In the previous circuit, when the ESP32 tried to connect to WiFi, the voltage dropped and the ESP32 rebooted. This time, I'll change to a DCDC converter that can output a larger current and experiment to see if I can make the ESP32 work with about two batteries.
The ISL91107IRTNZ is a 3.3V, 2A maximum output step-up/step-down DCDC converter IC that operates from 1.8V to 5.5V. It is very simple, requiring only ceramic capacitors for input and output, and an inductor. The graph on the datasheet shows that it can output more than 1A at 2.5V. This should be enough to cover the high current required for the ESP32's WiFi connection.
Here is the IC. As before, it is a very, very small IC with a pitch of 0.5mm.
It was so small that I was barely able to solder it while looking at the digital microscope and shaking my hands.
A digital microscope is a very useful tool. Click here for details of the digital microscope I use.
I managed to complete the soldering. I used 0.1mm polyurethane wire for the wiring.
Check the operation
I supplied about 5V of electricity from a regulated power supply. The DCDC converter seems to be working properly.
For more information about the regulated power supply used in this project, please refer to here.
The moment I tried to connect to WiFi, the output voltage of the DCDC converter dropped to 0.8V. This DCDC converter IC may be a fail-safe, but once it goes to 0.8V, it will not return to 3.3V even if the load is disconnected. It is designed in such a way that 3.3V will not be output until the power is turned off and then back on.
Slowly increasing the supply voltage from 0V is also not recommended, as it will cause an overload at low voltage, and the output will remain at 0.8V even if the input voltage is increased to 5V.
Once it enters the overload state, it will not output power unless the power is turned off and on again.
How many V is needed to prevent overload?
I tried to find out what input voltage is required to prevent the ESP32 from being overloaded when connected to WiFi. We found that the input voltage needed to be at least 3.5V. It seems that even a DCDC converter with an output of 2A cannot drive the ESP32 at such a low voltage as two batteries to handle its high current.
I added a resistor
I'm going to try a little idea I had. To limit the overcurrent, I'll put a 1 ohm resistor between the DCDC converter and the ESP32, and add a 100uF tantalum capacitor to absorb the shock of the sudden high current.
Then, perhaps because the inrush current during WiFi connection was mitigated, the DCDC converter did not become overloaded and did not stop even when the input voltage was lowered from 3.5V to 2.9V.
Lowering the voltage any further caused the ESP32 to brownout reset; it seems that by adding a 1Ω resistor, the output voltage of the DCDC converter drops and reaches the brownout voltage when connected to WiFi.
When I tried setting the resistor value to 0.5Ω, the DCDC converter quickly overloaded and there didn't seem to be much change from when I didn't put in the resistor. On the other hand, when I set the value to 2Ω, the voltage drop was too large, and the brownout reset occurred frequently. Also, when the 100uF tantalum capacitor was increased to 470uF, the DCDC converter was overloaded by the inrush current at power-on and did not operate properly.
For this reason, a combination of 1Ω and 100uF seems to be a good choice.
Maybe it works for the previous circuit too
The fact that it worked in this circuit means that it might also work in the TPS63001 step-up/step-down DCDC converter circuit that I made before. So I added a resistor and a capacitor to see if it would work.
I added a 1 ohm resistor and a 100uF tantalum capacitor to the output of the DCDC converter, and found that when I gradually lowered the input voltage to the DCDC converter, it worked fine down to 2.8V. This is a big improvement from the 3.6V without resistors.
The TPS63001 circuit was able to communicate with WiFi up to 0.1V lower than the ISL91107IRTNZ without brownout reset.
It seems that the ESP32 needs some kind of current buffer with resistors and capacitors because the current is too large when connecting to WiFi.
However, it seems impossible to operate the ESP32 with two batteries, which was the original goal. At least three batteries are needed.
I updated the mailbox, which notifies me by e-mail when mail arrives, with this circuit. It used to stop functioning when the battery voltage dropped to about 3.6V, but now it keeps working up to 2.9V. It should work longer than before.
You can read more about this mailbox here.
I'm going to use it for a while and see how it works.