist am Start!

Voltage regulator for primary battery