ESP32 Forum

Hello,
I’m designing my own PCB, and I’m using the esp32-wrover-b module. My question is about the use of capacitors on the 3V3 line. It appears that this power input must be filtered with a 100nF (0.1uF) capacitor -standard practice. It also appears that a 22uF (or even bigger?) capacitor must be added for power reserve purposes. I’m using a 1.4 MHz step down DC-DC converter, and it can provide up to 1.5A output current. This converter has a resistor feedback regulation network. Given this type of regulator, do I still need the extra capacitance on the 3V3 line? Can I get away by just using the 0.1uF filtering cap? Thanks in advance for your help.

It depends on a bunch of things, like the distance (and resistance/inductance) of the trace between the DC/DC and the ESP32, as well as how quickly the DC/DC can react to changes in power usage.

I'm using a dual layer design (as opposed to the 4 layer recommended one). I designated the top layer 3.3V and the bottom layer Ground, and mounted the ESP on the top power layer. This means the esp32 can just get the power by making the 3v3 pad a solid or thermal relief connection. The DC/DC reacts very fast to voltage changes, and it has proven itself very reliable in another design. I'm tempted to just put the 100nF filtering cap in there and leave the beefier one out. I just didn't want to send the PCB to prototype manufacturing knowing I could potentially have problems powering it (I know, that why it's called prototyping). Because of this, I'm looking for a development board that has just the GPIO pins and nothing else (and by the looks of what's out there, I might have to make those first). We have already designed the DC/DC converter separately on it's own board (from the previous project).

If it's a prototype, why not leave a footprint for a 22uF cap but not populate it?

Yes, that's what I'm going to have to do. I don't have much room for it. I'm going to have to remove something in order to make room for that, at least I'll know it works for sure before anything else. I know this is trivial but this is due to the fact that we're trying to move from an existing design (another MCU), to the ESP32 without making the PCB bigger and without compromising any component layouts from that design. So far, it's looking good, except for this little quirk.

Thanks for your input!

I thought i upload a screenshot of what I did in KiCad, I had a few boards produced and I'm doing some electrical testing of them. So far, everything is holding good and no need for the extra 22uF cap. I still need more testing though , it's looking like I can remove the cap and make more room in the area. Also, I'm designing an external programming tool for when the ESP needs reprogramming once it has been soldered onto the board (no 10k resistor, no 0.1 capacitor soldering per unit should the ESP need reprogramming). Since I was in a rush, I added some header pins for the serial, enable, and boot connections (those latter two with a simple jumper for when the programming is needed). I plan to replace those with solid pads for pogo testing pins.