I thought I should share my experience having a seemingly somewhat common issue that other threads here advise solving through firmware updates, but actually turned out to be a hardware problem.
Recently I bought a R4 WiFi from what I believe is the official store on Amazon. Right out of the box I couldn't upload sketches to it from Windows or Linux with the known "no device" error from bossac (with the board led properly switching to flashing in the bootloader mode):
No device found on COM4
No device found on ttyACM0
I couldn't update the firmware from Windows either, but on Linux it worked and seemed to fix the problem.
Fast forward to yesterday, when I uploaded a sketch that flooded the serial output line due to a print in a too tight loop, making the TX led basically stay constantly on, like here.
This effectively killed the ability to upload either sketches or firmware from the IDE on either OS, just like in that thread.
Using the stand-alone ESP32 firmware update script it was possible to flash that chip firmware from linux just fine, but it didn't affect the problem. The bootloader flash procedure with 2400 baud and rfp-cli also wasn't working with
[Error] E3000105: The device is not responding.
I tried soldering the RA4M1 jumper and flashing the bootloader using the Minima procedure (and deleting that sketch from memory while at it), which worked perfectly fine, but didn't solve the problem after the jumper was removed.
Finally I broke out my oscilloscope to test the serial lines between the two chips, and found a bad solder joint on the A8 pin of the TXB level converter chip, which effectively cut the serial data line transmitting from RA4M1 to ESP32, so the upload software could not get any reply from the chip. The sketch putting high load on that serial line may have fried the already flaky joint.
While trying to directly test the pad on the side of the chip with a sharp probe I must have mechanically pinched the solder blob to the pad, because it actually started working; but it clearly needs a proper reflow.
I must say that under a strong magnifying glass the SMD solder joints on the board look quite grainy and bad, compared even to through-hole connector joints on the same board, so I think the manufacturing process for these boards may need a few tweaks.
