Left here for archival purposes.

User avatar
By martinm
#8891 - fails if DS18B20 data is sent to thingspeak, but also fails with just calling node.dsleep(1000000) a couple of times
- connected GPIO0 to Vcc (which is a bad idea in general) and pull-ups to RESET, XPD, GPIO2 as suggested in post 7078, still fails
- Voltage at GPIO0 depends on the capacitor/resistor connected to GPIO0, voltage > Vcc is seen with nothing connected
- faulty GPIO0 signal does not look like a digital signal, had a quick look only for the first post

As written above, it is actually the 26MHz oscillator

- measured with an Agilent U1604A, not the best equipment, will try with a better tool

- looks like the XTAL_OUT looks bad (heavy 3rd harmonic) during zombie mode, looks fine during normal operation

Maybe the oscillator is not starting properly?
User avatar
By gwizz
#8895 I was re-reading the whole thread, and noticed there was a comment that this only was a problem in firmware > build 20150126.

Can anyone confirm this? If so, perhaps we can report it to zeroday?
User avatar
By GeoNomad
#8902
gwizz wrote:I was re-reading the whole thread, and noticed there was a comment that this only was a problem in firmware > build 20150126.

Can anyone confirm this? If so, perhaps we can report it to zeroday?


Not for me. This problem has occurred in numerous builds all January. I sure what firmware is in the module I am testing today, but it is probably around 20150107.

BTW. Has anyone checked if the problem occurs when the module goes to sleep vs when it tries to wake up?

As for me, it is very rare, up to once every 2 days, I never actually see it happen.

I have now added a datalogger (using another ESP module, of course) on the current measurement to figure that out next time it goes zombie.
User avatar
By GeoNomad
#8945
GeoNomad wrote:BTW. Has anyone checked if the problem occurs when the module goes to sleep vs when it tries to wake up?


Just to add to the information. I confirmed during the night using a datalogger that the zombie mode occurred on wake up from dsleep. Not on going into dsleep mode, or during program execution.

Also, the previous WiFi connection terminated normally. I had thought before that zombie mode might have been related to WiFi being in some "hung" state at the time the processor was shut down.

So, the non-wakeup seems totally random and occurred after 14 hours of flawless operation, cycling on a 6 seconds on 54 seconds sleeping cycle.

Power is provided by a 3.3 volt 1A power supply, that ran this same board and same software for 40 days with zero errors (not using dsleep).

Firmware is NodeMCU 0.9.5 build 20150108.

Has this problem been seen on firmware without Lua? Or is the problem Lua/NodeMCU related?