Stability issues
Posted: Sat Jul 18, 2015 8:47 amFrom the "just saying" department... 
I have only been investigating the ESP8266 for about a month, therefore I do not post as a knowledgeable experimenter, but as a tinkerer.
The #1 issue that I have found troubling in playing with the ESP8266 modules is that prototyping on a quality breadboard is problematic: small, incremental increases in DC resistance from the breadboard and jumpers can cause strange malfunctions and even the inability to have a clean upload session after a compile. I went through all of my jumpers using a new breadboard and my lab supply locked to constant current of 500mA. The jumpers essentially became the shunt that I connected my DVM across to read the voltage drop. I found a lots of bad jumpers this way.
Another issue was the inexpensive banana plugs on the cable from the PSU to the breadboard!
Finally, to get around all of this mess, I simply used a LiFePO4 battery to power the projects. Unlike other chemistries of lithium cells, the LiFePO4 batteries are environmentally "green" and far safer than other types because lithium metal never appears in the charge/discharge cycle of the cell.
Since switching to the new cell to power my projects, I am consistently seeing 100% uploads from the Arduino IDE and 100% resets - far improved overall stability and I have yet to have a reset not start user code. This obviously will not solve any memory leaks or silicon issues, but for me it has gone a long way to settle down the breadboard projects. A benefit is that remote projects using solar charging are ideally suited to this particular battery... many solar-powered garden lights now include this particular cell. In low quantity, the cells are approximately $4 each (AA size), but Amazon and AliExpress offer them for much less.
Ray
I have only been investigating the ESP8266 for about a month, therefore I do not post as a knowledgeable experimenter, but as a tinkerer.
The #1 issue that I have found troubling in playing with the ESP8266 modules is that prototyping on a quality breadboard is problematic: small, incremental increases in DC resistance from the breadboard and jumpers can cause strange malfunctions and even the inability to have a clean upload session after a compile. I went through all of my jumpers using a new breadboard and my lab supply locked to constant current of 500mA. The jumpers essentially became the shunt that I connected my DVM across to read the voltage drop. I found a lots of bad jumpers this way.
Another issue was the inexpensive banana plugs on the cable from the PSU to the breadboard!
Finally, to get around all of this mess, I simply used a LiFePO4 battery to power the projects. Unlike other chemistries of lithium cells, the LiFePO4 batteries are environmentally "green" and far safer than other types because lithium metal never appears in the charge/discharge cycle of the cell.
Since switching to the new cell to power my projects, I am consistently seeing 100% uploads from the Arduino IDE and 100% resets - far improved overall stability and I have yet to have a reset not start user code. This obviously will not solve any memory leaks or silicon issues, but for me it has gone a long way to settle down the breadboard projects. A benefit is that remote projects using solar charging are ideally suited to this particular battery... many solar-powered garden lights now include this particular cell. In low quantity, the cells are approximately $4 each (AA size), but Amazon and AliExpress offer them for much less.
Ray