Building a Custom Low-Power ESP32 Dev Board

The idea behind this project was straightforward enough: build an ESP32 development board that could run for months on a single lithium cell. I wanted something genuinely low-power, not the usual development board drawing a couple of milliamps even when it’s supposedly asleep. And, truthfully, I wanted to make it myself. There’s not much satisfaction in just buying one.

I started with the core design goal of reducing quiescent current wherever possible. That meant stripping away the usual luxuries, no onboard USB-to-serial converter, no linear regulator, and no indicator LEDs sitting there quietly wasting energy. To achieve this I opted for the Texas Instruments’ TPS63802, a 2 A buck-boost converter chosen for its impressively low standby current. Power would come from a single Li-ion cell through a small charging IC, while USB-C would handle data and charging duties. It looked good on paper: simple, efficient, modern.

It was only when I ordered the parts that I realised what I’d done. The TPS63802 comes in a 12 × 2 mm SDFN package with ten pads hidden under the body. I knew it was small, but seeing it in person was another matter entirely. The USB-C connector was no better; the pins are so fine they almost disappear without magnification. These two components alone set the tone for the entire build.

Figure 1: Final PCB design of my development board – the TPS63802 is U3!

The boards arrived from JLCPCB about ten days later, in a box so large it looked like I’d ordered an oscilloscope. In reality, the size was there to protect the solder stencil, which has to stay perfectly flat during shipping. I followed a few online suggestions and taped three spare PCBs together to make a simple alignment frame for the stencil. It worked well enough, though I realised too late that the stack of boards had lifted the stencil slightly above the surface of the target PCB. That tiny gap would later cause a lot of trouble.

Figure 2: Framework of PCB’s to keep the central one fixed in position

The solder paste went on with an expired bank card and a bit of misplaced optimism. I used a cheap, low-melt paste rated for around 180 °C, reasoning that it would be gentler on the small parts. The paste looked uneven in places, particularly around the USB-C connector, but I convinced myself it would sort itself out during reflow. Surface tension is supposed to take care of that, after all.

Figure 3: Smearing on the solder paste was easy, still I missed a capacitor!

Component placement was fairly painless. Ceramic tweezers handled the smaller parts, and I used my fingers to nudge the ESP32 module roughly into position and the alignment turned out surprisingly accurate. In total, placement took maybe fifteen minutes, including the time spent hunting for stray 0805 capacitors that had vanished into the carpet.

Figure 4: Component placement was easy and didn’t take as long as expected

The hot-plate reflow was the most satisfying part of the whole process. I don’t own a reflow oven, so I used a magnetic-stirrer plate set to about 250 °C, comfortably above the paste’s melting point. After a minute or so the solder turned shiny and the components literally moved into place as the surface tension pulled them onto the pads. It was oddly mesmerising to watch. When it cooled, the board looked flawless.

Figure 5: Everything looked to be in the correct place after heating

I set up my budget HDMI microscope and immediately understood how easy it would be to destroy anything that plugged into the USB port. Under normal lighting, the pads looked perfectly separated; under magnification, they were a minefield of potential shorts.

Figure 6: The USB-C port looks like a complete mess!

Then came the real battle: the USB-C connector. At a glance it looked fine, but under the microscope I could see multiple bridges along the data and power lines. I tried to clear them with hot air and solder wick, but the results only got worse. My cheap iron without proper flux didn’t help matters. Eventually I accepted defeat and removed the entire connector using a small hot plate I’d bought from AliExpress. The removal itself was easy; the aftermath was not. I hadn’t cleaned up the excess solder, and when I placed a new connector on the still-warm board, a perfect solder bridge appeared between VBUS and GND. The second attempt at cleanup was more careful: flux, patience, and a lot of gentle brushing with copper wick until the microscope finally showed clean pads again.

Figure 7: Installing a new USB-C port, making the same mistake again!

The first power-up was a disappointment. The 3.3 V rail was dead, and the TPS63802 was getting warm enough to be worrying. I put the board back on the hot plate, removed the chip with tweezers, cleaned the pads with solder wick and plenty of flux, and tried again. This time the converter behaved itself, and I finally had a stable 3.3 V output.

After all that, the board powered up perfectly. A basic LED blink sketch compiled and ran without issue, and even the Wi-Fi test passed first time. More importantly, the power system remained cool and stable throughout. The current draw in deep sleep was exactly where I wanted it, suggesting that all the effort might actually have been worthwhile.

In the end, the board performs exactly like any commercial ESP32 development board, except it idles at microamp levels instead of milliamps. The experience taught me several humbling lessons about hand-soldering modern SMD parts. Stencil flatness matters more than I’d ever realised, and flux really is the difference between success and disaster. The USB-C connector, in particular, demands a level of precision that’s almost absurd for hobbyist assembly.

Despite the frustrations, I’d call it a success. The board works, it meets its design goals, and I learned more about reflow technique in two evenings than I had in the previous year. I’ll take that as a win.

Here’s a link to the mini hot plate I used to re-work the USB-C port:

MHP30 PD65W Mini Hot Plate Digital Soldering Preheating Rework Station (434 sold, 4.9*)
https://s.click.aliexpress.com/e/_c3YUSfnV

There’s also the low temperature solder paste:

Lead-containing and 181-degree silver-bearing solder paste (1000+ sold, 5*)
https://s.click.aliexpress.com/e/_c4MxvGAr

For the soldering microscope, here are the components I used:

HD1080P HDMI VGA Industrial Video Microscope Camera (4.7*, 60 sold)
https://s.click.aliexpress.com/e/_oD10RtT

Magnification Adjustable 180X 150X 130X Zoom C Mount Lens (4.9*, 371 sales)
(get the x180 version)
https://s.click.aliexpress.com/e/_ooByeXb

SmallRig P96L RGB Video Light Portable Metal LED (5*, 19 sold)
https://s.click.aliexpress.com/e/_oBrd4M5

(My favorite!) TS80P Mini Smart Portable Digital Soldering Iron (4.7*, 63 sold)
https://s.click.aliexpress.com/e/_okuOplT

To clean the tip, I like these copper wool things: Desoldering Cleaning Ball (4.9*, 2000 sold)
https://s.click.aliexpress.com/e/_oEkgdY1

To hold the iron when not in use, I use this stand: Soldering Iron Bracket (4.9*, 4000 sold)
https://s.click.aliexpress.com/e/_olF15yh