I think I found the problem. I think it misbehaved because of an illegal character in the "HOSTNAME". Changed the hostname yesterday and It was still alive this morning even if I have not pinged it.
Still, there seems to be 'someting rotten in the state of Network' where my LAN is concerned. Although the Arduino serial monitor says:
13:03:55.821 -> connected with XXX, channel 6
13:03:55.868 -> ip:192.168.178.4,mask:255.255.255.0,gw:192.168.178.1
13:03:55.915 -> ip:192.168.178.4,mask:255.255.255.0,gw:192.168.178.1
13:04:04.826 -> pm open,type:2 0
when I ping the IP-address I get:
Pinging 192.168.178.4 with 32 bytes of data:
Request timed out.
Request timed out.
Request timed out.
Request timed out.
Ping statistics for 192.168.178.4:
Packets: Sent = 4, Received = 0, Lost = 4 (100% loss),
When I use the tool 'Wireless Network Watcher' I do see that a device is connected at 192.168.178.4 with the correct MAC address.
However, when I look at the management page of my FritzBox 7490 modem/router no device is mentioned at 192.168.178.4.
I remain at a loss here.
The final version of the boards contains few notable improvements
Removed the PCB under the ESP8266 antenna. The previous design had PCB + Ground Plane, which resulted in VERY low reception. It worked just few meters away from the router.
Removed the Temperature Probe - On the one hand - it does not really measure real temperature, on the other hand it populates the Single Analog port, and makes it unavailable for other usages
Added place for an optional LDR (in case the Analog port is not used - one could add an LDR for added value)
The connection between GPIO 16 and RST (Deep Sleep) is not bridged via a 0603 0ohm resistor (jumper). Depending on application it may be enabled or not
Added a Power 5V line. That is - controlled via a SOT232 mosfet and one GPIO. This way you can connect power hungry modules and then turn them on or off. There are small mosfets rated ~ 2-3 amps for couple of cents.
The big 12mhz crystal for the CH340G was replaced by a small patch ceramic oscillator to save space
Added an additional mounting hole
Added additional test pads for the unused GPIOs
Added one Diode in series with the power to the WS2812B LED. It turned out, when powered directly via a USB Battery bank, the voltage can be 5.30-5.40v. And apparently WS2812B, does not light if VIN is > 5.15V. So putting a generic small Diode would drop 0.3-0.4V
Here is how the latest design looks like
Qu3Uk,
I did get a chance to test with a Fitbit recently. It technically worked, but not all that well. The thing is, the Fitbit only advertises every 2 seconds so latency is a bit high. But worse, its apps really want to be connected to it often to sync; and whenever it's connected, it stops advertising and the detector then can't pick it up.
TommySharp,
At this point I don't plan to modify the board much but having it read environmental sensors would be a cool feature!
For that, would be nice to make little beacons that read sensors and advertise/broadcast the readings every few seconds. Then they could be very low-power, run on coin-cells for months, and could be placed anywhere instead of needing it to be hooked up to USB power like the main board is. Like these: https://sen.se/peanuts/ I'm curious if it could read them.
As for enclosures, I know. I would love to have some nice ones but at this point I'm making too many boards to 3D print enclosures, but too few to afford injection-molding tooling to make a custom case.