Reboot, Soft- or Hardware-Problem. My Solution

[schnurzel59]

27 hours without reboot.ESP-12E.
Build: mega-ESP_Easy_mega_20190315_dev_ESP8266_4M.bin
Lately I've had reboots more and more often. The reason was always the hardware watchdog. Sometimes after an hour. Then I tried a lot to get rid of the cause.
Today he already runs 27 hours. It would be good to know if the latest changes have contributed, or if it was a coincidence. Of course, I'll have to wait a few days
to see if things will go on like this.Since I control only 3 shutters, and I transfer some data from and to the Homematic, there is no reason that the hardware watchdog reacts.
I control 6 relays only with longpulse and without delay. The whole thing happens time and temperature controlled. Since the current shutter positions are lost in a reboot,
I secure this on the Homematic. If a reboot has taken place, these values ​​will be automatically restored. Now all I have to do is to fix all the shortcomings of the ESP.My actions:
Since I initially had software watchdogs, I tried it with short timer rules. Since then I did not have SWD's anymore. Coincidence? But i have enough HWD's.
Now I have an unused GPIO switched regularly:
on Rules#Timer=2 do
Pulse, 2,1,10
timerset, 2.1
endon
The intervals between the reboots became longer. But that was probably a coincidence.
The last measure was this: I ping my ESP from the Homematic every minute. Since then he has been running for 26 hours. Are disconnections now the cause of HWDs? After the reboot a connection is
immediately available. This is automatically restored without any problems. Because 10s after the reboot, the backed-up data is transferred back to the ESP.
The backing up and restoring of the data works perfectly well. I would like to show my approach here. So I'm not bored with it, but I wait, first, if someone is interested.

addendum:
Just in this moment my ESP reboots. This time not because of hardware watchdog. Hooray. This time the reason is a new one.
And right now my cat has come through the chip-controlled cat flap. (RFID) But restoring my data has worked again. I'm glad.

Reset Reason: Exception

that is new.jpg (188.16 KiB) Viewed 2194 times

[TD-er]

I know that pinging the node does have a positive effect on the wifi stability.
And I do also believe that the hardware watchdog reboots almost all are related to network interactions. Either something is waiting for a response, or something is trying to make a connection.

As of now, I really have no clue what is causing this effect of the ping.
It does seem to be related to power save settings somewhere in the core libraries (likely in the NonOS SDK being used)
If I look at the power consumption, the node hardly goes into power save mode when a ping is being sent to the node.

[schnurzel59]

Hi,
in the next days I will connect my memory oscilloscope to look at the supply voltage.
Although the power source should deliver 25W, but that's always only information from
the manufacturer. Voltage dips are always possible. I actually see that as the most obvious cause.
Now all I have to do is create a big current spike. Then I see if the voltage source is stable or not.
Then the whole thing again with a big Elko. Then we'll see.