A quick guide to building your first browser-flashed internet radio without the usual headaches

That UART port catch got me too, though in my case I wasted an evening flashing the wrong socket and wondering why nothing showed up in the browser. Now I put a strip of tape over the other port so I stop doing it every time I reflash. My build runs a PCM5102 board into a little 3W amp from an old computer speaker set, and I had a diffrent noise problem than you describe. Mine buzzed worse the moment the wifi connected. Turned out the antenna end of the board was sitting right next to the amp input wires on my perfboard. Shuffled the layout so the DAC sits at the far corner and the buzz dropped to almost nothing. So for anyone planning a layout, distance from the radio side matters as much as wire length.

I also skipped the paper planning step and paid for it. Had to desolder three jumpers because I forgot the rotary encoder needed pins I'd alredy given to the display. The editor warning popped up and I dismissed it, thinking I'd work around it later. I did not work around it later. One thing I'd add from my setup: I power the whole thing from a 5V phone charger and it reboots whenever the volume gets loud. Suspect the amp pulls more than the charger likes at peaks, so a beefier supply is on my shopping list.

I wonder if anyone has empirically measured the thermal profile of thier finished assemblies, because sustained high-bitrate audio decoding and streaming places a surprisingly significant load on the silicon. If the ESP32 core reaches elevated temperatures, the clock stability and the radio frequency output coud become compromised, introducing subtle timing artifacts that manifest as audible digital noise. I suggest paying close attention to adequate passive cooling solutions, or even forced air movement if the device is housed in a restricted enclosure.

I totally agree with the part about power supply limits; my own radio reboots happened for the same reason, expecially when the audio processing spiked. It’s frustrating when you spend all that time on the code and the board just quits because of current draw. I fixed this by adding a couple of large electrolytic capacitors right across the power rails near the board. It smoothed out those peak demands and stopped the unexpected reboots without needing to upgrade to a massive power supply.