I have noticed that wifi brick is not reliable. I have a prototype sensor based on the wifi extension (sensor_1) and second one using RPI zero W (sensor_2) running at the same time in the very same location on the uni campus.  Wifi signal is perfect at around -55  The infrastuture is CISCO enterprise AP's

 

The tinkerforge bricklet will drop from the network occasionaly and will always do that for 3h45m  an then it will recover. Why? What is wrong with it?

 

On a graph when the response is 1k the sensor is unavailable.

Hard to say, this is definitely not a general problem. We have had WIFI Extensions running for many months without any connection drop.

 

Do you have any more information about your setup? Are you using the WIFI Extension V1 or V2?

This is V2 latest extension with updated firmware. connection running simple WPA2. The only unusual setup item is that the SSID of network is not broadcast.

The drop period is always the same, is there some kind of watchdog in the brick? 

There is nothing in the firmware that should stop it from reconnecting to a network for 3:45 hours.

 

Does this happen in a regular interval? Like every XXX hours? Do you have any logs in your network infrastructure that show if the Extension tries to reconnect or just does nothing?

Is the AP in N-only mode? If so then try to configure it in B/G/N mode. This can be an issue for hidden (no SSID broadcast) APs as discussed here, https://github.com/esp8266/Arduino/issues/3457.

 

Another thing to try would be to make sure the AP is not switching to a different channel. For example some APs have the feature of switching to a different channel if it detects the currently used channel is over crowded with other APs around. Particularly applicable for 2.4GHz APs.

 

Cheers.