All Tasmota versions >12.x.y spontaneously resetting when connecting via Android for initial configuration

[DurvalMenezes]

Per title.

With the latest version (v15.0.1), here's what the serial console shows:

00:00:06.924 QPC: Reset

 ets Jan  8 2013,rst cause:4, boot mode:(3,3)

wdt reset
load 0x4010f000, len 3584, room 16 
tail 0
chksum 0xb0
csum 0xb0
v3969889e
~ld

00:00:00.001 HDW: ESP8266EX
00:00:00.009 UFS: FlashFS mounted with 2000 kB free
00:00:00.050 CFG: Loaded from File, Count 55
00:00:00.056 QPC: Count 1
00:00:00.067 Project tasmota - Tasmota Version 15.0.1(release-tasmota-4M)-2_7_8(2025-06-14T10:30:39)
00:00:00.270 WIF: WifiManager active for 3 minutes
00:00:00.778 HTP: Web server active on tasmota-4FBDCF-7631 with IP address 192.168.4.1

And then I try to connect again to the device by selecting its Wifi network on my Android phone, the above repeats, and the phone reports an error connecting,

v12.2.0 and v12.5.0 works as expected: the phone connects, shows the initial WebUI configuration page showing available Wifi networks, and asking which one and with what credentials I want Tasmota to connect to, and I can proceed normally from there.

But with all versions newer than v12.5.0 that I tested (to wit: v13.0.0, v13.2.0, v13.4.0, v14.6.0, and v15.0.1), the Android phone reports a failed connection and (obviously) the initial WebUI configuration screen never shows.

And yes, in all cases I followed https://tasmota.github.io/docs/Getting-Started/ to the letter.

Any clues?

EDIT: in case it matters, device is a Sonoff POWR2, straight from factory, never used (I can't stomach Sonoff's proprietary firmware nor their ecosystem).

[sfromis]

In case you're not powering via mains, but wired from some USB/serial adapter, the issue may be lack of sufficient power to get on Wifi. Known issue with many borderline devices.

And do not try powering via mains when having USB connected. Could destroy hardware, at the other end of the cable. If you want to try to add power, you could add an additional 3.3V power source.

[sfromis]

It does explain, as you can have no "logical" explanation when your device is "borderline". Tiny variations can push it over the edge to failing. Different Tasmota versions are not exactly the same.

[DurvalMenezes]

But a "borderline" device would cause "borderline" (ie, random) behavior, no? What I see here is 100% reproducible behavior, and that's what your hypothesis do not explain. And of course different Tasmota versions will never be exactly the same, otherwise there would be no pointing in releasing them, right? Anyway, thanks again for your insightful, considered and intelligent comments, much appreciated.

[sfromis]

No, borderline is very different from random, it is instead about barely working vs barely not working - depending on things like power draw. And a device with Tasmota might draw a bit more power with a newer version, not only due to Tasmota code, but also upgrades library code Tasmota depends on. Meaning that borderline hardware may just reach the point of no longer being able to work.

As I already said:

Tiny variations can push it over the edge to failing. Different Tasmota versions are not exactly the same.

[DurvalMenezes]

Posting this for the benefit of those that will eventually come a-googling:

1. Disconnecting the adapter and plugging the POWR2 into mains right after flashing for the initial setup did not work with any version >v12.x.y, therefore falsifying @sfromis hypothesis that the issue was being caused by what he called a "borderline adapter" not supplying enough power for the initial operation.

2. What did work was flashing Tasmota v12.5.0 first, then doing the initial setup (on which, as I reported, v12.x worked without fail every time), then upgrading to v15.0.1; it all worked perfectly with that same "borderline adapter" and with no other power source, therefore corroborating the conclusion of 1) above.

3. Besides doing the initial setup and then upgrading, I was also able to fully operate the device, including toggling the relay on and off (verified by its clicking noise and using a DMM to confirm the resistance on its contacts went from infinite to zero and then back) both with v.12.5.0 (after the initial setup but before the firmware upgrade) and v.15.0.1 (after the upgrade), all the time without any other power supply but the so-called "borderline" adapter. This further corroborates 1) and 2) above.

I hope that this report can be of benefit to anyone else caught with a "non-recommended/borderline" adapter and more interested in solving the issue than engaging in useless discussions. I'm now out (and unsubscribing from this discussion).