1
\$\begingroup\$

I recently created a prototype PCB for ESP32-C3 WiFi-connected LED earrings. These were just quick proof-of-concept boards, and peak efficiency was not the goal. The issue is, I am unable to get WiFi to work. I was confined by space in the RF section of my PCB and was hoping to just get something that works enough for code development and testing. As you can see in the photos below, I have a Pi matching network which I based off a Phil's Lab ESP32 PCB Design video. The traces are not impedance matched but the total trace distance is very small. The circuit connects to a header which has a 2.4GHz dipole soldered onto it.

I understand that this setup is far less than ideal, and does not follow the proper RF standards. The purpose of this was to just get something that would work within a foot or two of the ESP32, so I wasn't concerned about getting the best efficiency. Both in AP and Client modes I get no signal from the ESP32 with my phone scanning for WiFi networks or acting as a hotspot, even right next to the antenna. Is the RF circuitry so sensitive that a poor matching circuit would completely prevent any effective signal in or out of the ESP32, or could there be another hardware/software issue at play?

PCB Schematic PCB Layout

\$\endgroup\$
2
  • \$\begingroup\$ Coaxial connectors exist for a reason. using a 100mil TH connector for your antenna will reduce your link budget drastically. \$\endgroup\$
    – Lior Bilia
    Aug 4, 2023 at 15:46
  • \$\begingroup\$ QFNs are hard to work with sometimes. Is it possible your antenna pin is shorted to the exposed pad, or not soldered at all? What soldering process was used? \$\endgroup\$
    – Bryan
    Aug 4, 2023 at 17:32

1 Answer 1

2
\$\begingroup\$

Even though the antenna and RF traces aren't perfectly impedance matched, you should have been able to detect a signal. I've worked with many wireless PCB designs (WiFi and BLE), and when you hold a smartphone nearby to scan, the RSSI is usually higher than -30 dBm. With receiver sensitivities down below -90 dBm, your imperfect antenna and matching network would have to cause a 60 dB loss. That's unlikely unless you have a dead short.

As far as I can tell, the ESP32 does some checking on its own RF power output -- they call it calibration -- to ensure the output power is within the specified limit, but I don't believe it checks for a high VSWR that would be caused by an impedance mismatch.

If you're willing to unsolder the antenna and matching network, you could just substitute a 1" long wire at the RF output (ungrounded end of C20) as a substitution that should give you a reasonable signal. If you still don't detect it, look for firmware/software and other human errors. Are the WiFi initialization routines correct? Are you sure your smartphone WiFi scan isn't limiting itself to 5 GHz?

\$\endgroup\$
2
  • \$\begingroup\$ Thank you for the tips. I removed the matching circuit and directly soldered on a 1 inch wire with the same results. I checked the CPU debug and saw that the WiFi functions were running correctly, and I checked to make sure there was no short on the RF pin. I also confirmed that everything was on 2.4GHz. Very weird \$\endgroup\$
    – Vera Fodor
    Aug 6, 2023 at 22:52
  • \$\begingroup\$ I had a similar experience. Used the ESP32-C3 chip directly, instead of the SOC. My PCB antenna chip didn't work, and soldering on a solid core copper wire of about that length boosted the signal, but it just wouldn't connect to the Wifi access point. In the end, I swapped it out for the ESP32-C3 SOC verson and it connects right away, no issues. Size was sacrificed, but it works. \$\endgroup\$
    – legoblocks
    Sep 16, 2023 at 7:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.