Slow WiFi connection on one ESP32-C3, but not on the other

[requireiot]

I have two identical ESP32-C3 modules (WeAct ESP32-C3FH4 Mini Core Board), running identical Arduino-based software. They connect to a WiFi AP (AVM Fritz!box 7530) about 3m away.

One of them (let's call it the breadboard) taken 400-500 ms to connect to the WiFi network, i.e. the time between calling WiFi.begin(user,pass) and when WiFi.status() returns WL_CONNECTED ... this is what I would expect, no complaints.
The other module (let's call it the box) takes >2000 ms to connect to the WiFi AP, every time ... this is very slow, imho.

The breadboard is the module on a , well, breadboard, powered by a USB3 connection. The box is in a small plastic case, powered by a battery, or also by a USB3 connection, makes no difference.

Can't be because of location ... they both sit side by side on my desk.
Can't be because of a blocked WiFi antenna ... in both units, there is no electronics/metal/pcb underneath or above the module's WiFi antenna

Can anyone explain this difference in behavior?

[Aria James]

The slower WiFi connection on the "box" module could be due to antenna performance (affected by enclosure), power stability (voltage drops during WiFi init), firmware corruption, or factory calibration differences in NVS. Try reflashing firmware, erasing NVS, and ensuring stable power. Testing another board can help diagnose hardware issues.

[requireiot]

The slower WiFi connection on the "box" module could be due to antenna performance (affected by enclosure), power stability (voltage drops during WiFi init), firmware corruption, or factory calibration differences in NVS. Try reflashing firmware, erasing NVS, and ensuring stable power. Testing another board can help diagnose hardware issues.

Thank you for those suggestions. Let me address them one at a time:

• "antenna performance (affected by enclosure)": this is a plastic enclosure, with no other PCB or chunks of metal above or below the antenna region of the ESP module

• "power stability": That was my initial thought, too. However, I get the difference in performance between the 2 modules even when I power them from the same USB3 outlet

• "firmware corruption": I have re-flashed the module several times, no change

• "factory calibration differences in NVS": that I can't influence, of course. By "NVS", do you mean a system-reserved section of flash memory, or RTC memory ? Either way, wouldn't know how to erase this

[Sprite]

• "factory calibration differences in NVS": that I can't influence, of course. By "NVS", do you mean a system-reserved section of flash memory, or RTC memory ? Either way, wouldn't know how to erase this

It's in flash memory. Easiest way to clear this is to do a full flash erase, e.g. using `esptool.py erase_flash'

[lbernstone]

Cheap boards use cheap components. There is a ceramic antenna on those boards that was matched with a CLC filter and tuned in a network analyzer. This was probably done once in a lab with better quality parts than were actually used in manufacture. So, if the capacitors were not well matched, or the antenna is turned just so on the pads, then the reception will be poor.

[requireiot]

@ESP_Sprite: I tried your erase_flash suggestion, but it made no difference

@lbernstone : this board does not have a ceramic antenna, it has the typical zigzag PCB layout antenna