Let's Control It

TD-er wrote: ↑15 Jun 2021, 23:38 Just to be sure... have you tested with reduced WiFi TX power?

Yes, did that in steps of -2.5 till 0. No success:

Tebogo wrote: ↑09 Jun 2021, 12:31

Can you make a better close-up of the one that doesn't work?

See attachement's

I have never seen a module like that.

I can send you one if you like, pm your address.

It does look like a beefy AMS1117 as voltage regulator, but I don't see a lot of other larger capacitors.
Only one larger yellow one.

See, pictures, this is beyond my knowledge..

Where is the antenna located?

Under the WeM*s logo

Also I would like to see a close-up of the parts next to the ESP chip.
On the right I do see a crystal, but since I have no clue where the WiFi antenna is located, I wonder if there are parts interfering with the antenna tuning.

Can you also get a close-up of the parts surrounding the 3v3 regulator? (the chip next to the CH340 and the yellow capacitor)
If you look at the NodeMCU board, you can see there are quite a few larger capacitors next to the voltage regulator.
When I design a board for myself, I always use 100 - 330 uF on 3v3 and a 22 uF close to the ESP.
Also 100 nF capacitors as close as you can next to the Vcc and GND pins of any chip like the CH340, ESP, flash chip.
On the right board I don't see as many capacitors as I would like to see.

See pictures

N.B. they are also missing on the 'middle' board, but those fake Wemos D1 mini boards without shield over the ESP are notorously bad. It really is a wonder that board isn't causing issues for you.
They have a really underpowered voltage regulator (often max. current of 150 mA) and missing some beefy capacitors.
Also these boards lately have really bad USB connectors which are really picky on what USB cable you're using.

As far as I can see it is not a fake board, it as this board..https://www.wemos.cc/en/latest/d1/d1_mini.html, only version v3.0.0

Maybe "fake" isn't the correct term for the "Wemos" or "Lolin" board.
Problem is that lots of "new" versions appeared on the market after Wemos stopped making boards under their brand as it was merged into Lolin.

Some were names with names like "WEMS", but others didn't even bother trying to rename the boards.

So it is very hard to tell fakes from original boards.

Fact is that those later boards (also the Lolin D1 mini you linked) do have a voltage regulator which is far less powerful compared to the original (original as in "rev1") Wemos D1 mini boards.
As you can see in the image, the voltage regulator is this 5-pins SOT-23 chip next to the pin labels "D8" and "D7"
Especially those boards without a shield covering the ESP differ a lot per board with respect to stability.
Even modules which appear to be from the same batch, may look completely different when looking at stability.
I know components do have some tolerance, especially capacitors have a reasonable high tolerance in true capacity.
So maybe it is just a matter of having a design with component choices too close to the absolute minimum. If parts do have a tolerance on the negative side, the whole design becomes unstable.

Also those boards, without shield over the ESP, have been reported to experience issues when sending at max. TX power.
That's why I asked it again. But now you mention it, you did indeed check in a lot of steps to see if it has any effect.
N.B. the issue with max. TX power wasn't related to power delivery, but more about the RF signal flipping bits inside the ESP causing crashes.

TD-er wrote: ↑16 Jun 2021, 11:34 Fact is that those later boards (also the Lolin D1 mini you linked) do have a voltage regulator which is far less powerful compared to the original (original as in "rev1") Wemos D1 mini boards.
As you can see in the image, the voltage regulator is this 5-pins SOT-23 chip next to the pin labels "D8" and "D7"
Especially those boards without a shield covering the ESP differ a lot per board with respect to stability.
Even modules which appear to be from the same batch, may look completely different when looking at stability.
I know components do have some tolerance, especially capacitors have a reasonable high tolerance in true capacity.
So maybe it is just a matter of having a design with component choices too close to the absolute minimum. If parts do have a tolerance on the negative side, the whole design becomes unstable.

I do understand that a right design is necessary for a stable working ESP-board. So if this board doesn't work, I would have throw is away.
But this board works very stable with the builds mega-20190226 and older. So something must have change in the build that maybe asks more from the components why it get's unstable.

Also those boards, without shield over the ESP, have been reported to experience issues when sending at max. TX power.
That's why I asked it again. But now you mention it, you did indeed check in a lot of steps to see if it has any effect.
N.B. the issue with max. TX power wasn't related to power delivery, but more about the RF signal flipping bits inside the ESP causing crashes.

I have tried to put the package-foil in where the ESP is delivered, to use as a shield by folding it, and just lay it over the ESP8266EX-chip, not blocking the antenna. But also that did not solve the problem.

As far as I can see now, the problem starts when the ESP switches from AP tot STA mode. As long as it is in AP mode it works like designed..

As I suggested, would you like me to send you one? So you can experiment with it?

If those boards are the ones (multiple) that do not work, while others do, then I guess it may be a good idea to send one.
How is it powered?
Via USB, or do you supply your own voltage? On which pins?

TD-er wrote: ↑16 Jun 2021, 14:44 If those boards are the ones (multiple) that do not work, while others do, then I guess it may be a good idea to send one.
How is it powered?
Via USB, or do you supply your own voltage? On which pins?

Via the USB port, with a iPad 5.2V 2.5A adapter

TD-er wrote: ↑16 Jun 2021, 14:44 If those boards are the ones (multiple) that do not work, while others do, then I guess it may be a good idea to send one.

PM me your address, I'll send one today.